Two Sigma’s Ownership Trusts were built so that no single life event at either founder could touch who controls the $75 billion firm, and John Overdeck’s own sworn testimony in his New Jersey divorce trial is now the evidence a court would need to treat that same trust interest as illiquid enough to discount, under a legal standard that otherwise forbids exactly that discount.

Photo: Don Ramey Logan · CC BY 4.0, resized · via Wikimedia Commons

A billionaire divorce, irrelevant to the fund

Read the coverage this week and the story is a celebrity divorce wearing a hedge fund’s clothes. Two billionaires. No prenup. A $6.2 billion number that makes headlines on its own.

That reading isn’t naive. Family offices see marital splits among the ultra wealthy constantly, and the standard playbook exists because it usually works. The business keeps running. The founder writes a check or a note. A few years later nobody outside the family remembers the number. That’s the base rate.

Two Sigma’s own numbers back that up. The firm is managing more money than it ever has, a record north of $75 billion, per Hedgeweek’s reporting on the 2023 investor departure, which the firm replaced and then grew past. A quant manager whose systems and headcount are intact doesn’t need Overdeck’s personal balance sheet to keep functioning.

On this view, the right professional reaction to “hedge fund cofounder’s divorce” is the same reaction a PM has to any billionaire’s marital news. Interesting. Irrelevant to the fund.

Overdeck’s testimony is doing legal work

I think that reading skips the one detail that actually matters to anyone who allocates to, competes with, or works inside a multi strategy manager. Overdeck is not describing a personal liquidity problem. He’s describing a control problem, under oath, in a jurisdiction whose leading precedent says his kind of interest should not get a discount for exactly the reason he’s describing.

That is a specific legal maneuver. Two Sigma’s own regulatory filings are what make the maneuver credible.

New Jersey values marital business interests at fair value, a specific legal standard distinct from fair market value. The state’s default rule, from Brown v. Brown (App. Div. 2002), holds that marketability and minority discounts do not apply absent “extraordinary circumstances”: “lack of liquidity does not affect the fair value of the minority interest,” full stop, unless the case clears that bar. Brown never defined what would.

Overdeck’s testimony reads like an attempt to clear it. Family law commentators have argued, following a 2022 state Supreme Court case about a closely held business buyout, that “extraordinary circumstances” should reach fact patterns like a buy sell agreement limiting what an owner can receive, or an owner who lacks real control over compensation and capital calls.

I went looking for the actual case behind that argument. It isn’t what it’s cited as. I want to be precise about what that 2022 case, Sipko v. Koger, actually is and is not: it is a shareholder oppression dispute between family members over two closely held companies, not a divorce case, and its own outcome ran the other way. The Court denied a discount because the controlling party had acted in bad faith to deplete the company’s value, the opposite fact pattern from Overdeck’s. The buy sell and compensation control framework is the commentary of one family law practitioner writing after that case, not a holding any court has applied to a fact pattern like this one. No court has yet found “extraordinary circumstances” in a case resembling Overdeck’s.

What his testimony has going for it is narrower than “new precedent, directly on point.” It reads like a serious attempt to build the record Brown itself said would matter, using the one argument NJ’s family bar has actually been making. I cannot prove that is his legal team’s deliberate strategy. I can show you it is a fit with the argument, not with a decided case.

The mechanism: a trust built to protect control becomes the exhibit that explains why it might not

Start with what Two Sigma actually built. Per Two Sigma Investor Solutions LP’s own Form ADV Part 2A brochure, an SEC-registered Two Sigma affiliate, public and current as of the March 2025 update, trusts established by Overdeck and Siegel, called the Ownership Trusts in the filing, are “the principal owners of the Adviser.” Those Ownership Trusts “indirectly effectively control” TSM, the general partner, and the advisory business itself.

That structure isn’t decoration. Routing founder control through a trust is a standard way large private managers wall off the firm from a single founder’s estate, incapacity, or a life event like a divorce. It’s the same logic that puts a founder’s operating company shares into a family trust or a voting trust well before anyone expects a fight: name the vehicle as the owner, and no single life event touching the individual is supposed to touch the vehicle. My read is that the design intent here was precisely what Overdeck is now testifying against: making sure nothing that happens to John Overdeck the person can, by itself, touch who runs the business.

Layered on top sits a two person Management Committee, one seat per Co-Chairman. Each Co-Chairman can unilaterally remove his own appointee. That’s 50/50 by construction, with no tiebreaker. I’d call that the single most fragile governance shape a manager this size can run: it works exactly as long as two people keep agreeing, and it has no built in answer for the day they stop. No committee. No board. No vote.

The firm’s leadership actually moved through this exact structure recently, and the dates matter. On September 30, 2024, Overdeck and Siegel both stepped back from the Management Committee itself, each naming a designee in his place. As of March 31, 2025, Overdeck reversed course and reclaimed his own seat, replacing his designee. A structure this sensitive to who occupies two chairs is a structure where the identity of the equity holder behind each chair isn’t a footnote.

Here’s the chain, in the order the risk actually moves. Four links. Money has to move through all four before the story is real, not hypothetical.

One. A New Jersey court values Overdeck’s trust interest for equitable distribution, somewhere between his $4.9bn and her $6.2bn.

Two. The court orders a payment Overdeck can’t fully cover from outside liquid assets alone. His own testimony is that this is a real possibility above his $723M offer.

Three. Meeting that order means monetizing part of an Ownership Trust interest that was never designed to be sold in pieces, through a pledge, a secondary sale, or a structured payout that still changes who holds an economic claim on the trust.

Four. Because the trusts are what “indirectly effectively control” the Adviser, any change to what sits inside Overdeck’s trust is a change to the mechanism that was supposed to keep 50/50 parity intact.

Each link is disclosed somewhere on the record. The ADV brochure supplies links one and four. His testimony supplies two and three. No single document states the whole chain, which is why it reads as a divorce story instead of a governance story to anyone reading only one source.

The tell that predates the trial by three years

The independent tell is the ADV brochure itself, and it predates the trial. Two Sigma’s Item 4 risk factors, filed under the heading “Certain Risks Associated with Management and Governance Challenges,” disclose that the Management Committee “has been unable to reach agreement on a number of topics,” specifically naming C-level role definitions, “organizational design and management structure,” “corporate governance and oversight matters,” and succession planning.

That sentence sits in a document the SEC can act against if it’s false or misleading. That’s a materially higher bar than anything said in a divorce deposition. I wouldn’t weight Overdeck’s trial testimony alone as proof the control risk is real. I weight it alongside a filing the firm had every incentive to soften and didn’t, filed year after year, long before a divorce trial gave anyone outside the firm a reason to read it closely. Nobody was reading it for this. It’s also not the first time Two Sigma’s own regulatory record has cut against its public reputation: I wrote up its $90 million SEC fine earlier this year, on a separate compliance failure the firm itself disclosed.

The same two week window in August produced two more data points, and both firm the mechanism up. On August 17, entities linked to Siegel’s family filed arbitration against Overdeck and co-CEO Carter Lyons, alleging undisclosed conflicts tied to a former employee’s severance and deferred compensation claim. Hedgeweek reported the dispute escalated all the way to formal arbitration, something the Management Committee itself was evidently unable to resolve internally.

The clearest account of the second dispute’s mechanics isn’t in any press report. It’s in Two Sigma’s own Form ADV, filed with the SEC on March 31, 2026. Siegel designated Seth Platt his Executive Manager and co-CEO on March 16, 2026, replacing Scott Hoffman, who Hedgeweek reported had resigned citing “continuing governance difficulties.” Platt then moved to terminate co-CEO Carter Lyons, arguing Lyons had undermined his authority. Overdeck called that move “imprudent and baseless” and placed the termination into dispute resolution, meaning it cannot take effect without Management Committee agreement or an arbitrator’s ruling. Whether an Executive Manager appointment even confers the co-CEO title at all, the filing states plainly, “is in dispute.”

That is Two Sigma telling the SEC, in its own words, that one Co-Chairman’s designated lieutenant tried to fire the other Co-Chairman’s designated lieutenant, and the firm’s own governance apparatus could not settle it without outside intervention. The same regulatory filing names the underlying ownership structure precisely: the “John A. Overdeck Revocable Trust” and the “David M. Siegel Revocable Trust,” both Delaware entities, both Limited Partners in the 25-to-50-percent band, each founder listed as trustee of his own trust. As of February 2026, a third party, Thomas Rafferty, was added as an additional trustee on Siegel’s trust. Overdeck’s trust carries no such addition on the same filing. I can’t tell you what that difference means. I can tell you it’s the kind of structural fact this system was supposedly designed to make irrelevant to any single person’s circumstances, and it isn’t. One trust got a second trustee. One didn’t.

Three separate legal proceedings. Two founders. Sixteen days. That’s what a firm running its command structure on ad hoc litigation looks like, and it’s happening at the exact moment a family court is being asked to decide how much one founder’s economic interest is worth. The Platt-Lyons succession fight itself already has a second life: I tracked how fast the public record moved versus how slowly an actual institutional allocator reacted once the story broke.

What “monetizing part of an illiquid trust interest” would actually look like is worth spelling out, because it isn’t a hypothetical this industry has never seen. Founder and GP-level stakes at private managers get sold every year to specialist buyers: Petershill, and Blue Owl’s GP Strategic Capital (the business Dyal Capital Partners became after its 2021 merger). A founder’s economic interest in a fund manager is illiquid by design, and a market exists to price that illiquidity anyway. A structured minority sale of a slice of Overdeck’s trust interest, or a margin loan collateralized by it, are the two realistic mechanisms for closing a multi billion dollar gap without selling the whole interest. Either one still changes who has a claim on the trust that “indirectly effectively controls” Two Sigma, even if Overdeck’s own voting seat never formally moves. The control risk doesn’t require a headline sale to be real. It only requires a lender or a minority buyer with a seat at the table that didn’t exist before.

The timeline: this was never a sudden story

Lay out the dates in order and the divorce trial stops looking like the origin of the problem. It starts looking like the latest entry in a ledger that has been running for three years.

March 31, 2023. Two Sigma first disclosed the Overdeck-Siegel rift to regulators as a business risk, in a Form ADV update Fortune reported on the same day it became public. The filing’s own language warned that “if such disagreement were to continue, the adviser’s ability to achieve client mandates could be impacted over time.” Around the same window, a major investor exited the platform citing that governance concern. The firm replaced the capital. The disclosure did not go away. It is still there, almost word for word, in the March 2025 update I read for this piece. Two years of an unresolved risk factor is not a disclosure. It is a standing admission that nobody inside the firm has found a structural fix.

September 30, 2024. Overdeck and Siegel both stepped back from the two person Management Committee itself, each naming a designee, and became Co-Chairmen of the general partner instead. The idea, I would guess, was to put daily governance in less personally entangled hands while the founders kept ultimate control through their trusts.

March 31, 2025. Overdeck reversed that, rejoining the Management Committee directly and replacing his own designee. Whatever the designee arrangement was supposed to fix, Overdeck judged it insufficient inside six months.

August 17 and 18, 2026. Two arbitrations, a day apart in filing terms and a world apart in subject: Siegel-linked entities accusing Overdeck and co-CEO Carter Lyons of undisclosed conflicts, and Seth Platt disputing whether his own committee seat carries the co-CEO title he believes it does.

September 2, 2026. Overdeck testifies that preserving voting parity with Siegel is a “principal concern” in his own divorce, the same figures and quote Bloomberg’s original report carries, independently confirmed here.

My read of that sequence: every governance fix Two Sigma has tried since 2023 has lasted less than two years before someone inside the firm needed a court or an arbitrator to settle it instead. Designee committee seats. A Co-Chairman layer. A title clarified by memo, never formalized in a document. None of it held. Every fix bought months, not years. A structure that keeps needing outside adjudicators is not stable. It has been quietly outsourcing its own governance function to litigation, and the divorce trial is simply the first instance visible to people who do not work there.

What any LP in a founder controlled fund should actually check

This is a design pattern common to nearly every founder led multi strategy manager, and Two Sigma’s public filings happen to be unusually legible about it right now. Three checks translate directly to any fund where control sits with one or two named individuals rather than an institutional board.

Read the ADV Part 2A’s risk factors section for governance language, not just performance and conflicts disclosures. Most LPs skim Item 4 for AUM and skip straight to fees. Two Sigma’s own “Certain Risks Associated with Management and Governance Challenges” heading has been sitting in plain text, publicly filed, for anyone who reads past the first page.

Ask directly whether founder economic interests sit in a trust, a direct partnership stake, or something else, and whether that vehicle has ever been tested by a life event. A trust that has never faced a forced valuation is an assumption. Nothing more. Two Sigma’s is being tested in real time, in public, right now.

Check whether governance depends on a fixed number of named seats with no tiebreaker, versus a board with independent members or a supermajority mechanism. A 50/50 committee with unilateral removal rights per seat, which is what Two Sigma discloses, has no structural way to break a genuine deadlock short of the two principals agreeing or an outside body forcing the question. That’s the exact condition under which one founder’s personal legal exposure becomes the whole firm’s problem.

None of those three checks require anything beyond documents every LP is already entitled to receive. Most never ask. That’s the gap, not the filing.

I ran this exact three clause check end to end on Two Sigma’s own filing, with the precise document retrieval path, the CRD lookup, and the dated trigger this specific case turns on: Two Sigma’s Own SEC Filing Shows Exactly How a $4.9bn Ownership Trust Works.

Strategic testimony, or a real balance sheet event?

The rival explanation deserves its full weight, because it is the one a skeptical family law practitioner would raise first. Spouses routinely overstate illiquidity and control risk specifically to argue the business interest down. It is close to the single most common maneuver in a high net worth divorce, and Overdeck’s counsel would be negligent not to make it.

Under that reading, “maintaining voting parity is a principal concern” is a line written for a judge. The ADV language is just risk factor boilerplate that every multi strategy manager with founder tension eventually files.

What separates the two readings is timing and audience. A risk factor written for a securities regulator, in a document that predates the trial by years and carries its own legal exposure for misstatement, is a different kind of evidence than testimony written for a family court judge weighing a settlement.

If the Management Committee gridlock were purely a talking point invented for this trial, it would not already be sitting in a 2025 client facing brochure describing succession and governance disagreements the firm had every reason to minimize. Two Sigma’s compliance and legal teams reviewed and filed that language before anyone knew a divorce trial would ever quote it back.

The ADV disclosure does not prove the control risk framing is honest. It proves the underlying gridlock is real and independently documented. That is the one fact a purely strategic reading of his testimony cannot explain away. The paper trail predates the motive.

Four things I can’t verify from public documents

I can’t verify the actual terms of the Ownership Trusts. The buy sell provisions, whether either founder holds a contractual right of first refusal on the other’s interest, whether the trust documents contain a specific divorce triggered transfer clause: none of that is public. The ADV brochure describes the control effect, not the trust mechanics that produce it.

I also can’t confirm the New Jersey court will accept a forced sale theory at all. Overdeck’s testimony argues for it, but no ruling on valuation methodology has been made public, and the trial doesn’t resume until October.

And the AUM record itself is a genuine confound. If the market believed governance risk at Two Sigma were acute, a multi strategy manager that depends on continuous capital raising would be the first place it showed up. It hasn’t shown up there yet. The market isn’t pricing this.

A fourth complication, and it cuts against my generous reading of the trust structure’s design intent. Laura Overdeck has a separate, still live lawsuit against the law firm Seward & Kissel, reported by The Wealth Advisor, alleging the firm moved several billion dollars of marital assets into new Wyoming trusts in 2018 without disclosing that the new terms would strip her beneficiary status the moment either spouse filed for divorce. Those are different vehicles from the Ownership Trust this piece is about, so the allegation doesn’t touch the ADV facts above. But it means I can’t take “the trust structure was designed as neutral governance protection” as a given about this particular person: his family has a separately litigated history around trust vehicles and divorce triggered terms, and a reader should weigh that when judging how much good faith to extend to the design intent reading. Good design and good faith are not the same claim.

I also can’t tell you how contested the $6.2bn figure currently is inside the actual courtroom: Laura Overdeck has been barred from testifying and from calling her own valuation expert at trial, a sanction Judge Buechler imposed for discovery misconduct: accessing Overdeck’s private computer and mailbox, and photographing emails before deleting the evidence. The $6.2bn number is on the record, but the witness who produced it may never take the stand. A number nobody can defend live is a number, not a fact yet.

That gap, between a record asset base and three simultaneous legal proceedings over control, is real. I’m not going to resolve it by assuming the quieter number is the one that’s wrong.

What would change this view

Watch three things, each dated and public.

The underwriting question for every LP in a founder controlled fund

The professional question this week isn’t whether Two Sigma survives a divorce. At $75 billion and growing, it almost certainly does.

The real question is whether “control via trust” holds up the first time a court outside the firm’s own governance documents gets to test it, or whether it only ever worked because nobody had tried.

My read: the ADV brochure is the tell that the Management Committee’s gridlock was real before any of this reached a courtroom. The control question Overdeck is litigating isn’t manufactured for the case. It’s the same gap the firm has been disclosing to its own clients since March 2023, arriving at a different address.

Which side of that gap would you underwrite? The trust that has held through more than three years of documented gridlock, or the trust that has never yet faced a court order it couldn’t simply absorb.

The Decision-Grade Version

This piece is complete on its own. The thesis, the evidence, and what would kill the view are all above, and nothing was held back to sell you a next step.

The Patreon note is a separate piece of work rather than a deeper cut of this article. It takes the exact SEC filing method this piece runs on Two Sigma, i.e. the three clause Schedule A and Item 4 checklist, and writes it as an instrument any reader can run themselves on any founder controlled fund they hold. Written for people who do their own due diligence.

→ Read the Two Sigma governance checklist note

→ Or join the Patreon community for every note

More research like this: YouTube · LinkedIn

Cover photograph: Tdorante10, CC BY-SA 4.0, via Wikimedia Commons.

A hedge fund’s own compliance paperwork disclosed a new pod’s brand and website before any press outlet reported the hire.

The figures behind all of it, pulled straight from the filings:

Below the paid line:

• The full timeline: July 31 (data as of), then August 19 (form date), then September 3 (Hedgeweek)

• The exact Schedule D miscellaneous language ExodusPoint filed, quoted verbatim, plus the CIK, CRD and 801 numbers to pull it yourself

• ExodusPoint’s June 30, 2026 13F snapshot: $15.53bn reported long book, 1,510 positions, the options split

• The $2bn Abu Dhabi backed raise this pod draws against, and what fraction of it $1bn+ represents

• WestWind Asset Management: ExodusPoint’s only other prior other name, the 2021 loss event tied to its era, and why its ADV entry is still a live check today

• A dated, falsifiable trigger: what to watch for at ExodusPoint’s next annual amendment, due by March 2027

The consensus: a hedge fund hire is a press event

Trade coverage of the multi strategy pod shops treats a new portfolio manager’s arrival the way it treats an earnings beat. A fact exists the day it gets reported. Hedgeweek’s story this morning follows the standard shape. A senior macro trader with a name, Ben Atlas, formerly of BlueCrest, leaves one platform and lands at another with a headline allocation of “more than $1bn.” The story runs. The aggregators pick it up. LinkedIn does the rest. Simple. Familiar.

That reading captures something real. It’s where the information turns usable for a reader. State it at its strongest: trade journalism is the primary record for personnel moves at private funds, because private funds don’t issue an 8-K when they hire someone. There’s no earnings call. There’s no proxy statement. A private multi manager platform’s staffing stays opaque, by design, until somebody with sourcing inside the industry writes it down.

One correction before the variant view: this isn’t the first time someone read an ADV filing

The consensus above has already moved once, and I want to be honest about where. Mining Form ADV for staffing news isn’t unknown. eFinancialCareers already tracks headcount, year over year, at Millennium, Point72 and Jain Global, straight from ADV’s own employee count item, published before those firms say a word about it themselves. That practice is real, and it isn’t what I found. Headcount is one number, filed once a year, aggregated across an entire firm. Different field. Different clock. Schedule D’s other business names section is a different field entirely, filed the moment it changes, naming one pod at a time. Nobody appears to be reading that field yet.

I don’t think the press comes first, not here. I can point to the document that says so.

The variant view: the SEC already had it

ExodusPoint Capital Management, LP is registered with the SEC as an investment adviser under CRD number 294156, SEC file number 801-112901. I expected its filing history, if it showed anything at all, to read as routine housekeeping. I pulled its SEC investment adviser record directly, on the IAPD system the SEC runs specifically for adviser filings, separate from EDGAR, to check. It did not read as housekeeping. The most recent filing on record is not the annual amendment every adviser owes once a year. It is an “Other-Than-Annual Amendment, All Sections,” and the generated form itself carries the date 8/19/2026 5:50:30 PM. A specific name. A specific date. Not housekeeping.

Schedule D’s miscellaneous section exists so an adviser can explain, in its own words, what an amendment actually changed. Here is ExodusPoint’s own ADV brochure, quoted verbatim:

“This Other-than-Annual Amendment reflects the addition of Atlas Capital in Schedule D, Section 1.B and the associated website for Atlas Capital in Schedule D, Section 1.I. Further, Item 5.B(3), Schedule D(1.F), and Schedule R(4.A) were updated as of July 31, 2026.”

Section 1.B of that filing lists “Name: ATLAS CAPITAL,” registered in New York. I checked the actual jurisdiction grid rather than trusting the label list a plain text export throws back: of the 55 state and territory boxes on that page, exactly one is checked. Section 1.I lists its website, atlas-capital.com, next to ExodusPoint’s own site and its LinkedIn and Glassdoor pages. That’s the same name AlphaMaven’s aggregation of the Ben Atlas story carries for the pod he’s said to be launching later this month, though AlphaMaven’s own byline field credits Hedgeweek, so this is one wire syndicated twice, not two separate confirmations. Weigh it that way rather than as double corroboration.

Three dates, one document, one wire story. July 31 is when the underlying change was already true. August 19 is when ExodusPoint told the SEC. September 3 is when Bloomberg told the market, with Hedgeweek and AlphaMaven both relaying the same reporting. The compliance record beat that story to this hire by at least fifteen days. Measured against the date the underlying change was reported true, it beat it by more than a month. Paper first. Wire second.

I checked the obvious falsifier before writing a word of this. Is Atlas Capital a name ExodusPoint has used for something unconnected to Ben Atlas? I found no other candidate. The name is new to this amendment. Nothing in ExodusPoint’s filing history before August 19 carries it. No prior use. The falsifier does not land. I also checked the domain live while drafting: atlas-capital.com doesn’t currently resolve. A registered brand, waiting on its own launch.

The mechanism: why the paperwork runs ahead

The credit belongs to SEC Rule 204-1, a function of how the rule actually works, no leak and no clever detective work on my part.

Registered advisers file Form ADV’s annual amendment once a year, within 90 days of fiscal year end. The SEC’s own General Instructions require something else too. Additional amendments, filed promptly, whenever information reported under Item 1 becomes inaccurate. Other business names live inside Item 1, in Section 1.B. The instructions carry real teeth: “failure to update your Form ADV, as required by this instruction, is a violation of SEC rules or similar state rules and could lead to your registration being revoked.” Real teeth. Compliance does not wait for a press cycle.

A firm cannot wait for its annual cycle to disclose a new trading name once that name is actually in use for advisory business. That’s the exact phrase the form uses: Section 1.B asks for any additional names under which the adviser conducts its advisory business. Registering Atlas Capital is not a marketing flourish. It is ExodusPoint telling the SEC it is, or will imminently be, legally doing business under that name. A regulatory act with a deadline attached. Nothing softer than that.

A press announcement runs on an entirely different clock. It waits for the hire to close. It waits for both sides to agree on language. It waits for a trade reporter’s sourcing to come together. None of that carries a regulatory deadline. The compliance filing has one. The news story doesn’t. That asymmetry is the whole mechanism, and it isn’t personal. Nobody at ExodusPoint is trying to bury this in EDGAR. The filing exists because the rule requires it, on a timeline the rule sets, whatever the trade press decides to do with the story later.

I turned the actual checking process into its own walkthrough, the exact CRD lookup, the PDF field to read, and the jurisdiction checkbox mistake that cost me a correction inside this very piece. The Five-Minute ADV Check has all of it, so you can run this same check on any adviser yourself, not just this one.

Read more

Below the paid line:
- The full disclosure inventory: each of the three subpoena asks mapped to the rule, form or system that already carries it, with the citation
- The 13F growth series read off five EDGAR cover pages, $2.12 billion to $20.24 billion, and the General Instruction 4 note that let Form ADV Items 5 and 7 stay lawfully wrong
- The proposed $10 billion Form PF threshold worked against this fund’s own $9,278,344,000, and the $721,656,000 that separates them
- The Archegos theory stated at full strength, in the SEC’s own charging language, and what survives it
- Rule 13h-1(d)(2)(xii) and the next morning route to trade timing that needs no subpoena
- Five confounds I can’t rule out, including the one that most damages this piece

The consensus, stated at its strongest

Here is the reading in circulation this week, in the form its own believers would sign. A fund built a concentrated, heavily financed book in AI infrastructure names and ran it through four of the largest prime brokers on the Street. In late July the trade went against it, margin calls landed, and the portfolio was sold into a market with no natural bid. Someone in that sequence may have done something wrong, and the SEC has gone to the lenders to find out.

That’s a coherent story and parts of it are certainly true. Reporting on the asks is specific. The subpoenas sought information on when Situational Awareness executed its trades and on how the fund communicated with its lenders about borrowed capital, and the banks were told to preserve records (Quartz, via Yahoo Finance). I read preservation notices as what a regulator sends when it wants the option to look later. Nobody has been accused of anything, the SEC declined to comment, and none of the four banks commented (CNBC). Both accounts are relays and it is worth knowing whose: CNBC is carrying Reuters, the Quartz piece is carrying the New York Times.

Situational Awareness’s own response fits the frame. “It is to be expected that regulators would closely examine any funds that are high profile, produce significant returns or have particularly dramatic drawdowns,” a spokesman said. “We are a highly regulated business and will cooperate to the fullest extent with any regulatory request.”

I don’t think that reading is foolish. I think it’s incomplete, and the thing that completes it never appears in the reporting. It sits in the filings.

What the filings say, and the part nobody checked

Quartz, summarising the New York Times account, carries a clause that is easy to skim past: the banks are known “according to a regulatory filing cited by The New York Times.”

I know of one public filing that names a private fund’s prime brokers, and I could not find a second. It is Form ADV, Schedule D, Section 7.B.(1), Question 24. I pulled it.

Situational Awareness LP, CRD 333011, SEC file 801-132039, filed 11 June 2026, signed under penalty of perjury by chief compliance officer Kimberly Summe. Question 24(a) asks whether the private fund uses one or more prime brokers. “Additional Prime Broker Information: 4 Record(s) Filed.”

Four names. I count the same four. No addition and no omission. The reported list stops exactly where the filing stops. Morgan Stanley, a top tier US prime broker, appears on neither.

The subpoena list is a filing, not a finding. That is the first thing the documents say. The second is stranger, and I had to read it twice.

A registration record less than half the size of the fund’s own 13F

Question 11 of that same Form ADV asks for the “current gross asset value of the private fund.” Filed on 11 June 2026, the answer is $9,278,344,000. Item 5.F reports identical regulatory assets under management, across two accounts.

Nineteen days after that signature, at the 30 June quarter end, the fund’s own Form 13F reported $20,242,292,228 in US listed long equity across 26 positions. That comes straight off the cover page, and I did none of the sums in it. I make the ratio 2.1817, so 2.18 times, and the gap $10.96 billion. A gap between what an adviser files and what anyone prints is not unique to this fund. I took the same two fields apart on Citadel, where the filed number and the circulated number differ by a factor of eight.

Both figures cover the same book: the ADV reports a single master-feeder fund, the 13F the adviser’s discretionary holdings. Two limits on the comparison, and both cut the same way. A 13F reports long US listed positions at a quarter end and nothing else, no shorts, no derivatives, no non-US holdings, and it lands 45 days late, so $20.24 billion understates the book. Regulatory assets under management is a gross figure, deducting no debt and no leverage, which is precisely why it should be the larger of the two. Both corrections widen the gap. Pull the whole 13F series off EDGAR and the reason becomes obvious.

Each row is long US listed equity at that quarter end, and each is the cover page total, not my sum of the holdings.

$2.12 billion to $20.24 billion in four quarters. I make that 9.53 times. That $9.28 billion on the ADV sits between the December and March rows, which is roughly where a fiscal year end gross asset value would land for a fund whose December long book was $5.52 billion. So the number is stale. Here is the part that matters. It was allowed to be stale. The Form ADV General Instructions, Instruction 4, on what an adviser must refresh when it files an interim amendment:

Item 5 is regulatory assets under management. Item 7 is private fund reporting, which is where Question 11 lives. The 11 June filing is captioned, in its own header, “Other-Than-Annual Amendment - All Sections.”

I re-checked that filing on 26 August, six weeks after the fund reportedly lost around $35 billion, an unwind I traced through Citadel’s own subsequent selling. It is unchanged. Nobody did anything wrong here. That’s my whole point. “Current” in Question 11 means current as of the last time the form was obliged to carry a number, and for a fund that grew ninefold inside a year, the obligation arrives once. A regime that updates annually cannot see a fund that moves quarterly. No rule was broken and the record was wrong anyway.

If you want to run this on your own book. I pulled Question 24 for six advisers this week, this fund among them, to see whether naming four prime brokers is unusual. The census, the two step recipe for pulling that field on any SEC registered adviser, and what the spread does and doesn’t tell you are in the members’ note on Patreon.

Read more

The $14.6bn refinancing everyone priced in August was not Jane Street’s largest new fixed charge of 2026. A fifteen year absolute triple net lease on an Oklahoma data centre, signed the next day and parent-guaranteed, commits it to $289m to $311m a year against roughly $175m of incremental interest on the notes. The notes are counted, rated and public. The lease is none of those.

Below the paid line: the committed charge arithmetic, 149MW at the EIA’s June 2026 Oklahoma tariff of 7.37c/kWh, to a fifteen year undiscounted $4.34bn to $4.66bn. Moody’s 30% trigger worked line by line at 26.5% and 30.6%, and the 8x-rent method that lands it at 29.5% instead. Then the full 2026 obligation stack, the capacity table, the ICE BofA control, and five confounds I cannot rule out.

What Jane Street’s $14.6bn refinancing bought, and why it read like the whole story

On 12 August Jane Street priced $14.6bn of senior secured notes into a private group led by JPMorgan with PIMCO in the book: $5.86bn to 2031, $5.13bn to 2033, $3.64bn to 2036, the long piece at 8.088%. Roughly $11.2bn retired a $5.5bn term loan and $5.7bn of existing notes, as Bloomberg reported. The reading that followed was coherent and mostly right. New money costs about 311bp against roughly 152bp on the paper it replaced, about $175m a year of incremental interest, and it takes the funding private. Bloomberg Opinion called the swap smart self-defense. A firm that had just printed its first losing month in a decade paid nine figures a year to control who reads its numbers. I wrote about that two weeks ago and still think the disclosure motive is right. This piece is about the deal nobody set beside it.

The three obligations of 2026, and the one that got priced

Jane Street has been buying compute all year, four ways, and nobody has put the year in one table.

CoreWeave’s announcement is unambiguous: “Jane Street has committed approximately $6 billion to use CoreWeave’s AI cloud platform,” alongside a $1bn equity investment at $109.00 a share. It states no term, a gap I return to. Bloomberg reported in June that the firm is separately weighing its own 100MW to 200MW data centre, self financed, on a plan to go from tens of thousands of GPUs to hundreds of thousands.

The two off metric lines come to $8.25bn in one year, against $1.90bn of headroom under Moody’s test, worked below.

⛔ Those two are not the same instrument. A guaranty of third party debt is close to debt and is the kind of contingent obligation a rating methodology can capitalise. A cloud purchase commitment is a vendor contract, and no agency’s leverage metric counts those as debt, for any issuer anywhere. The lease is the sharp case. CoreWeave is not evidence of a gap in anyone’s metric. That one costs me the headline number. It’s there for scale: its April date turns the August lease into a pattern.

Zenith Arc, 149MW, and a lease with no bid

Zenith Arc LLC, a subsidiary of Prairieland Rivers LLC, sold $2.25bn of green bonds due 2031 to fund a data centre and substation in central Oklahoma: 149MW, leased for fifteen years to JS Data Center LLC, a Jane Street Group subsidiary, with the parent guaranteeing. That lease is no normal tenancy: under an absolute triple net lease with a guaranty Jane Street owes base rent and, separately, every property expense: maintenance, property taxes, utility bills, insurance premiums, and electricity. All of them, uncapped, fifteen years, whether the machines inside earn anything or not. Moody’s rated the notes Ba2, two notches below investment grade. I read the guaranty as why they cleared at all.

That shape is why the credit prices where it does. A firm at 6.2x balance sheet leverage borrows at investment-grade-adjacent levels because its book is self-liquidating: if the business stops on a Tuesday you sell it and pay everyone back by Friday. Jane Street Capital’s own Form X-17A-5 for 2025 shows it: assets $60.70bn, liabilities $54.42bn, equity $6.28bn, and securities sold short of $45.01bn against long positions of $41.70bn. A book made of things with bids. A fifteen year lease on an Oklahoma substation has no bid, cannot be sold or hedged, and does not shrink when the trading book does. July was the first down month in a decade; the rent would have been due anyway. Jane Street has always had office leases, so non trading obligations aren’t new. The scale is, and so is a public rating letter on one.

Baa3, Ba1, Ba2: Moody’s three notches, and why they prove less than they look

Moody’s affirmed Jane Street Group at Ba1 on 22 July, outlook to positive, and affirmed Baa3, investment grade, on the four wholly owned operating subsidiaries. Two days later Fitch upgraded the group to BBB- from BB+, citing “strong through-the-cycle growth in net operating income.” Three weeks later the same agency looked at a bond whose economics are almost entirely Jane Street’s promise to pay rent, and rated it Ba2.

I wanted that ladder to mean the market prices Jane Street’s duration. It doesn’t, and I’ll say why before someone else does. Clifford Chance’s July briefing on exactly these structures says data centre lease bonds are “notched down for structural reasons (e.g., completion risk or refinancing risk at a bullet maturity)”, with the notch being about structure, not credit. Zenith Arc has both: a construction financing with a 2031 bullet against a fifteen year lease. And S&P didn’t notch it at all, assigning a preliminary BB, where it already rates Jane Street.

The ladder is real. It is not evidence of what I wanted, and I am not going to dress that up. What survives: a Jane Street obligation is now rated in public, on its own, by two agencies, and neither puts it above the firm. The rest runs on the cash.

The rent floor is $202.5m. The power bill is bigger than I expected.

Nobody discloses the rent, so I built a floor from the debt it services. $2.25bn priced at 99.5 to yield close to 9%. That is $202.5m of interest a year, and under an absolute triple net structure rent must cover it. Real rent sits higher: it carries the developer’s equity return, fees and reserves too.

Electricity is where I was wrong. I expected a rounding line; on 149MW of critical IT load it isn’t, so I priced it. The EIA’s Electric Power Monthly, Table 5.6.A puts Oklahoma industrial electricity at 7.37 cents per kWh in June 2026 against 6.57 a year earlier, a tariff rising 12% a year that the tenant absorbs uncapped. Two cases, bracketing the uplift for cooling and losses:

Add them. I make the committed annual floor $289m to $311m: $202.5m of rent plus $87m to $108m of power, before the property taxes, insurance and maintenance the lease also assigns to Jane Street and which public data cannot size. Fifteen years, undiscounted: $4.34bn to $4.66bn, or 69% to 74% of the $6.28bn of equity in the broker-dealer that does the trading. Now put the two August deals side by side.

The obligation nobody wrote about runs between 1.65 and 1.78 times the size of the one everybody did. One caveat, because it isn’t like for like: $175m is the INCREMENTAL cost of the refinancing while $289m to $311m is the lease’s GROSS charge. Incremental is the right basis here because the subject is what Jane Street ADDED in 2026: the refinancing added $175m, the lease at least $289m. My load and PUE assumptions are moveable and a reader should move them, though not far enough to reverse this: set the power bill to zero and the rent floor alone still exceeds $175m. That’s the gap.

Read more

Griffin’s own letter proves the market misread the July 30 trade. My read: distributing the risk was the plan from day one. Citadel says so itself. The receipt is public: more than 80% of the risk gone in three weeks, while the stocks Citadel exited kept rallying without it.

Photo: Federal Reserve Board · Public domain, adapted · via Wikimedia Commons

This piece is also a film: sixteen minutes working Griffin’s own letter against the block trade tape, the capital base arithmetic worked on screen against both of Citadel’s measures, and the Q3 falsifier stated on camera with its date.

The consensus, stated at its strongest

On July 30, four names in Leopold Aschenbrenner’s forced sale book ripped 21 to 28% in a single session. The market read that as Citadel underwriting the AI infrastructure trade. I wrote about the mechanics of that day three weeks ago, and one detail from it still holds up: a name that never appeared in Situational Awareness’s disclosed holdings rallied just as hard as the ones that did. That argued for an overhang clearing. A fresh conviction call reads differently, and I trust the cross section over the headline. Still, I want to steelman the confidence reading properly here. It deserved a hearing then, and it deserves one now.

The backdrop makes it easy to believe. Situational Awareness had returned 439% through the first half of 2026 on leveraged AI infrastructure bets, then lost roughly 67% in July alone, as SanDisk fell as much as 56% and Nebius dropped about 48% from its peak. Assets went from as much as $45 billion at the start of July to roughly $10 billion after the sale, half of that a private Anthropic stake the fund never sold. Its three prime brokers, Goldman, JPMorgan, and Bank of America, had issued margin calls, and the book was being shopped with nobody stepping up. Then, the morning after the Federal Reserve held rates at 3.50 to 3.75% on a 9-3 vote, Citadel bought the whole thing in a single block. A firm managing hundreds of billions in regulatory assets paid real size for a book three of the largest banks on the Street could not place. That is not commentary. That is capital at risk, and reasonable investors updated on it. The story was clean. Almost too clean, in hindsight.

Griffin’s own words at the time reinforced the reading. ZeroHedge’s account of the same letter has Citadel describing itself as front footed over its nearly thirty six year history, arithmetic that checks out against a firm founded in 1990. A firm that wants credit for a fundamental call talks like that. It does not usually talk that way about a service it planned to resell inside three weeks. I take that phrasing seriously as a data point, even though I think the letter three weeks later reads differently.

I am not going to argue that reading was foolish. My claim is narrower: it was incomplete. Griffin’s own words on August 21 are the evidence that closes the gap.

What the letter actually says

Line up the dates first. July 29: the Fed holds. July 30: the block trade. August 14: the Q2 13F confirms the book. August 21: Griffin’s letter names the exit. Four dates, three weeks, one story. Consider the sentence that matters, from Griffin’s letter dated Friday, August 21, reproduced identically across CNBC, Yahoo Finance, and two further wire renderings: “Our ability to distribute this risk was central to our investment thesis. These moments highlight our ability to quickly evaluate complex risks, deploy capital with speed and conviction, and execute with precision.”

The word doing the work is “distribute.” Not “hold.” One verb. That is the whole disclosure. Citadel is not telling its limited partners it bought conviction in memory chips, data center capacity, or power generation. It is telling them the thesis was the transaction itself: buy a forced sale book at a discount nobody else could underwrite in the window available, then sell it back into a market that had time to normalize. The asset was never the point. The interval was.

The receipts back the letter. The July headlines do not hold up next to them. Griffin says Citadel completed nearly 100 block trades worth more than $4 billion in market value, including the largest intraday block trades of the year in 10 separate names, and shed more than 80% of the aggregate risk it took on. Those two figures describe different things and the letter never reconciles them: $4 billion transacted, 80% of risk, and Citadel does not publish how it weighs the second, so I cannot recompute one from the other. That is a full public turn on the risk, in three weeks, on a position the market spent a week in late July calling a conviction bet. I do not think a firm distributes 80% of a genuine long term fundamental call that fast. I think it distributes a warehoused position exactly that fast, because every day it holds one is a day of uncompensated exposure to a name it never wanted to own past the point where somebody else would.

What makes this checkable, beyond merely plausible, is what Situational Awareness’s own Q2 13F shows. Filed August 14 for the June 30 snapshot, the same filing that reported its total value twice on the same day, a separate reporting problem I covered on its own. Its five largest disclosed positions were SanDisk, Micron, Bloom Energy, Taiwan Semiconductor, and Nebius: a $20.2 billion book across 25 names. Five of those are the same names multiple outlets report Citadel bought, out of the 10 total names Griffin’s letter says the block trades spanned. The forced seller’s own filing and the forced buyer’s own trades line up. Same five names. Same order of size. I do not read that as a coincidence worth a footnote. I read it as confirmation that the book everyone is discussing is the actual book, which is why I trust the block trade figures in the August letter enough to build an argument on them.

The mechanism: why “distribute” beats “hold” here

Grossman and Miller’s 1988 result in the Journal of Finance is the right starting frame, and I am reaching for it again for the reason I reached for it three weeks ago. Market liquidity is the demand and supply of immediacy. A market maker gets paid for bearing risk during the interval between a seller who must transact now and a buyer who has not shown up yet. He is not paid for being right about the asset. He is paid for standing in the gap.

Citadel stood in that gap on July 30. Situational Awareness needed to sell an entire book, longs and shorts together, before the next margin call. The natural buyers of levered AI infrastructure equity were the other funds running the exact same trade, nursing the identical July drawdown. The bid did not disappear because the stocks were bad. It disappeared because everyone who would normally set the price was impaired at once. That is Shleifer and Vishny’s 1992 mechanism, and it is why a 10% discount to already depressed prices was available to whoever showed up with clean capital. The empirical literature on so called upstairs block markets backs the shape of this: large negotiated blocks executed away from the open order book routinely carry a temporary discount that reverses once the immediacy premium is paid off, distinct from a permanent repricing.

Here’s the part I think most coverage missed. The August letter and the July trade explain each other; each one alone tells half the story. A pure liquidity trade has a specific, falsifiable signature: a short holding period, uncorrelated with the asset’s own catalysts, ending when the forced sale discount closes rather than when the story plays out. Compare that against what actually happened. Bloom Energy posted a strong earnings quarter two days before the block trade. Microsoft’s July 30 earnings, a 43% cloud revenue growth print that added roughly $480 billion of market value in a single session, landed the same morning as Citadel’s block. None of that is Citadel’s research edge. It’s the market’s own catalyst calendar, arriving on schedule. Citadel supplied none of it. Citadel got to sell into every bit of it.

Read more

The claim: TCI’s newly disclosed $636mn Italian hotel debt book, widely read as diversification away from Chris Hohn’s famous concentrated equity bets, is the same single name concentration philosophy in a different wrapper.

Sir Christopher Hohn built TCI Fund Management into a $77bn hedge fund doing one thing better than almost anyone else alive. He finds businesses with pricing power, buys a handful at a size most funds would call reckless, and holds through the volatility. Eleven US listed equities carry $52.8bn of that book as of TCI’s most recent quarterly filing. GE Aerospace alone is roughly a third of it (TCI’s Q2 2026 13F-HR via 13f.info). That concentration produced a record $18.9bn net gain for investors in 2025, the largest single year dollar profit any hedge fund has ever booked, and an estimated $4.9bn for Hohn personally, the most any hedge fund manager has ever made in one year (Institutional Investor). The same concentration cost the fund 11.9% in a single month this March, when the US strike on Iran hit GE Aerospace, Visa, Microsoft, Moody’s and S&P Global all at once. The quarter closed down 9.4% (Institutional Investor). A brutal month, inside a good year.

So when the Financial Times reported this month that TCI has built a $636mn book of loans secured against four of Italy’s most exclusive hotels (Financial Times; reproduced in full by Hedgeweek and independently covered by Il Sole 24 Ore), the coverage read it as continuity, not departure: Hedgeweek notes the hotel book “fits closely with Hohn’s broader investment philosophy” of pricing power and scarce supply, the same case he makes for GE Aerospace. That framing is correct on strategy. My disagreement is narrower: continuity of PHILOSOPHY is not the same claim as diversification of RISK, and the second claim is the one that matters for sizing this.

The bull case: four hotels, four brands, uncorrelated income

The case for calling this diversification is a good one, and I want to state it before I argue against it. TCI’s $392mn position sits in debt against the Danieli in Venice, a fifteenth century palazzo converting from Marriott to Four Seasons. A further $132mn backs the Caesar Augustus in Capri, $74mn sits against the Six Senses on Lake Como, and $38mn against the Mandarin Oriental in Milan (Hedgeweek). Four different cities. Four different brands. Four different loan sizes. On a spreadsheet, that reads as a portfolio. Not a position.

I will grant the asset class itself is structurally uncorrelated to TCI’s equity book, and that matters. Real estate debt pays cash coupon income. It sits senior against hard collateral. It does not reprice every morning against an Iran headline the way GE Aerospace does. Italy’s luxury hospitality market backs the thesis with real numbers too: RevPAR across the country rose 53% between 2019 and 2025, the strongest of any European market, with average daily rates up 55% over the same stretch, per Cushman & Wakefield’s year end 2025 report (Hospitality Net).

Scarce five star inventory in cities wealthy travelers keep visiting is the same pricing power Hohn has spent two decades buying in public markets, just financed through debt instead of equity this time. TCI is not originating these loans itself, either. It takes participations in loans arranged by TCI Real Estate Partners, a credit shop Martin Fräss-Ehrfeld has run since 2011, which gives Hohn’s fund a specialist underwriter’s judgment on every deal rather than its own (Hedgeweek). Read only the headline number, and a sophisticated allocator walks away thinking TCI just bought itself some ballast.

The variant view: one sponsor owes every dollar

My read is different, and it rests on one fact none of the initial coverage foregrounds. All five loans, in both countries, sit with a single sponsor: Gruppo Statuto, the hotel group controlled by Giuseppe Statuto (Hedgeweek; BeBeez). This isn’t a new relationship TCI is testing with a diversified basket of first time borrowers. In my reading, it’s the deepening of a credit exposure that is a decade old, concentrated in a single name, and has already lived through one shock.

TCI financed Statuto’s purchase of the San Domenico Palace in Taormina in 2016. It was already holding debt linked to the Danieli with a 2022 maturity well before this year’s expansion, per BeBeez, the Italian trade publication that tracks the country’s private debt market (BeBeez). By May 2020, with Covid lockdowns having emptied every hotel in Italy at once, Statuto needed to recalibrate its capital structure across the whole group. TCI sat alongside Banco BPM (roughly €500mn of exposure, down from €900mn three years earlier), Amco (around €126mn inherited from the failed Veneto Banca, one of two regional lenders the Italian state wound down that year (European Central Bank)) and bonds held by Cale Street Partners’ own investment vehicle, all exposed to the same sponsor at the same moment. I read that as the closest real test this relationship has faced, though I should be precise about what it tested: Covid closed every hotel in Italy at once, regardless of who owned them, so this shock arrived at the sector level, not as something unique to Statuto. What it still shows is that Statuto’s OWN response to a shock was coordinated across the whole group in one negotiation, not handled property by property. The shock didn’t stay contained to one property. It hit the sponsor’s whole book at once, whatever triggered it. One negotiation, not five.

I make the current exposure $698mn, not $636mn, because the FT reporting Hedgeweek reproduces adds a fifth position: $62mn against the Six Senses in Ibiza, also owned by Gruppo Statuto, sitting outside the Italy total the headline figure describes (Hedgeweek). $392mn plus $132mn plus $74mn plus $38mn. That is $636mn before Ibiza even enters the count. Nobody covering this story so far has added that fifth number to the other four. Once you do, the four city framing understates the position by almost ten percent, and the “diversified across Italy” framing drops a hotel that isn’t even in Italy. Five trophy assets. Two countries. One name on every note. That matters.

And this isn’t an allocation TCI made at arm’s length into somebody else’s fund. TCI Fund Management Limited’s own SEC Form ADV, amended June 29, 2026, lists TCI Real Estate Partners Limited and four associated general partner entities, one per fund vintage, as related persons of the firm (SEC Form ADV, TCI Fund Management Limited, CRD #269954). A 2023 investment memorandum Pennsylvania’s public pension system published when its staff recommended a $200mn commitment to TCI Real Estate Partners Fund IV names the firm behind the vehicle as TCI Fund Management Limited itself, with Fräss-Ehrfeld running origination and underwriting day to day (Pennsylvania PSERS memorandum, TCI Real Estate Partners Fund IV LP, August 2023). The same memo discloses a key person clause naming both Fräss-Ehrfeld and Hohn: if either stops being involved, the fund’s investment period ends. I read this as Hohn’s own balance sheet, run by a specialist he backs. His book. Not a diversifying check into someone else’s.

Why five hotels behave like one credit

Here is the part that turns a coincidence of ownership into an actual risk claim, not just an observation.

Concentration risk in a lending book is not primarily about how many loan agreements exist. It is about how many truly independent sources of repayment sit behind them. A lender holding four mortgages on four unrelated homeowners has four independent credit risks. One borrower’s job loss does not touch the other three. A lender holding four mortgages on four hotels owned, managed and centrally financed by one family group has, in the scenario that actually matters, one credit risk wearing four collateral packages. The number of loan documents tells you about legal structure. It tells you nothing about correlation. And correlation, not the document count, decides whether a shock produces one bad quarter or a portfolio event.

The 2023 PSERS memo makes this point explicit for the fund structure generally, not for the Statuto loans specifically. It names “Concentrated Portfolio” as a disclosed risk category for TCI Real Estate Partners’ funds, which run just eight to twelve underlying loans at any time with an average size of roughly $200mn each. Its stated mitigant is “the focus on high quality properties in prime locations… with experienced, institutional quality equity sponsors and borrowers” (PSERS memo). That mitigant addresses collateral and sponsor quality. It says nothing about sponsor correlation, and that, in my read, is exactly the gap the Statuto book is now testing. When four or five of a small fund’s loans trace back to one sponsor, the fund’s own stated diversification, eight to twelve names, shrinks toward one in precisely the dimension that matters under stress. Eight to twelve becomes one.

I worked through this same mechanism at more length in a separate note, running the sponsor concentration math against TCI Real Estate Partners’ own stated averages: TCI’s $698M Statuto book, on how eight to twelve names becomes one.

What ring fencing blocks, and what it cannot

This is the strongest objection the piece has to answer. Institutional real estate lenders structure a book like this as without recourse, ring fenced special purpose vehicles, so one asset’s distress cannot legally reach another’s collateral. That structuring is the industry default, and I have no evidence these five loans depart from it. That argues against my own framing, and I am not going to dress it up.

Ring fencing blocks legal cross default. A missed payment at the Caesar Augustus cannot, on its own, put the Danieli into default. It does nothing to block correlated performance. What generates each loan’s cash flow is not just the building. It is Statuto’s management bandwidth, its brand conversion execution, and its access to the next refinancing, all priced at the sponsor level whatever the loan documents say. Own five ring fenced loans to five different sponsors, and a stumble by one team says nothing about the other four. Own five ring fenced loans to one sponsor running three simultaneous luxury conversions, and a stretched management team is the one variable every loan shares. Ring fencing or not.

The Danieli position tests that construction discipline directly. $392mn against a stated average of $200mn is not a marginal overweight. I make that 392 divided by 200, or 1.96, close enough to double the size TCI Real Estate Partners itself uses to describe a normal position in this strategy (PSERS memo). A fund that sizes one loan at roughly two average positions, inside a book that already runs on only eight to twelve names, has made a concentrated bet inside an already concentrated strategy. I read the extra sizing as conviction in the Danieli’s brand conversion specifically, not an accident of underwriting. It is also, by definition, the single loan most exposed if that conversion runs into execution trouble before 2027.

I would draw the same distinction Hohn’s own equity book forces on anyone analyzing it. GE Aerospace, Microsoft, Visa, Moody’s and S&P Global are five different tickers. They stop being five independent risk factors the moment a shock like a war headline hits every large cap US industrial and financial name at once, which is exactly what happened in March, when GE Aerospace fell 17% and dragged the fund down 11.9% for the month (Institutional Investor). I do not think Hohn is averse to diversification by accident. He has run an 85 percent in five names US book for years and built his career on the returns concentration can generate when it is sized right. My point is narrower than “concentration is bad.” The same manager applying the same logic to hotel debt has built something concentrated on the one axis, sponsor credit, that a five city, two brand loan book is least likely to disclose on a first read.

Signa, and the €623 million lesson in what “diversified” hid

There’s a live example of what happens when this exact structure fails, and it’s instructive precisely because the scale is so much larger. René Benko’s Signa Group collapsed into insolvency in late 2023 across a reported 1,000-plus interlinked entities (Euromoney), with Signa Holding, the entity at the top of the structure, later facing €8.35bn in creditor claims (Global Banking & Finance). Julius Baer alone booked a €623mn loss tied to the group (PEI Private Real Estate). Signa’s portfolio spanned trophy hotels, department stores and offices across several countries under separate legal names, exactly the structure that let lenders feel comfortable extending credit against each asset one at a time. Euromoney’s read of the collapse names the mechanism plainly: “debt service costs are going up while property values are going down,” a shock that hit across a structure so opaque that creditors were “struggling to trace the flow of funds between what may be up to 1,000 interlinked entities,” every one of them tracing back to Benko’s control (Euromoney). Statuto is not Signa. Add up TCI’s own $698mn, Banco BPM’s roughly €500mn and Amco’s €126mn, low billions across Statuto’s named lenders, against Signa’s €8.35bn in creditor claims and 1,000-plus entities, and nothing suggests Statuto is near that scale of complexity or distress. But the mechanism that broke Signa’s lenders, a sponsor level shock reaching every asset sharing that sponsor’s balance sheet at once, is the same class of risk the 2020 Statuto episode showed at a smaller scale, even granting Covid as the trigger rather than something Statuto-specific. Scale differs by orders of magnitude. The mechanism does not. Same shape, smaller frame.

There’s a structural reason this kind of book exists at all, and it explains why TCI keeps adding to one name instead of spreading $700mn across ten unrelated hotels. Bank lending against European commercial real estate has been retreating since the post-2022 rate cycle, and part of Statuto’s older debt, per BeBeez, traces back to Veneto Banca, one of two regional lenders the Italian state wound down in 2017. The PSERS memo names the opportunity for a lender like TCI Real Estate Partners directly: wider spreads and lower loan to value ratios, available because “retrenchment of regional banks and heightened liquidity pressures facing many levered lending platforms” have thinned the field (PSERS memo). That’s a real edge, and every private credit fund raised since 2022 is chasing it. It doesn’t change my argument. Spread is not safety. It never was. A bank retreating from Statuto’s debt for its own capital reasons is a different signal than a bank concluding Statuto’s credit is safe.

The repeat borrower defense, and where it stops working

The obvious rival explanation is that this is simply how private real estate credit works. Concentrated lenders routinely back repeat borrowers, because a repeat relationship gives the lender better information than a stranger would, and a sponsor with a proven record through one cycle really is lower risk than an unknown one. TCI Real Estate Partners has completed more than $14bn of transactions since 2011 with what the PSERS memo calls “an attractive loss ratio,” and Fund III generated a 10.8% net internal rate of return against a leveraged loan benchmark as of March 2023 (PSERS memo). On that reading, five loans to Statuto aren’t a concentration failure. I’d call them the payoff from a decade of diligence on a borrower TCI now understands better than any new counterparty it could underwrite cold. A decade of homework, not a shortcut.

What separates my reading from that one is the size trajectory, not the relationship itself. Financing a trusted borrower again and again is a different claim from watching that exposure grow from one Sicilian hotel in 2016 to five trophy assets and roughly $700mn today, much of it in one asset still under renovation. Trust earned through history justifies extending credit to a known borrower. On its own, I don’t think it justifies letting that borrower become the largest disclosed concentration inside a fund whose own paperwork names concentration as a risk category. My reading and the rival one agree on every fact here. Somewhere between one hotel and five, relationship became risk. That is the whole disagreement. Nothing more, nothing less.

What I can’t verify

I can’t see the loan documents, and that matters more here than in most pieces I write. I’ve already granted that without recourse, ring fenced structuring is the industry default for exactly this kind of book, and I have no evidence these five loans depart from it. What I can’t verify is anything finer than that default: whether Statuto’s facilities carry any cross collateralization or shared covenants at all, and whether TCI’s own internal risk limits treat this as one name or five for capital purposes. Both would matter to how correlated the legal exposure actually is, and both are private. Both stay private.

I also can’t verify what share of each loan TCI itself holds versus its other investors, since every account describes TCI’s positions as “interests” and “stakes,” never as sole exposure. The $392mn Danieli figure could be TCI’s full participation or one tranche of a larger loan. Nor do I have Statuto’s current balance sheet. BeBeez’s reporting that the group needed to “recalibrate its capital structure” is from May 2020, inside the Covid shock described above, not a current finding, and I have not found a 2026 source making the equivalent claim.

What would change my mind

My falsifier here is concrete and dated. If Danieli’s performance after renovation, once the Four Seasons conversion finishes in 2027, tracks or beats the sector’s RevPAR growth while Statuto’s other four properties keep servicing debt normally, that’s direct evidence the loans behave as independently as the headline count implies, and my correlation claim is wrong. If Statuto instead goes back to its lenders for another capital structure conversation, the way it did in 2020, and that conversation touches more than one of the five properties at once, that confirms the loans move together exactly as I argue. I’d also revise this view if TCI’s next Form ADV amendment shows the real estate sleeve holding steady in size rather than growing.

Three things to watch, none of them a trade

A PM tracking TCI doesn’t need to short anything here to act on it. The trade, as I see it, is watching, not positioning. I’d track three things over the next four quarters. Whether Danieli’s occupancy and rate performance after the 2027 reopening validates the RevPAR thesis in isolation. Whether any new TCI Real Estate Partners vintage adds further Statuto exposure or deliberately caps it. Whether Banco BPM, Statuto’s other major lender, shows any sign of moving its own exposure, though its current number isn’t public. The only figure on record is BeBeez’s 2020 read, so this is a lead to chase rather than one to track precisely. A widening gap between what TCI is doing and what its other lenders are doing would tell you more than any single loan level disclosure can. Divergence is the tell. Watch for it.

My question isn’t whether Hohn still believes concentration beats diversification. Obviously he does. My question is whether anyone reading the coverage of this deal has actually noticed he just made the same bet again, in an asset class where the collateral is marble and the correlation stays invisible until the sponsor’s cash runs short across every postcode at once.

I post this same research on YouTube and LinkedIn as it publishes.

The Decision-Grade Version

This piece is complete on its own. The thesis, the evidence, and what would kill the view are all above, and nothing was held back to sell you a next step.

The Patreon note is a separate piece of work rather than a deeper cut of this article. It works through the sponsor concentration mechanism itself against TCI’s own stated fund averages, with the ring fencing counterfactual and two other limitation cases argued at more length than this article has room for. Written for people who put capital behind a view.

→ Read the TCI sponsor concentration note

→ Or join the Patreon community for every note

Where do you draw the line between a trusted repeat borrower and a concentrated single name credit risk, when the collateral changes but the name on every note stays the same?

Cover photograph: Armin Kleiner, CC BY-SA 4.0, via Wikimedia Commons.

Citadel, D. E. Shaw, Elliott, and Farallon each answer the SEC’s independent valuation question zero percent on every private fund they report, three years after SEC staff said the correct answer for their exact fact pattern is 100 percent.

Every SEC-registered adviser to a private fund answers the same question every year: what share of the fund’s assets, by value, was priced by somebody who isn’t you. I pulled the current Form ADV for eighteen of the largest multi-strategy and quantitative managers in the business and read every answer myself, fund by fund. As far as I can tell, nobody has ever tabulated them. What I found is a split so clean it worried me at first. Fourteen houses run overwhelmingly at 97 to 100 percent, real exceptions included. Four cluster at exactly 0, on every fund they report, with no exceptions at all. And the SEC’s own staff has already told the industry, in writing, which of those two answers this fact pattern is supposed to produce.

The question everyone answers, and almost nobody reads

Schedule D, Section 7.B.(1), Question 27 of Form ADV asks each private fund a version of the same thing: during the last fiscal year, what percentage of the fund’s assets, by value, was valued by a person who is not your related party, such as an administrator. The instruction attached to the question is specific about what counts. It only counts if that outside person carried out the valuation procedure the fund actually uses. And it only counts if the resulting number is the one applied to investor subscriptions, redemptions, and fee calculations. A shadow number nobody trades on doesn’t qualify. I read the underlying instruction directly off the SEC’s own Form ADV general instructions, which tell advisers to “treat all of the assets of a private fund as a securities portfolio” and never deduct liabilities. Question 27 is written with the same precision.

That specificity matters, because the SEC has already ruled on the exact ambiguity a 0% answer could be hiding behind. In an FAQ the agency posted on October 26, 2023, staff addressed an adviser who reasoned this way: my administrator prices 100% of the fund, but I’m ultimately responsible for the number, so shouldn’t I enter 0? The staff’s answer, verbatim: “In the staff’s view, an adviser should enter 100%.” I confirmed this myself, on the live SEC site. sec.gov blocks automated fetches outright. A plain HTTP request and a scripted fetch both come back with a flat 403. I had to load the page in a real browser session and pull the text off the rendered page. The guidance has been public for almost three years.

One caveat on the guidance itself, stated plainly: it is a staff FAQ, not a Commission rule. Nobody is bound to follow it, and I found no enforcement action or deficiency letter citing any adviser for answering Question 27 the way the FAQ says not to. Three years of an unchanged 0% is three years of not adopting non-binding guidance, not defiance of a rule. I have tried to keep that distinction in view throughout.

A 0% answer, filed by a firm that names a real, unaffiliated administrator, is not a neutral disclosure. It’s an assertion. It says the administrator’s valuation work is not the number the fund’s own investors trade on.

What “a good administrator” is assumed to mean

Ask most allocators what an independent administrator buys them and you’ll get the standard operational due diligence answer. A big, recognizable name like Northern Trust or Citco or State Street on the fund’s paperwork means an outside party is checking the manager’s marks. Full stop. Writing on ODD practice, like this overview of the discipline, asks who the administrator is and whether it’s independent. It rarely asks whether that number is the one that actually prices a subscription or redemption, and the industry has generally assumed those are the same question. Two law firm alerts, from Mayer Brown and Simpson Thacher, told compliance departments the 2023 guidance existed. Neither asked which large advisers had actually changed their answer. That’s the strongest form of the consensus: a large, unaffiliated administrator is the disclosure that matters, and a single percentage field on page 40 of a 300-page form is a compliance detail nobody outside the back office reads.

I don’t buy it once you look at what the four 0% firms are actually reporting, fund by fund, alongside that one answer.

The variant view, and what forces it

Citadel Advisors LLC’s most recent Form ADV, filed June 11, 2026 (the same filing where Citadel’s own AUM runs eight times the headline number), reports thirty-seven private funds. I extracted every one of them from the filing’s Schedule D myself, matching each fund’s gross asset value and Question 27 answer to its own top level entry. Master-feeder structures nest sub blocks inside a master fund’s own record, and a naive text search will happily attach the master’s dollar figure to the wrong feeder’s name. I built my extraction around the form’s own item numbering to avoid exactly that trap, and I paid for the caution once already: my first pass misassigned Millennium’s $687.7 billion flagship to a feeder fund called Millennium USA LP. I had to trace the item hierarchy back to find its real owner, Millennium Partners, LP, the entity that actually carries item 27’s 100% answer alongside 1,697 beneficial owners. The full six-step version of that method, and a nineteenth adviser I checked against it after this piece first went out, are in a companion note on Patreon.

Every one of Citadel’s thirty-seven funds answers 0%. That’s $1,094,345,899,742 in combined gross asset value. Every dollar of it. Including the $331.3 billion Global Fixed Income Master Fund, Citadel’s single largest reported fund. Including the $30.9 billion Wellington fund, which reports 309 beneficial owners and a named administrator, Northern Trust Hedge Fund Services LLC, that the filing says sends account statements to every one of them. I pulled up that specific page as a rendered image instead of trusting the extracted text, because Question 26’s sub answers are radio buttons, not typed fields, and extraction can silently drop which button is actually marked. The image confirms it. “Is the administrator a related person of your firm?” is marked No. “Does the administrator prepare and send investor account statements?” is marked Yes, provided to all investors. Question 27, right below both, still reads 0%.

The other three all-zero houses

D. E. Shaw’s current filing shows the identical shape: thirteen funds with a reported gross asset value, all thirteen answering 0%, $213.4 billion combined, Citco Fund Services (Cayman Islands) Ltd named as administrator on every one. Elliott Investment Management reports thirteen funds, all 0%, $128.6 billion, administered by Harmonic Fund Services, with Bank of New York Mellon also named on its two largest. Farallon Capital Management reports forty-eight funds with a gross asset value on file. Every single one is 0%. $64.2 billion combined. Twenty of the forty-eight name State Street (Cayman) Trust Limited or International Fund Services as administrator; the other twenty-eight report none at all, which makes a 0% answer on those twenty-eight unremarkable, since there is nobody to override. It is the twenty with a real, named, unaffiliated administrator that carry the same weight as Citadel’s and Elliott’s. Four firms. Four different unaffiliated administrators, at least where one is named. One identical answer, repeated across a hundred and eleven separate fund filings, with zero exceptions among them.

I checked the other side of the comparison with the same discipline, because a finding this clean is exactly the kind that turns out to be a parsing artifact. I downloaded and parsed fourteen more advisers, matching item hierarchy the same way: Millennium, Point72, Two Sigma, Bridgewater, AQR, Schonfeld, Balyasny, ExodusPoint, Squarepoint, Verition, Tudor, Voleon, Third Point, and Capstone. Across those fourteen, 511 funds report a Question 27 answer, for 622 across the full eighteen-adviser sample combined. Two Sigma answers 100% on all sixty-two of its funds, from its $38.3 billion Absolute Return Portfolio down to its smallest vehicle. Bridgewater, all thirty-eight, including its $48.4 billion Pure Alpha Trading Company II. Schonfeld, thirty-nine of thirty-nine, topped by a $171 billion master fund. Balyasny, forty-three of forty-three. ExodusPoint, Squarepoint, Voleon, and Capstone answer 100% on every fund each of them reports. AQR reports 100% on 199 of its 203 funds; the four exceptions sit at 91 to 99 percent, never zero. Verition answers 99% flat across all five of its funds. Tudor is the real outlier on this side of the comparison: ten funds ranging from 100% down to a single fund at 3%, a genuine spread inside one firm’s own book. That’s the clearest illustration in my whole sample of what an honest, asset-driven answer looks like when it varies. Real operational variation shows up as a spread. It doesn’t show up as a wall.

Point72 is the one mixed house in my whole sample, and it’s instructive precisely because it isn’t clean. Point72 Associates, LLC, its $266.3 billion flagship, answers 99%. Twelve of its eighteen funds run 99 to 100%. But six smaller Point72 vehicles answer differently: Point72 Credit at 46%, three Point72 Hyperscale entities at 0% (running $75 million, $283 million, and $67 million), and two more, CPV Real Estate Holdings at 0% ($22 million) and CPV Holdings II at 0% ($1.52 billion), all presumably because those specific funds hold assets the administrator can’t price. That’s what an honest, asset-driven answer looks like. It varies with what the fund actually holds. It moves fund by fund, the way a real operational difference should, and Tudor’s 3-to-100 spread makes the same point from a different firm. Citadel, D. E. Shaw, Elliott, and Farallon never vary. Not once, across 111 funds combined, spanning fixed income, equities, credit, and multi-strategy books alike.

I went a step further and checked whether Point72’s spread might explain the broader pattern on its own, since it’s the clearest case in my sample of a firm reporting both answers honestly. It doesn’t. Point72’s zero-percent funds are mostly small relative to its $266 billion flagship, with one real exception, the $1.52 billion CPV Holdings II. Citadel’s zero-percent funds are not small either. They include the largest fund among the four all-zero houses, at $331 billion, larger than Point72’s own flagship and every one of its exceptions. Scale doesn’t explain a firm answering 0% either way; if anything, the biggest fund among the four zero-percent complexes is the one where an outside valuation would matter most to the investors who hold it.

Read more

I’ve spent enough time inside Form ADV filings to know most of what Item 26 and Item 27 turn up is boring. An adviser either pays an administrator or it doesn’t, and plenty of funds answer “doesn’t” every filing cycle with nothing interesting behind the answer. Renaissance Technologies is one of the advisers that answers “doesn’t” for at least one of its funds. What makes its answer worth an article isn’t that Medallion Holdings LLC has no administrator. It’s that Renaissance’s own Form ADV, filed July 1, 2026, on the same schedule, in the same section, under the same signature, tells the SEC something completely different about the two funds that hold outside money.

The legend assumes the plumbing works

The version of Medallion everyone already knows is a closed shop. It stopped taking outside capital in 1993 and, by 2005, had bought out its last remaining outside investor. As of late 2016, Bloomberg reported the fund was open only to roughly 300 Renaissance employees and their families, of whom about a hundred qualified as “qualified purchasers” (generally $5 million or more to invest), with the rest admitted as “accredited investors” (generally $1 million or more). Both figures are a decade old, the last public snapshot of who was actually in the fund. That’s a structure that answers to nobody but its own payroll, and the strongest version of that story is worth taking seriously before I depart from it: a fund with no outside limited partners has, on paper, no outside limited partners to protect. Nobody’s redeeming a pension fund’s capital against a number Renaissance made up, and if the same people setting the number want to mismark their own retirement savings, I’d argue that’s their business, not the SEC’s.

I think that argument is wrong on the facts, and the filing is what shows it. Medallion isn’t the roughly $10 billion fund most write-ups still describe. Renaissance’s July 2026 ADV puts its gross asset value at $61,177,067,368, spread across a master fund and four feeders: Medallion Associates L.P., Medallion Fund L.P., Medallion International Ltd., and Medallion USA L.P. All five are reported under a single Schedule D Section 7.B.(1) filing, exactly what the SEC’s own instructions permit for a master-feeder arrangement. And it isn’t run for roughly 300 people any more either. The same filing reports 549 beneficial owners, nearly double the 2016 headcount and more than five times the roughly one hundred qualified-purchaser employees Bloomberg reported a decade ago. Item 14 puts Renaissance and its related persons at 80% of that pool, by value. The remaining 20%, by my own arithmetic on the disclosed gross asset value, is roughly $12.2 billion ($61.177bn times 0.20) beneficially owned by people the filing doesn’t classify as related to the adviser.

The minimums Renaissance sets tell the same story a different way. Medallion’s own Item 12 answer lists a $25,000 minimum investment commitment. RIEF and GF each require $5 million. A fund priced two hundred times cheaper to enter than its institutional siblings is built for broad internal access. It isn’t a handful of founders parking bonus checks, and my read is that a $25,000 minimum only makes sense for a fund Renaissance expects hundreds of ordinary employees to actually use.

One filing, two numbers: 0% and 99%

Here is the part that does not fit the closed-fund story. Item 27 of Schedule D asks a specific question: what percentage of a fund’s assets, during the last fiscal year, was valued by “a person, such as an administrator, that is not your related person,” where that person’s valuation is the one actually used for subscriptions, redemptions, distributions, and fee calculations. For Medallion Holdings LLC, Renaissance’s answer is 0%. Item 26(a), a checkbox asking whether the fund uses an administrator other than Renaissance itself, is answered No. That answer did not come from a text scrape, which cannot reliably show which radio button is filled on a form like this. It was confirmed by pulling the filing and checking page 32 of Renaissance’s PDF directly.

Two entries later in the same document, Renaissance answers the identical pair of questions for RIEF Trading LLC, gross assets $21,767,779,497, and GF Trading LLC, gross assets $4,214,510,777. Both report an administrator, The Bank of New York Mellon, named and located in New York. Both report 99% of assets valued by that non-related administrator. RIEF’s owner count is 705, only 8% of it owned by Renaissance and related persons. GF Trading has 193 owners, 43% related-party. These are the Renaissance funds that raise money from pensions, endowments, and other institutional allocators. I checked whether this pattern is idiosyncratic to Renaissance’s paperwork or closer to a house style across the industry, because a single adviser’s quirk is a much weaker story than an industry pattern. It is not unique to Renaissance. Citadel Advisors’ own ADV, filed June 11, 2026, three weeks before Renaissance’s, shows the identical shape at smaller scale across its employee vehicles: CEIF Partners LLC at $1.83bn, 95% adviser-owned, 0% independently valued; CEIF International Ltd. at $1.42bn, 407 owners, 0%; CEIF Partners International Ltd. at $144m, 0%; and Citadel Voluntary Investment Fund Ltd. at $204m, 163 owners, 100% adviser-owned, 0%. Four employee-facing funds, four identical answers, at a firm with no public connection to Renaissance beyond filing the same form the same way.

My leading read is that this is a genuine double standard, not a coincidence of fund type, though I lay out a real competing explanation later in this piece and I don’t think it’s fully resolved by what’s public. Renaissance can clearly run a fund with an outside administrator checking 99% of its book. It does exactly that for RIEF (Renaissance Institutional Equities Fund) and GF. It does not do it for the fund that is nearly three times RIEF’s size and pays its own people.

How a master-feeder filing turns into a single percentage

The SEC’s own instructions to Form ADV explain why one number can stand for a $61 billion complex spanning five separate legal entities. Under Instruction 6.d, an adviser filing a single Section 7.B.(1) for a master-feeder arrangement may report Items 13 through 16 by “aggregating all investors in the master-feeder arrangement,” and Item 11’s gross asset figure covers “the master-feeder arrangement as a whole.” That is not a loophole. It is the documented, intended mechanism, and it applies identically to Medallion, RIEF, and GF. The comparison is apples to apples: three master-feeder complexes, one adviser, one filing date, three different answers on who checks the number.

What does Item 27’s percentage actually certify? The SEC’s own standard is worth checking, because “administrator” can mean a rubber-stamp mailbox or a real check. Division of Investment Management staff guidance on the Form ADV FAQ addresses this directly: an adviser should report the full percentage of assets valued by an independent administrator on Item 27, even where the adviser still considers itself ultimately responsible for the valuation. This describes the substance of that guidance rather than quoting it verbatim, since a clean fetch of the SEC’s own FAQ page to check its exact wording was not obtainable this session (sec.gov blocks automated access). That guidance treats this field as a real measure of whether an outside party actually performed the valuation work behind what an investor’s stake is worth. It is not a formality Renaissance can satisfy just by retaining ultimate sign-off. Renaissance’s RIEF and GF answers of 99% mean BNY Mellon is doing that work on nearly the entire book. Medallion’s 0% means nobody outside the building is doing any of it.

Read more

Schonfeld Strategic Advisors reported three put positions totalling $56,352,600 in its 13F for the quarter ended 31 March 2026, against a reported book of $14,211,293,816. The quarter before carried 60 put lines worth $1.310bn. The quarter after carried 129 worth $6.202bn. Inside the anomalous filing, the number of index and ETF put lines is zero, while the same document reports 21 index and ETF call lines worth $1.583bn.

The filing is on time, complete on its face, and reproduces its own declared totals to the dollar. I checked this morning and no amendment has been filed in the 93 days since. I scanned all 8,400 managers who filed original 13Fs for both quarters, and two of them show this signature. If you run any signal derived from 13F option composition, you ingested a 95% collapse in one manager’s put book, and nothing upstream of you was built to tell you whether it happened.

The consensus is right about 13F, and it is right about the wrong three things

Anyone who works with this data knows its defects and states them without prompting. The filing arrives 45 days after quarter end, so it describes a portfolio that has already moved. It covers long positions in a defined universe of exchange-traded equities and options, so short stock never appears and can never be netted against anything. And a manager can request confidential treatment under Rule 24b-2, delaying disclosure of specific positions while the rest of the filing goes out on schedule.

That is a fair and strong account, and I’d sign it. It’s also the version that supports the entire industry built on this data, because all three defects are known unknowns. A lag’s easy to model. A long-only universe can be treated as a lower bound. A confidential treatment request leaves a trace on the cover page. You can price all three, and serious users do.

My argument isn’t that the consensus underrates those three. There’s a fourth defect, it’s structurally different from the others, and nobody can price it, because the document gives you nothing to price it with.

The observation: one side of the book, one quarter

Here is what forces the claim. I counted every line myself, straight out of the filed XML. The aggregators restate whatever was filed, and what was filed is the thing in question.

Across Schonfeld’s 19 quarters from Q4 2021 to Q2 2026, reported put lines run from 22 to 129. Every quarter. The single exception is Q1 2026, which reports three.

The window starts at Q4 2021 for a reason you should have rather than discover. Further back, four quarters report no options at all in either direction: Q4 2018, Q2 2019, Q1 2020 and Q2 2020. Both sides absent is a different shape, and one of those quarters matters later. Schonfeld’s options reporting only becomes continuous from Q4 2021. Over the same window call lines run from 37 to 172, and Q1 2026 reports 107, which sits above the 19-quarter median of 77.

So the anomaly is one-sided. In the quarter where the put book falls 95%, the call book rises 35% by value and the call line count rises from 83 to 107. A manager reducing gross exposure cuts both sides. This filing cuts one.

The index cut is sharper, and it’s the number I’d put in front of a risk committee:

Not small. Zero. A multi-strategy manager reporting $14.2bn across 2,149 lines reported index calls on 21 tickers and not one index put.

The index and single-name split is mine, sorted by issuer name, so the 16 and the 21 would shift under someone else’s rule. The zero wouldn’t move under any rule: Q1 2026 holds three put lines and all three are single companies.

Two filers out of eight thousand four hundred

My first question was whether this is common, and I couldn’t find anyone who’d asked it. So I pulled the SEC’s own Form 13F structured data sets for both quarters and ran the diff across the whole universe.

8,400 managers filed an original 13F-HR for both Q4 2025 and Q1 2026. Most hold no options at all, so I restricted the set to filers carrying at least 20 put lines in Q4, which is my proxy for having a put book to lose. That’s 157 managers. Of those, five lost 80% or more of their put lines. Only two did so while their call book held, which is the one-sidedness that separates this from an ordinary de-risking.

Those two are Schonfeld, and Franklin Resources, which went from 28 put lines to zero. That is the whole list.

I used originals only, deliberately. A 13F-HR/A isn’t one thing: RESTATEMENT replaces the book, NEW HOLDINGS merely appends rows. ExodusPoint’s Q1 2026 amendment holds 41 rows against an original book of 1,454, so reading it as the portfolio understates that manager by 97%. Originals are also what the market saw on the day.

Names aside, the rate is the part worth keeping. A one-sided put collapse shows up in 1.27% of managers who carry a put book, which is what makes the check worth running: it fires rarely enough that a hit is worth a human’s attention.

Why the document cannot answer the question it raises

Dullness is why this mechanism survives. Three properties of Form 13F combine into a gap that no single property looks responsible for.

The cover page is derived from the table, so it can never contradict it. Schonfeld’s Q1 2026 filing declares tableEntryTotal of 2,149 and delivers 2,149 rows. It declares tableValueTotal of $14,211,293,816 and the value column sums to exactly that. I checked both identities on the three surrounding filings and all four reconcile to the dollar. That reconciliation feels like a completeness check and is not one. Both figures are computed from whatever rows the filer chose to include, so a table missing an entire instrument class still produces a perfectly self-consistent document.

Nothing in the form asserts completeness beyond a signature, and nothing links a quarter to the one before it. No field for “this report is comparable to my last one”, no prior-period column, no reconciliation of opening to closing positions. Each quarter is a standalone assertion. Quarter-over-quarter differencing, which is the basis of essentially every hedge fund positioning product sold, is an inference the form was never designed to support. Nobody built that bridge. Users assumed it.

And no one checks. This is the part I expected to be out of date and found was not. The SEC’s own Office of Inspector General reported in Review of the SEC’s Section 13(f) Reporting Requirements that “no SEC division or office conducts any regular or systematic review of the data filed on Form 13F”, that “no SEC division or office monitors the Form 13F filings for accuracy and completeness”, and that “there are no checks built into the EDGAR system, through which the Forms 13F are filed, to scan for obvious errors in the forms.” I expected to find a caveat softening this and did not. The report is explicit about what is left in place of monitoring: errors “are typically detected only in connection with IM’s processing of Section 13(f) confidential treatment requests, or when a member of the public notifies IM of an error.”

That report is dated September 2010, so I went looking for its replacement. There’s none. In September 2024 the Society for Corporate Governance, NIRI and the NYSE petitioned the Commission and wrote that they “believe that it would be helpful for OIG to conduct another review of the 13F program”. Fourteen years on, the constituency that cares most about this data was still asking for the first follow-up.

I’d rather not rest a live argument on a 2010 audit, so here’s the current version in the Commission’s own words. The SEC’s Form 13F FAQ states that an EDGAR acceptance message “does NOT refer to the correctness of the content of your filing”, that it “does not address the substantive accuracy or correctness of the filing”, and that after submitting, the manager “should review your accepted Form 13F on the SEC’s website and check your filing for accuracy and correctness.” The SEC instructs the filer to audit their own filing, because nothing else will. That is not an old finding. It’s live staff guidance today.

And the Commission has been back to this form recently. On 23 June 2022 it adopted amendments, effective for filings from 3 January 2023, which added a Summary Page checkbox flagging confidential treatment requests and required managers to report additional identifiers. So Form 13F was opened, revised and re-issued three years ago, and what got added was a better flag for the disclosure gap that already leaves a trace. No completeness attestation. No prior-period field. The one defect that leaves no trace survived a redesign.

Put those three together and you get the result that matters. A filing that omits a position and a filing that reports the absence of that position are byte-identical. No signature distinguishes them. That’s why I can tell you exactly what Schonfeld’s Q1 2026 filing says and can’t tell you what it means.

There’s a fourth property that makes the first three worse, and it’s about timing. Nothing about the Q1 filing looked wrong on the day it landed. It was filed on 15 May 2026, and read on its own it is a manager with a $14.2bn book, 2,149 positions, no puts to speak of and an active call book. That is an unusual portfolio, and unusual portfolios are legal. The discontinuity only became visible on 14 August, when the Q2 filing arrived carrying 129 put lines and $6.202bn, because the anomaly is not in either filing. It is in the pair. Anyone who consumed Q1 data during those 91 days had no signal available to them even in principle. This is also why the 45-day lag and the completeness gap compound instead of adding: the petitioners want the lag cut to five business days, which would surface the raw numbers faster and would do nothing whatever to tell you a quarter was incomparable to the one before it.

Read more

The claim: Every outlet led with Situational Awareness. Jane Street’s internal note says that stake round tripped to roughly flat on the year and the firm “largely lost on the same portfolio of trades that had strong outperformance in the second quarter.” The money went on its own book.

The numbers: $15bn hit in July against more than $40bn of trading revenue year to date. A 13F filed the same day the loss broke showing $575.58bn of puts against $472.94bn of calls. A credit spread that widened to about 330bp over a window in which the high yield market tightened.

The catalyst: The $14.6bn refinancing that priced on 12 August moves Jane Street’s debt from broadly held 144A notes into a narrow private placement, which collapses the audience for the only reporting that has ever shown this firm’s earnings.

Wrong if: Jane Street’s notes keep appearing in registered funds’ holdings filings after the refinancing settles, or the firm’s quarterly figures keep reaching the press at the same cadence.

What Reuters reported, and why it read as survivable

On 14 August, Reuters reported that Jane Street took a $15 billion hit in July, citing two people familiar with the matter and an internal note it had seen. Every version I read framed it the same way: the firm was an investor in Leopold Aschenbrenner’s Situational Awareness, that fund collapsed from $45bn to around $10bn on margin calls and a fire sale of its book to Citadel, and Jane Street wore the consequences. I wrote about the other side of that block when Citadel bought it.

I understand why it read as survivable. Trading revenue still runs above $40bn year to date, against $39.6bn for all of 2025. Fitch had upgraded the firm to BBB- from BB+ on 24 July, citing “strong through-the-cycle growth in net operating income.”

That account is coherent. It’s also widely repeated, and wrong about the thing that matters.

The note says the stake is flat on the year, and the loss came from the book

Read the internal note past the sentence everyone quoted. Here it is whole:

“We have an investment in Situational Awareness, an externally managed AI-focused hedge fund, that became large by performing well in the first half of the year. They had a large drawdown that left our stake about flat on the year, but still up over the entire period we have been invested.”

Flat on the year. Still up over the holding period. The position every headline blamed gave back its first half gains and finished roughly where it started. I’d call that a round trip. A hole is something else.

So where did I find the $15bn? The note answers directly, and this is the part nobody printed:

“The firm also took a hit from its long positions in non-AI stocks in Asia, many of which had outperformed the markets earlier during the year.”

“We largely lost on the same portfolio of trades that had strong outperformance in the second quarter. AI-exposed stocks were down a lot during July, several of the largest memory and semiconductor stocks were down around 50%.”

“Largely” is Jane Street’s own word, and I take it to point at Jane Street’s own book. The firm adds that it has “closed a significant portion of our risk in the specific areas we lost on in July, and have also reduced risk taking in other strategies.”

My read is that this reverses the story. The external allocation is the part that worked out. The firm’s own positions are what broke. A market maker’s edge is spread capture, roughly indifferent to direction and scaling with volume. July delivered plenty of both, in a month when chip stocks alone shed more than $1tn. What the note describes instead is a book that made money holding things in the second quarter and lost money holding the same things in July, which is a risk taking business wearing a market making name.

There is a second mechanism inside that first quote, and it is the one that generalises. The stake “became large by performing well.” Nobody chose to concentrate into it. An unrebalanced winner compounds into a concentration by arithmetic alone, and the moment it is largest is the moment immediately before it is marked down. That’s the oldest failure in allocation. It happened here to one of the most sophisticated risk operations in the world, and it happened to end flat.

$575bn of puts at 30 June, and what a 13F cannot see

Jane Street Group filed its Q2 13F on 14 August, the same day the loss became public, covering holdings as at 30 June. I haven’t seen anyone read the two documents against each other, and I think that’s the gap worth standing in.

The headline number is $1,205,633,745,715 across 14,718 line items. I parsed every row and the values sum to that total to the dollar. That’s my check. Nothing got dropped.

Only 13% of that trillion dollar headline is actual stock. Puts run about 1.22 times calls across the book.

One technical note, because it decides whether any of this means anything. The value column reports notional value of the underlying shares. Premium would paint a completely different picture, so I checked which one it was: for each issuer the implied dollars per share is identical across the share, call and put lines, which only happens if all three are struck against the same quarter end mark. Micron computes to $1,154.29, and Micron’s record close six days earlier was $1,213.37. The tape agrees.

Now to the memory and storage complex, which is the corner of the market the note names:

Across those four names Jane Street held $67.80bn of puts against $29.54bn of calls, a 2.30x skew on $102.87bn gross. NVIDIA sits at 0.91 in the same book on the same date.

Here is why that sits alongside the note. A 13F is a long only snapshot, filed on a 45 day lag and consolidated across eight Jane Street entities. Under the SEC’s instructions a manager reports neither short positions nor written options, and may not net either against a long holding in the same issuer. It also covers US listed securities only. The note points at Asia, which a 13F structurally cannot see. So the US listed book going into July leaned the other way in those names: real long exposure, with put notional running 2.30x the calls against it. And much of what the firm describes losing sits where this filing simply cannot look.

I get the same read from the audited numbers. In the statement of financial condition that Jane Street Capital LLC filed for 31 December 2025, that broker-dealer held $41.70bn of financial instruments owned against $45.01bn sold not yet purchased. Shorts exceed longs. That is a market making inventory at a point in time, seven months before the month in question.

Read more

The claim: The “$71 billion” attached to Citadel in every 2026 ranking is the smallest of four defensible measurements of the same firm, and the largest is 15 times bigger. The one Citadel files with its regulator, $570,621,709,022, is the number that describes what the market actually faces. Size counterparty, financing or crowding risk off the league table figure and you are out by a factor of eight, on a correction that costs one free PDF.

The numbers: $71bn investment capital as of 1 July 2026 (press, two independent chains) · $570,621,709,022 regulatory assets under management across 37 accounts (Form ADV, filed 11 June 2026) · $618,473,172,395 in long 13(f) holdings (13F-HR, Q1 2026) · $331,320,173,932 in one fixed income master fund, 58.1% of the regulatory total.

The catalyst: Citadel’s Q2 2026 13F is due 14 August 2026, unfiled as this went out. Its next annual Form ADV amendment restates every fund’s gross asset value by roughly 31 March 2027.

Wrong if: the Global Fixed Income Master’s gross asset value falls below the Multi-Strategy Equities Master’s on the next annual amendment, or the regulatory to capital ratio compresses below 5x.

The $71 billion is accurate, and it answers a question about investors

Start with the consensus at its strongest, because it isn’t wrong.

CNBC’s reporting has it that Citadel “managed some $71 billion in assets on July 1”; the FT, reporting separately, called it “more than $71 billion in assets”. Two newsrooms, one figure. It’s investor capital: the equity clients and insiders have committed, the base performance is struck on, the denominator behind “Wellington returned 5.9% in July.”

Citadel controls that number deliberately. The firm returned about $5 billion of 2025 profits at the start of this year, taking capital from $72bn to $67bn, and has handed back $32bn since 2017.

I checked whether the capital path since then is anything other than performance, and it isn’t. Wellington returned 5.7% in the first half. Multiply $67bn by 1.057 and you get $70.8bn, against a reported “about $71 billion” on 1 July. Work it backwards from 12% through July and 5.9% in July alone and the implied first half is 5.76%. Both routes land in the same place. The base is closed. Nobody subscribed, and the growth is the portfolio’s.

So my quarrel isn’t with the figure. It’s with what people think it measures.

Item 5.F, filed 11 June 2026: $570,621,709,022 across 37 accounts

Citadel Advisors LLC files its Form ADV under CRD 148826, from 830 Brickell Plaza, Floor 15, Miami. The current filing is dated 11 June 2026. Item 5.F reads:

Discretionary regulatory assets under management, $570,621,709,022. Non-discretionary, zero. Total accounts, 37. Of that, $384,834,614,984 sits with clients who are not United States persons, or 67.4%. Item 5.D assigns every dollar to one category, pooled investment vehicles. Item 5.A reports 3,284 employees, 1,314 in investment advisory functions.

That is 8.04 times the $71 billion.

A third measurement sits in a different filing. Citadel Advisors’ 13F-HR for the first quarter of 2026, filed 15 May on a 45 day lag, reports positions as they stood on 31 March: a table value total of $618,473,172,395 across 15,589 lines. A fourth sits back inside the ADV: Schedule D requires a separate entry for every private fund, each carrying its own current gross asset value. Summed across all 37, those entries total $1,094,345,899,742.

One firm, one reporting window give or take eight weeks. $71bn, $570.6bn, $618.5bn, $1,094.3bn. The spread from smallest to largest is 15.4x, and I’d call every one a real number somebody filed on purpose. Citadel disclosed all four accurately and on time, so my argument is with the people quoting them and not with the filer. One became the shorthand, and now does work it was never built for.

The SEC instruction that creates the gap: “Do not deduct any outstanding indebtedness”

I do not read that gap as a discrepancy. It is written into the form.

The SEC’s Form ADV instructions, at Instruction 5.b, tell an adviser how to compute Item 5.F. Two sentences do the work. First: “treat all of the assets of a private fund as a securities portfolio, regardless of the nature of such assets.” Second, and this is the operative one: “Do not deduct any outstanding indebtedness or other accrued but unpaid liabilities.“

Regulatory AUM is a gross assets measure by construction. Buy a Treasury bond with repo financing and the bond enters Item 5.F at full market value while the repo liability enters nowhere. Investment capital is a net equity measure. These are not two estimates of one quantity. They are the numerator and denominator of a financing ratio, and 8.04x is that ratio.

That reframing is the whole argument. Nobody is quoting the wrong number. Two numbers that divide into each other are being treated as rivals for the same job, and the division is where the information sits.

I should be honest about how novel this is. Securities lawyers have written the gross point up for years, in these words: a fund adviser “is required to include all gross assets without any deduction for debt or leverage”. What I can’t find anyone doing is the next step: taking it out of the compliance filing to size a named firm’s footprint, then running it across the cohort. The rule is documented. The application isn’t.

The 13F number is gross in a third and stranger way. It captures long positions in 13(f) securities only: no shorts, no bonds, no futures, no non-US listings. It also aggregates other included managers, and Citadel’s names one: Citadel Securities GP LLC. The market maker’s inventory rides inside the hedge fund adviser’s 13F, which squares with the ADV listing fifteen affiliated broker dealer and clearing entities as related persons. Read it as “Citadel’s portfolio” and you’re reading a market maker and a hedge fund added together, shorts deleted.

The feeders hold $74.6bn. The masters they feed hold $530.1bn.

Schedule D showed me exactly where the $71 billion lives.

Three of the funds are the investor facing vehicles. Citadel Wellington LLC, the onshore flagship, reports gross assets of $30,935,735,414 across 309 beneficial owners, 0% non-US. Citadel Kensington Global Strategies Fund Ltd., the offshore twin, reports $30,105,876,076 across 372 owners. Kensington II adds $13,533,767,845. Together, $74,575,379,335, or 1.05x the stated $71bn of investment capital.

I didn’t have to guess at which three. KBRA’s affirmation of 21 April 2026 names exactly that set. The issuer pays for the rating and supplies much of the data, so I lean on it for facts Citadel has no reason to misstate, never for judgement. On that basis it corroborates the three vehicles I picked out of Schedule D by owner count. The investor sits in $74.6bn, inferred and then confirmed.

Now the funds those feeders feed:

$530.1bn of master fund assets standing on $74.6bn of feeder vehicles, a ratio of 7.11x. Double counting inflates the $1.09tn total, and it’s visible here too: GFIL Holdings Ltd. reports $290,469,065,838 with three owners, and it is a holding layer above the same fixed income assets. Item 5.F’s $570.6bn already nets that layering out, which is why I lean on it rather than on the sum.

One line in the Wellington entry deserves care, because it’s where I nearly got this wrong. Question 14 asks what share of the fund is beneficially owned by the adviser and its related persons, and for Wellington LLC the answer is 53%. Read fast, that says the onshore flagship is majority inside money. KBRA, with management access, puts principals and employees at 18% of the capital base. Both hold, because Form ADV counts as a related person any entity under common control, which sweeps in affiliated holding vehicles nobody owns personally. 18% is the number I’d use, and it’s still a large insider share.

58.1% of the regulatory total sits in one fixed income fund

Here’s the finding I didn’t expect when I opened the filing.

Citadel gets discussed as an equity firm: the pods, the stock pickers, the July headline about Situational Awareness. Its own filing describes something else. The Global Fixed Income Master Fund carries $331,320,173,932, which is 58.1% of the entire $570.6bn regulatory total and 4.67x the firm’s whole investment capital. Its Multi-Strategy Equities Master carries $92,049,214,132. The fixed income master is 3.6 times larger than the equities master. Group every fixed income vehicle in the schedule and you reach $692,146,644,790, or 63.2% of the gross sum, layering included on both sides of that ratio.

Here’s the objection I’d raise if somebody showed me this, and it nearly killed the finding. Gross asset value isn’t comparable across strategies. A cash and carry Treasury position books both legs; a long equity position books once. If the fixed income book runs at 15x gross to capital and the equity book at 3x, then $331bn and $92bn imply something like $22bn and $31bn of capital, and the equity business is the larger one. I can’t rule that out. The capital split lives in Form PF and Form PF is not public.

So take it narrowly, which is how I take it. This is where the BALANCE SHEET sits; capital and risk are separate questions. The balance sheet still determines financing, counterparty exposure and footprint, and it’s still the thing nobody quotes.

Performance points the same way. In March 2026 Citadel’s Global Fixed Income fund fell 8.2% in a single month, a drawdown the trade press attributed to rate volatility and widening basis spreads. I treat that attribution as a reporter’s reconstruction and I would not underwrite it. The magnitude is what I would hold onto. A strategy nobody puts in the headline moved 8.2% in a month, and it is the largest thing the firm runs.

Gross measures footprint. The Fed’s $4.0 trillion nets to $0.8 trillion.

An obvious objection: I’ve discovered leverage and dressed it up as insight. Eight times gross sounds alarming, so let me argue the other side properly, because a careful person reaches it first.

A relative value fixed income book is gross heavy by construction and carries little directional risk. Buy the cash Treasury, short the future, finance the cash leg in repo. Both legs count at full value. Net, you own a few basis points of price difference. Enormous in gross, nothing in net.

Precisely this has now been quantified by the Federal Reserve. In a FEDS Note published 22 June 2026, Phillip J. Monin decomposes large hedge funds’ Treasury exposures from Form PF: $4.0 trillion gross as of September 2025, made up of $2.4 trillion long and $1.6 trillion short. Net those and you have $0.8 trillion, 20% of the gross figure. The cash futures basis trade alone runs about $830 billion, roughly double its early 2020 peak and 35% of long Treasury exposure, financed inside $3.0 trillion of repo borrowing that the OFR attributes to multi-strategy, macro and relative value funds.

Citadel’s own balance sheet supports the steelman from the other side. KBRA reports the flagship funds hold “generally more than 30% of investment capital held in cash and cash equivalents (though the fund averaged more than 40% during 2025)”, with Level 3 positions at about 2% of the portfolio. A firm carrying 8x gross while sitting on 30% to 40% cash is not running out of rope. Those two facts belong in one sentence and almost never are.

So the steelman holds, and I’d rather concede it here than have a reader find it for me. Eight times gross does not mean eight times the risk, and anyone who tells you it does is wrong.

What I read it as is footprint. Gross assets tell you how much dealer balance sheet a firm consumes, how much repo it rolls every morning, and how much of a market it occupies when everyone reaches for the same exit. Those questions are answered by $570.6bn, and by the $331.3bn inside it. None is answered by $71bn, and $71bn is the only number most people see. Footprint and risk are different questions. The press quotes one number for both.

The capacity bound: returns fell from 38.1% to 10.2% while the industry’s gross doubled

If gross assets are the footprint, the next question is whether the footprint still buys what it used to. It doesn’t.

Wellington’s full published series runs 9.1% in 2018, 19.4%, 24%, 26%, then 38.1% in 2022, 15.3%, 15.21%, and 10.2% in 2025. I print all eight because the short version flatters my argument. 2022 was the best year in the fund’s history, and its $16bn was the largest single year profit any hedge fund had ever booked, so starting the clock there makes any subsequent path look like collapse.

The honest reading is narrower and it still holds. Returns have fallen since that peak, with 2023 and 2024 flat at just over 15%, and 2025’s 10.2% sits at the bottom of the eight year range with only 2018 below it, against 19.46% annualised since 1990. The Global Fixed Income fund, the firm’s largest book, returned 9.4% in 2025. Over that same stretch the industry was adding gross rather than shedding it: OFR data has hedge fund repo borrowing more than doubling, up 104%, between Q4 2022 and Q4 2024, and the Fed’s decomposition has the cash futures basis trade at roughly twice its early 2020 peak.

More gross, less return. I read that as capacity erosion, and it’s why I don’t read the capital returns as generosity. The binding constraint at Citadel isn’t investor demand, it’s alpha per unit of gross balance sheet, and a firm that can’t deploy another dollar profitably hands the dollar back. Griffin has said as much in the language of opportunity sets. The filings say it in the language of arithmetic.

The capital path says it arithmetically. Citadel ended 2022 with $62.3bn and stood at about $71bn on 1 July 2026, a 14% gain in three and a half years, while the portfolio earned multiples of that. The difference walked out as distributions. The money is generated. It is not retained.
A capital return works the same way from the other side. A profit distribution takes equity off the right hand side of the balance sheet and the left hand side never hears about it, so handing back $5bn leaves gross assets untouched and raises the gross to equity ratio mechanically. Same footprint, smaller cushion. A capital return is a financing decision as much as a capacity decision, and only one of those readings is ever reported.

The peer table turns this from a Citadel quirk into a strategy signature

I pulled the same Item 5.F from every large platform’s own Form ADV, because one firm’s filing proves nothing alone. A pattern shows up that no single document reveals.

Citadel sits second, and I’d call Millennium’s $720,845,951,000 on a comparable investor base the more levered of the two. Bridgewater is the row I’d study: comparable investor capital, a quarter of the regulatory assets, because macro views expressed through futures and swaps consume far less gross balance sheet than cash against derivative relative value. The ratio is a strategy signature. Size is a different table.

One more entity deserves naming. Qube Research & Technologies Limited files as an exempt reporting adviser and reports no Item 5.F, but its Schedule D lists nine funds totalling $580,434,724,327. Add Millennium, Qube and Citadel and the three largest carry $1.87 trillion between them. That is the number these firms run on, and it is not the number they get written about with.

Three things I can’t resolve from public documents

The valuation date of Item 5.F isn’t observable. The 11 June 2026 filing is an other than annual amendment, and Item 5.F need only be updated annually, so the computation may date from the last annual one. I tried the SEC’s historical ADV compilation feeds and they refuse automated access. A separate fetch on 11 August returned an identical $570,621,709,022, which tells me the figure is stable and nothing about when it was struck.

Gross asset value isn’t net asset value, and Form PF isn’t public. Schedule D asks for gross assets. The borrowings, the repo, the derivative notionals and the actual net exposures all live in Form PF, which the SEC collects and doesn’t publish at firm level. Everything I’ve said about Citadel’s financing is inferred from the gap between two published numbers. None of it is read off a liability schedule. And that private window is narrowing: on 20 April 2026 the SEC and CFTC jointly proposed lifting the Form PF filing threshold from $150m to $1bn, and the large hedge fund adviser threshold from $1.5bn to $10bn. Fewer filers, less collected. Which makes the free filing I have been reading from more load bearing, not less.

The 13F comparison is the weakest of the four, and I’d never use it as a portfolio figure. The Q4 2025 filing reported $665,872,168,045 across 15,403 positions, so it swings by tens of billions a quarter for reasons with little to do with the hedge fund.

What would change my view: the annual amendment due March 2027

My composition claim is the most exposed, so here is its test, stated precisely.

Citadel’s next annual Form ADV amendment is due within 90 days of its fiscal year end and restates Item 5.F alongside all 37 fund gross asset values. If the Global Fixed Income Master’s gross assets come in below the Multi-Strategy Equities Master’s on that filing, my reading that this is a rates balance sheet with an equities business attached is wrong, and I’ll say so here. If total regulatory AUM to investment capital compresses below 5x while capital is flat, the financing story is wrong. And if a future FEDS Note shows the industry’s net to gross Treasury ratio climbing well above 20%, the footprint claim weakens, because gross and net would be converging.

Nearer term, the Q2 2026 13F is due 14 August. I expect the table value to jump on the Situational Awareness purchase, and I expect that jump to tell you almost nothing about the firm. Which is rather the point.

$71bn for returns, $570.6bn for counterparty risk

Here’s the rule I’d apply, one line each.

Underwriting Citadel’s returns or its fee load? Use $71 billion. That’s the equity performance is struck on, and the honest denominator.

Underwriting Citadel as a counterparty, a repo borrower, a crowding risk, or a competitor in trades you’re in? Use $570.6 billion, and look inside it at the $331.3 billion. That’s the balance sheet you’re standing next to.

Reading a 13F headline about Citadel’s “portfolio”? Discount it hard. Long only, US listed, market maker inside.

The habit generalises better than the case, and it’s the part I’d keep. When you see a firm’s size quoted, ask which filing it came from and which question that filing was built to answer. Across every platform I pulled, the answer moves by a factor of five to fifteen. The filings are free. Almost nobody opens them.

The whole procedure is in the accompanying note: six steps, the eight adviser dataset, the three places it breaks. On the financing side I took apart how hedge funds trade the $30 trillion Treasury market, and before that how Millennium, Citadel and Point72 structure their pods. I publish this filing work as research on Patreon, for people underwriting a counterparty rather than reading about one. Shorter reads go up on LinkedIn, filmed versions on YouTube.

So here’s my question for anyone allocating to this space. When your risk team sizes counterparty and crowding exposure to one of these platforms, do they use the investor capital figure or the Item 5.F figure, and if it’s the former, what’s the argument for it?

Cover photograph: J. Crocker, Attribution, via Wikimedia Commons.

When Citadel took Situational Awareness's entire public book in a single block before the open on 30 July (WSJ), four of the AI-infrastructure names at the centre of it rallied 21–28% in a session. The consensus read that as confidence: a credible buyer had stepped in, so the selloff must have been overdone. I don't think that reading survives the cross-section. A stock the fund never owned rallied just as hard as the ones it did. That's the whole story.

The discount a forced seller pays is not a verdict on the asset. It's the fee for compressing time. What cleared on 30 July was not a portfolio. It was the market's belief that a levered book still had to come out.

I have named the observation that would prove me wrong, and it resolves within weeks.

The consensus, stated at its strongest

A large, solvent buyer publicly underwriting a battered asset class is real information. Citadel has the balance sheet, the risk systems and the reputation to make that signal credible, and it committed capital rather than commentary. If you believed the AI-infrastructure selloff was a solvency scare rather than a demand story, Citadel's bid was evidence for you.

I'm not going to argue that this is stupid. My claim is narrower: it's incomplete, and I think the incompleteness is measurable.

What the tape actually did

By mid-afternoon on 30 July, the cohort was up hard: Nebius +28.5%, Cipher Mining +28.3%, IREN +27.4%, Core Scientific +21.0%.

Cipher Mining does not appear in any reported holdings list for the fund, and SpotGamma's position mapping states it was never in the book.

Hut 8, which the fund had already exited, rallied with the group. A story about renewed confidence in AI predicts a broad rally roughly proportional to AI beta. It doesn't predict that a name with no connection to the seller trades in line with the very best of the names actually being sold.

That is what an overhang predicts, and it is why I read the session the way I do.

The objection worth taking seriously

These names move together every day. They're the same trade. So the fair challenge is: you have shown correlation that exists on any big up-day in AI infrastructure, and called it a mechanism.

Two things separate this from an ordinary co-movement day, to my eye.

The timing tracks a supply announcement. No fundamental changed. The cohort based through the evening of 29 July as the rumour circulated, then gapped at the 30 July open when the block trade hit the news. Nothing about the demand for AI compute changed between those two prints.

And the cross-section runs the wrong way for the alternative. If the rally were about those specific shares finding a home, the names actually being absorbed should have moved more than the ones that weren't. They didn't: a name with no reported connection to the seller moved with the best of them. Under the overhang read that's the expected result, because the discount was never on the shares.

To me that's a discriminator, not decoration, and it is the same logic the falsifier at the end proposes to test out of sample.

Liquidity is a service someone sells you

Grossman and Miller set this out in 1988 ("Liquidity and Market Structure", *Journal of Finance* 43(3), 617–633; the working paper is free to read at NBER w2641). Market liquidity is the demand and supply of immediacy. A liquidity event creates demand. So does the risk of waiting. Market makers supply it by being present and bearing risk during the interval between the arrival of final buyers and sellers.

I would read that last clause twice, because it is the entire mechanism. The market maker isn't paid for being right about the asset. He is paid for standing between a seller who must transact now and a buyer who has not arrived yet.

Worth sitting with what that interval actually is. A seller who must be flat by the close and a buyer who would happily own the asset next Tuesday are not in the market at the same moment. Somebody has to carry the position across the gap between them, wearing the price risk for as long as the gap lasts. That carry is the service, and the discount is its fee. Widen the gap and the fee rises, whatever the asset is.

Which means the fee is a function of two things that have nothing to do with the company: how fast the seller must be done, and how many balance sheets are willing to stand in the interval. A forced seller has set the first to zero. The second is where this particular week gets interesting.

So the discount on a forced sale is a price for time. What the securities are worth is a separate question. It tells you what the interval was worth. It tells you nothing about fair value.

Why the bid disappeared everywhere at once

Here is the part the confidence story cannot reach.

Shleifer and Vishny showed in 1992 ("Liquidation Values and Debt Capacity: A Market Equilibrium Approach", *Journal of Finance* 47(4), 1343–1366; free copy) that in industry distress, the seller and the next-best user of the asset are encumbered at the same moment, for the same reason. The people who would ordinarily bid are impaired by the identical shock. Assets then clear to buyers who value them less. The discount widens past anything the fundamentals justify.

Now apply it, which is where I think this gets interesting. Who are the natural best-use buyers of levered AI-infrastructure equity? Other funds running levered AI-infrastructure books. In July, they were nursing the same drawdown as the seller.

So the bid did not thin in the names being sold. It thinned across everything those buyers would otherwise have bought. That's the step I think the confidence reading misses. The cohort was marked down for a supply event that hadn't happened yet. Names the seller never touched were marked down alongside it, because the discount was never attached to the shares. It was attached to the absence of anyone able to buy them.

The paper's own conclusion is worth stating in full, because it is sharper than the way it usually gets quoted. When the natural owners cannot bid, the asset does not simply trade lower. It trades to somebody who values it less, which is a different and more permanent kind of loss. Price is the visible part; misallocation is the part that persists after the price recovers.

Carry that into a crowded equity book and the shape changes. The buyer who would pay the most for levered AI-infrastructure exposure is someone who already understands it, already models it, and already has the mandate to hold it. That describes the cohort. It also describes exactly who was nursing a drawdown in July. So the bid that disappeared was not the marginal bid. It was the best bid, and the best bid is the one that sets the price you get.

That is why a never held stock rallied like a held one. Both carried the same discount. The same news removed it.

I ran the full version of this for paying members. It carries the evidence table with the in book column, the sizing view, and why the standard overhang hedge is the wrong instrument for: The Overhang Trade: What the Citadel Block Actually Repriced.

Read more

Situational Awareness returned 439% in the first half of 2026 and still sold its public book at the bottom, in a week priced for a rate hike that never came. The decision block, the numbers, and what would kill the view:

On Wednesday, July 29, the Federal Open Market Committee voted 9-3 to hold the federal funds rate at 3.50-3.75%. Before Thursday’s open, Situational Awareness LP, the AI-concentrated hedge fund run by former OpenAI researcher Leopold Aschenbrenner, sold its public equities book in a single block, longs and shorts together, to Ken Griffin’s Citadel. How much of it went is itself disputed: the whole book, per SpotGamma’s account of CNBC’s reporting and The Next Web’s of the FT’s; the majority of it, per TechCrunch. On Thursday, the names in that book ripped: SanDisk rose 23.9%, Micron 17.0%, Bloom Energy 26.4%, with the Nasdaq 100 up 3.1%.

The consensus reading of this episode is already settled: a brilliant, over-levered kid met a margin call, and leverage kills. That reading is true and worthless. Every professional already knows leverage kills. What the consensus misses is what the public record shows on two specific points. The fund’s hedges were structurally incapable of protecting it, and the numbers proving that sat in its own filings since May. And the panic that filled its final week was measurably part hedging flow rather than forecast, a distinction visible in real time on two free screens. Neither required inside information. That is the part worth a professional’s time.

Six days in July

The sequence matters. Backdrop first, then the tape.

2026’s inflation was an energy story. The CPI energy index rose 10.9% on the month in March, and headline inflation hit 4.2% in May before June’s print, released July 14, came in cool: down 0.4% on the month, 3.5% annually, with core at 2.6% and gasoline falling 9.7% on the month. The Fed’s own July statement still described inflation as “elevated relative to the Committee’s 2 percent goal, in part reflecting supply shocks that have driven price increases in certain sectors, including energy.”

Meanwhile the AI infrastructure complex was breaking for its own reasons. Over the month through late July, SanDisk fell 56% and Nebius dropped about 48% from its peak; SpotGamma’s post-mortem, counting a slightly different window, puts the fund’s core holdings, Nebius, SanDisk, Micron, and CoreWeave, each down between 27% and 54% on the month, with the Nasdaq 100 off more than 10%. The drawdown was doing its damage well before anyone repriced the Fed.

• July 24: Aschenbrenner writes to clients that the fund, now at approximately $20 billion, has “not been immune” to the selloff, calls it one of the best buying windows since early 2025, and invites fresh capital starting August 1, with a possible Anthropic IPO highlighted as a catalyst. Whatever commitments came, they evidently did not come in time.

• July 27-28: Citadel Securities publishes a client note by Frank Flight, its head of macro strategy, arguing for a surprise 25 basis point hike at the July meeting rather than September: “The market may once again be underestimating the extent of the hawkish shift at the Fed.” A hike now, the note argues, would “emphatically end the forward guidance era” and reduce the total tightening needed later.

• July 28 (Tuesday): CME FedWatch puts hike odds at 37.9%, up from 25.7% a week earlier. Kalshi prices the same event at 28% on more than $45 million of volume; Polymarket at 27.5% on over $107 million. A Reuters poll of 104 forecasters taken July 17-21 finds none expecting a move.

• The final days (reported only as “late July”): Goldman Sachs and JPMorgan issue margin calls to funds holding concentrated AI positions; Bank of America, Goldman, and JPMorgan are marketing Situational Awareness positions.

• July 29 (Wednesday): The Fed holds, 9-3, with Hammack, Kashkari, and Logan preferring a quarter-point hike. The hike risk was real; a quarter of the committee voted for it. Equities stay heavy into the close. After the bell, Microsoft, which closed Wednesday at $390.54, reports results showing 43% cloud revenue growth.

• July 30, pre-open: the book goes to one buyer in a single block, after Microsoft’s print is public. The buyer is Citadel.

• July 30 (Thursday): Microsoft rises as much as 17%, its biggest one-day gain since 2008, adding roughly $480 billion of market value and dragging beaten-down semiconductors sharply higher, reversing Wednesday’s losses. SanDisk closes up 23.9%. Bloom Energy, up 26.4% on earnings. Nebius rises nearly 30% in morning trade. The Nasdaq 100 gains 3.1%.

A fund that returned 439% for the year through June and reached as much as $45 billion in assets at the start of July ended the week at roughly $10 billion, half of it a private Anthropic stake valued at $5 billion. It sold hours before the rebound, and after the rebound’s catalyst was already on the tape. That is not bad luck; that is what margin mechanics do, and it is worth being precise about which mechanics.

The panic had a basis, and the basis was the tell

Start with the instrument that carried the panic. CME’s FedWatch tool converts 30-day fed funds futures prices into implied probabilities of FOMC moves. The market treats those probabilities as forecasts. They are not. A futures price is a forecast plus a risk premium, and the premium is not small or stable: Piazzesi and Swanson’s canonical result is that excess returns on fed funds futures are positive on average, countercyclical, and predictable, and that “simply ignoring these risk premia has important consequences for the expected future path of monetary policy.”

One line of intuition for what that means in practice: when a levered investor fears a rate shock, he does not update a forecast, he buys protection, and protection demand moves the futures price exactly the way a genuine change in expectations would. FedWatch cannot tell the difference. Prediction markets mostly can, because their buyers are expressing the event probability itself, in capped-size, real-money contracts, not hedging a bond book. The forecasting literature backs the asymmetry: Snowberg, Wolfers, and Zitzewitz find prediction markets generally exhibit lower statistical errors than professional forecasters and polls and incorporate new information quickly. I have covered which funds actually trade these venues and the pricing math underneath the contracts before; July was the first time the two screens mattered this much in one week.

That is why July 28 was so informative. The same binary event was priced three ways at once, and surveyed a fourth (all four figures from the same Bloomberg-syndicated snapshot; the primary tapes are CME’s and the venues’ own):

Be precise about what moved. Real information was repricing everyone that week: both venues repriced sharply, with Kalshi’s hike contract gaining 8 points into the meeting, which puts its week-earlier price near 20. The hike risk was not imaginary, and Wednesday proved it, with three of twelve voters dissenting in favor of a hike. But the futures-implied number rose 12.2 points into the same meeting, 25.7 to 37.9. So the gap between the two measures widened as the date approached: from roughly 6 points to roughly 10. Information moves both prices together; hedging pressure moves only the instrument people hedge with. The widening is the flow signature. Four points of widening on one meeting is a lean signal, and it is all the demonstration this episode offers; it was also visible in real time, timestamped, and free.

The Fed then held, 9-3. One hold validates nobody’s probability. The prediction-market number was not a clean forecast either: Kalshi’s own contracts show a documented favorite-longshot bias, which means a 28-cent hike contract likely overstated the true probability, making the futures premium larger, not smaller. This episode is one data point for the premium reading, no more. The systematic test, futures-implied minus event-market odds across every FOMC date since liquid Fed contracts existed, has apparently not been published. Someone will run it.

Now connect this to the dead fund, carefully, because the honest connection is narrower than the dramatic one. The month’s damage was done by the AI book itself: the fund had already fallen from $45 billion toward $20 billion before Citadel Securities wrote a word and before the futures spike. No fed funds flow moved SanDisk’s price. What the wedge tells you is what kind of week the fund’s margin clock ran out in: a week when priced fear exceeded measured probability by a third, when margin calls were going out across the levered AI complex, and when prime brokers were marking collateral against panicked prices rather than against the outcome three days away. Margin calls settle against price, not against the eventual outcome. This is the spiral Brunnermeier and Pedersen formalized, margins turning destabilizing as market liquidity and funding liquidity reinforce each other, and the single-block sale that ended the episode is what heading off that spiral looks like in practice. A levered fund dying in a flow-dominated week sells at flow-dominated prices. The Fed did not liquidate Situational Awareness. The Fed did nothing, which is the point.

Read more

🎬 Prefer to watch rather than read? A NotebookLM-generated video overview of this article is available here: Watch the video overview → Full analysis, citations, and data remain in the article below.

Why These Numbers Are Different From Every Prior Record

The Q1 2026 figures are not a gradual step up. They represent a structural break. Bloomberg first reported on May 11, 2026 that Hudson River Trading recorded $6.4 billion in trading revenue in Q1 — more than half the firm’s total haul across all of 2025 — and that HRT’s profit climbed approximately 175% to $4.2 billion, with EBITDA of $4.5 billion and $20 billion in net trading capital at quarter-end. The Financial Times confirmed the figures and reported that HRT’s net trading revenue in Q1 exceeded that of Bank of America or Wells Fargo in the same period. On the same day, Bloomberg reported Jane Street’s Q1 results: $16.1 billion in trading revenue — more than double its haul from the same period in 2025 — and net income more than doubled year-over-year to $10.3 billion. Reuters independently confirmed the same figures, though it described revenues as “up more than 40% from the same period last year” — a phrase that Bloomberg’s own data contradicts, as Q1 2025 net revenue implied by the H1 2025 figure of $17.3 billion (Bloomberg, September 2025) minus Q2 2025’s $10.1 billion leaves Q1 2025 at approximately $7.2 billion — making the year-on-year growth closer to 124%, consistent with “more than doubled.”

These are not directional bets that happened to work. The business model involves providing liquidity across hundreds of venues simultaneously, capturing spread while managing inventory risk using predictive models. Understanding precisely how that generates $22.5 billion in ninety days requires tracing the complete evidence chain: the macroeconomic transmission mechanism, the mathematical architecture of modern market making, the operational specifics documented in regulatory filings and interview transcripts, and the primary records from two court proceedings and one regulatory order that expose these strategies at execution level.

The Macro Catalyst: Mapping Cross-Asset Transmission

The Iran war did not create a single spike that quickly resolved. It created a sustained, multi-asset repricing cascade that persisted across weeks. The US war in Iran fuelled large swings in oil prices as most ships were blocked from passing through the Strait of Hormuz, with the fallout spreading to Treasuries and currencies as investors grappled with longer-term consequences to the global economy. The US Energy Information Administration’s Q1 2026 quarterly price review confirms the precise trajectory: Brent began the year at $61 per barrel, had risen to $72 per barrel by late February in response to escalating conflict risk, then surged sharply after military action on February 28 and the subsequent de facto closure of the Strait of Hormuz — finishing the quarter at $118 per barrel, the largest quarterly price increase on an inflation-adjusted basis in data going back to 1988. CNBC’s oil market timeline confirms the pre-war reference price: Brent was approximately $72 per barrel on February 27 — the last trading day before hostilities began, and peaked near $120 during March. CNN confirmed the pre-war baseline from a different angle, noting Brent was at approximately $73 per barrel before the war, and subsequently reached a wartime high of $126.41 during the late-April escalation.

The EIA report also documents the specific structural dislocation mechanism that created value for market makers: the Brent-WTI spread widened from approximately $4 per barrel at the start of Q1, peaked at $25 per barrel on March 31, and averaged $11 per barrel for the month — the highest Brent-WTI spread in over five years — reflecting elevated shipping costs and reduced oil flows near the Strait of Hormuz pressing on Brent specifically while US inventories and SPR releases cushioned WTI. This Brent-WTI dislocation, combined with steep backwardation in crude forward curves, created persistent cross-instrument mispricings that a unified, multi-asset market-making book could arbitrage continuously across the quarter.

For market makers, duration matters more than peak volatility. A spike that resolves in hours generates one wide spread on a large position. A regime that persists for weeks generates wide spreads on millions of transactions, repeated daily. Investors moved aggressively to hedge risk as concerns grew around AI disrupting software companies and uncertainty tied to the Iran conflict, with market anxiety intensifying in March after the outbreak of the US-Israeli war with Iran. This wave structure of demand meant institutional hedgers were repeatedly entering the market over weeks, not once.

The cross-asset correlation structure was also mechanically valuable. Analysts observed a direct inverse correlation between crude oil and S&P 500 futures intraday — each crude move triggering systematic equity selling. Firms whose predictive models had mapped this transmission function could quote on the receiving end of the transmission — equity index options, Treasury futures — with better-calibrated uncertainty than firms reacting independently to each asset class. A crude print that a well-trained model predicts will move equity futures a predictable amount removes that component from the adverse selection term. That is direct incremental P&L.

The Mathematical Foundation: Reservation Price, Optimal Spread, and the Alpha Layer

Two academic frameworks explain what these firms are actually optimizing. Both are publicly available in peer-reviewed form.

The foundational architecture is the Avellaneda and Stoikov (2008) model, published in Quantitative Finance, which solves for the optimal bid and ask quotes of a dealer managing inventory risk through a stochastic control problem. The model derives two outputs: a reservation price that adjusts the midprice based on inventory imbalance, and an optimal half-spread. The reservation price is defined as:

r* = s − q · γ · σ² · (T − t)

Where s is the current midprice, q is current inventory, γ is risk aversion, σ² is variance, and (T − t) is remaining trading horizon. The optimal half-spread is:

δ* = (γ · σ² · (T − t)) / 2 + (1/γ) · ln(1 + γ/k)

Where k captures the sensitivity of order arrival rates to spread width. Both components scale directly with σ². When crude and equity vol were simultaneously elevated in Q1 2026, the model-prescribed optimal spread widened in both markets. Dealers quoting at wider levels captured more revenue per transaction while still attracting flow from institutional hedgers who needed to transact regardless of spread width. Guéant, Lehalle, and Fernandez-Tapia (arXiv:1105.3115) extended this framework with closed-form solutions under inventory constraints, now the standard implementation reference for institutional market makers.

The Avellaneda-Stoikov baseline, however, assumes no directional view. This is precisely where HRT and Jane Street operate at a different level. Cartea and Wang’s “Market Making with Alpha Signals” (SSRN 2019, published in the International Journal of Theoretical and Applied Finance 2020) formally proves how a market maker possessing a momentum signal about short-term price direction can simultaneously minimize adverse selection costs, execute directional trades in anticipation of price changes, and manage inventory risk — with expected profits from the alpha signal increasing monotonically with risk tolerance, because the strategy employs more speculative market orders and performs more round-trip trades as tolerance rises. In operational terms: HRT and Jane Street are running the alpha-signal-augmented version of market making, not the passive baseline. The directional edge and the spread capture compound each other.

Hudson River Trading: Four Primary Sources on Architecture and Edge

Source 1: The Bloomberg Odd Lots Interview Transcript

On October 31, 2025, Bloomberg’s Odd Lots podcast published a 55-minute interview with Iain Dunning, HRT’s Head of AI Research, covering the firm’s use of AI to make short-term predictions about price that give its traders an edge. Dunning joined HRT from DeepMind, where his work included a paper on population-based reinforcement learning in multi-agent environments published in Science in 2019. His personal website confirms his current role: running HRT’s AI team, building “some of the most advanced models of financial markets in the world, using state-of-the-art techniques combined with massive compute and data”.

A PodMine transcript of the episode documents three specific disclosures: HRT’s AI models achieve approximately 50.1% accuracy in predicting short-term price movements — slightly better than random but sufficient for profitability at scale across millions of daily transactions; the firm consumes “tens of megawatts” of electricity for AI operations, described as more than most towns and cities; and since approximately 2014, HRT moved toward neural networks that consume all available market data rather than handcrafted features. The 50.1% accuracy figure is the most analytically important: at millions of trades daily, a 0.1 percentage point edge over random is a statistically enormous advantage that compounds directly with the Cartea-Wang alpha signal mechanism.

Source 2: SEC Rule 605 Data

In August 2025, Global Trading reported that among major wholesale market makers, HRT posted the lowest (best) share-weighted median execution quality (E/Q) ratio at 0.315, with Susquehanna (SIG) next at 0.335 — both clear of their competitive field — while Citadel Securities showed the most pronounced deterioration in the month, with its median E/Q worsening from 0.405 to 0.515. The E/Q measure, derived from SEC Rule 605 mandatory monthly disclosures, is the spread realized by market makers versus the NBBO midpoint divided by the prevailing NBBO spread — so a reading of 0 is a trade at midprice and 1 is a trade at the NBBO. HRT’s 0.315 versus SIG’s 0.335 represents a material advantage; both outperformed Citadel at 0.515 by a substantial margin. HRT’s own Rule 605 filings, disclosed monthly on its website, are the primary data source underlying these comparative analyses.

The execution quality advantage is the observable signature of superior short-term price prediction. A firm that predicts where prices will move in the next few seconds can offer better prices to retail orders with confidence the market will not immediately move against the filled position. The monthly Rule 605 E/Q differential is the empirically mandated, publicly disclosed record of that prediction advantage.

Source 3: The Bancara Credit Analysis

Bancara’s February 2026 analysis, drawing on S&P and Fitch credit rating research, reported that HRT held net capital of $2.5 billion at end-2024 and achieved net trading revenue of $3.7 billion in Q3 2025, representing 81% year-on-year growth and exceeding all prior quarterly results at that point — a figure itself now far surpassed by Q1 2026’s $6.4 billion. This establishes the trajectory clearly: Q3 2025’s then-record of $3.7 billion was already the highest in HRT’s documented history, and Q1 2026 exceeded it by 73%.

Source 4: CoreWeave’s Q1 2026 SEC Earnings Filing

CoreWeave’s Q1 2026 earnings press release, filed with the SEC in May 2026, lists Hudson River Trading explicitly as a “partner of choice for leading AI pioneers and enterprises”. This is primary documentary evidence of HRT’s active AI infrastructure investment — corroborating Dunning’s podcast disclosures about electricity consumption and the shift to GPU-intensive neural network training at scale.

Where HRT concentrates its architecture on a single integrated prediction system across 200-plus venues, Jane Street’s Q1 2026 figure rests on a different mix — and the three components below are not equivalent. Only the first, medium-frequency trading, was the quarter’s primary active driver. The second, the private AI portfolio, contributed mark-to-market gains that are a different kind of income from spread capture (addressed directly below). The third, the India options franchise, was not a material Q1 2026 contributor at all; it is included because litigation and a regulatory order exposed the market-making mechanism at execution level more completely than any other public record.

Jane Street: Two Active Q1 Drivers, and One Documented Case Study

Stream 1: Medium-Frequency Trading (Bloomberg/Reuters, May 2026)

Bloomberg’s people-familiar-with-the-matter sourcing explicitly identified medium-frequency trading strategies — machine-powered positions held for days or weeks — as the primary driver of Jane Street’s Q1 2026 performance. Reuters independently confirmed the same characterization: strategies “ranging from several minutes to days with the help of machines” drove the quarter. This is the key structural disclosure. Medium-frequency holding periods are better suited to geopolitical volatility regimes than pure HFT, because the signal half-life for macro-driven correlation shifts is measured in hours and days, not microseconds. The Iran-driven repricing of cross-asset correlations took hours to fully propagate. Firms holding positions for minutes to days captured the full repricing move. Firms that needed to flatten books every 30 seconds captured a single tick.

Stream 2: Private AI Portfolio

A CoreWeave 8-K exhibit filed with the SEC on April 15, 2026 contains the formal press release confirming that Jane Street committed approximately $6 billion to CoreWeave’s AI cloud platform and made a $1 billion equity investment in CoreWeave Class A common stock at $109.00 per share. The release quotes CoreWeave saying Jane Street “operates like a frontier lab, continually breaking new ground in deep learning and pushing the scale and complexity of their models.” This is the primary documentary source on both the financial terms and the strategic rationale.

Reuters confirmed Jane Street’s Q1 results were partly buoyed by its stakes in AI companies including Anthropic and CoreWeave — Jane Street had held a pre-existing CoreWeave position of approximately 19.99 million shares since August 2025, disclosed in a 13G SEC filing that made it the company’s fourth-largest shareholder at the time, and CoreWeave’s stock appreciated during Q1 2026. The separate $1 billion equity investment announced on April 15, 2026 — after Q1 ended — was an incremental addition to that prior position. Bloomberg reported that Jane Street’s Anthropic stake has appreciated with each successive funding round, with Bloomberg confirming on April 29, 2026 that Anthropic was weighing a fresh round at a valuation exceeding $900 billion — more than double its February 2026 valuation of $380 billion. Wikipedia documents that Jane Street also invested in Thinking Machines Lab’s $2 billion founding round in July 2025 at a $12 billion valuation alongside Andreessen Horowitz, Nvidia, AMD, and Cisco.

The mechanics of this revenue are documented in Jane Street’s SEC-filed financial statements. The 2024 annual financial statement of Jane Street Options, LLC (Form X-17A-5 filed February 27, 2025) shows the firm operates under SEC Rule 15c3-1(b)(1), the net capital exemption available exclusively to registered market makers, with a revolving credit facility from the parent entity of up to $6.5 billion outstanding at $1.463 billion as of year-end 2024. Private equity stake mark-ups pass directly through to parent entity equity without a corporate tax layer — a structure that makes the AI portfolio P&L highly capital-efficient relative to standalone fund structures.

A caveat the public reporting does not resolve, and one that materially affects how repeatable the quarter looks: the $16.1 billion is described as trading revenue, but Reuters notes it was partly buoyed by these private stakes. Mark-to-market gains on illiquid venture positions are a fundamentally different income source from repeatable spread capture, and the public figures do not break out how much of the $16.1 billion is each. The two should not be read as one number. The timing, however, bounds the contamination: the largest markups — Anthropic’s step toward a $900 billion-plus valuation and the $1 billion CoreWeave equity purchase — are both late-April 2026 events that fall in Q2, not Q1, and so do not inflate this quarter’s headline figure. What touches Q1 specifically is the quarter’s appreciation on the pre-existing CoreWeave position and whatever Anthropic was marked at through March 31 — real, but far smaller than the headline valuation jump. The honest read: medium-frequency spread capture was the dominant and repeatable driver; private markups were a secondary and largely non-repeatable tailwind; and the public number blends them without disclosure.

Stream 3: India Options Franchise

The India strategy is the best-documented of the three, because litigation forced it into the public record from two directions simultaneously. One clarification matters before the detail: this franchise was almost certainly not a material contributor to Q1 2026. The conduct documented below was the subject of SEBI’s July 3, 2025 interim order; Jane Street deposited ₹4,843.57 crore into escrow on July 14, 2025, and although it was subsequently permitted to resume India trading under heightened surveillance, the specific high-margin expiry-day strategy described here was curtailed well before the quarter began. The most recent activity SEBI documents is May 15, 2025 — three quarters before the period this article examines. What follows is therefore a forensic case study of the market-making mechanism at execution level, drawn from 2023–2025, not an account of a live Q1 2026 revenue line.

The India Anatomy: Two Trading Days From SEBI’s Own Files

SEBI’s interim order, numbered WTM/AN/MRD/MRD-SEC-3/31516/2025-26 and dated July 3, 2025, is a 105-page document describing trading surveillance across 18 derivative expiry days between January 2023 and March 2025. The document lists two strategies: an “Intra-day Index Manipulation” strategy observed on 15 days, and an “Extended Marking the Close” strategy observed on the remaining 3 days — the latter also appearing in NIFTY options in May 2025. The order’s background section states explicitly that SEBI initiated its preliminary examination based on April 2024 media reports about the Jane Street-Millennium lawsuit, which had inadvertently disclosed that Jane Street’s strategy involved India options.

January 17, 2024 — sourced directly from SEBI’s order as reported by Moneycontrol:

Between 9:15 AM and 11:46 AM, Jane Street purchased Bank Nifty constituent stocks and futures worth approximately ₹4,370 crore while simultaneously selling Bank Nifty options for approximately ₹32,115 crore. After noon, Jane Street sold Bank Nifty futures worth approximately ₹5,372 crore, creating a peak short position of approximately ₹46,620 crore in Bank Nifty index options, and the index closed near 46,064.45 — Jane Street made a profit of approximately ₹735 crore in the options segment and an intraday loss of approximately ₹61.6 crore in cash and futures, for a net gain of approximately ₹673.4 crore on that single expiry day.

July 10, 2024:

On July 10, 2024, Jane Street sold Bank Nifty futures worth approximately ₹2,800 crore and created a short position of approximately ₹44,154 crore in Bank Nifty options, with the resulting softening of the index closing generating a profit of approximately ₹225 crore.

SEBI’s analysis attributed the entire positive price impact in Bank Nifty during the morning trading patch on several examined days to Jane Street alone, with the rest of the market exerting net downward pressure simultaneously. SEBI’s order confirms that despite a caution letter from NSE issued on February 6, 2025, and Jane Street’s own commitments to the exchange, Jane Street continued to run very large cash-equivalent positions in index options as late as May 15, 2025. Continued operation after regulatory warning is the key element SEBI uses to support its characterization of the conduct as a “deliberately devised device” rather than coincidental hedging.

SEBI directed Jane Street to deposit ₹4,843.57 crore into an escrow account, representing alleged unlawful gains — the highest-ever impounding order issued by the regulator, according to SEBI Chairman Tuhin Kanta Pandey, who stated publicly that “market manipulation is not going to be tolerated”. Jane Street deposited the full amount and subsequently sought an extension from SEBI to respond to the interim order, confirming it was “engaging constructively” with the regulator while maintaining its characterization of the trades as standard index arbitrage. Business Standard reported in September 2025 that SEBI had expanded its probe beyond Bank Nifty to Sensex and other indices, with early findings suggesting wider alleged manipulation than covered in the July order.

The Manhattan Court Record: What the Trade Secret Case Reveals About Strategy Valuation

Jane Street filed its complaint against Millennium Management and former traders Douglas Schadewald and Daniel Spottiswood in the Southern District of New York in April 2024 under civil case number 24-cv-02783 before Judge Paul Engelmayer, with preliminary motions recorded under miscellaneous docket 1:24-mc-00175 on CourtListener. The strategy involved India options and had generated $1 billion in profits for Jane Street in 2023, a disclosure that emerged when lawyers for both sides inadvertently identified the market during a court hearing, as Bloomberg reported.

Jane Street alleged that its profits from the strategy fell approximately 50% in March 2024 after Millennium began using the same approach. Schadewald and Spottiswood disputed this, with defendants arguing that Jane Street’s India options team actually posted record results in the months after they departed, per previously sealed motions released in redacted form. The case was settled on mutually agreeable terms and dismissed by December 2024 according to a federal court filing, with terms undisclosed.

The case’s primary analytical value lies in the $1 billion single-strategy revenue figure it forced into the public record — and in the second-order consequence: SEBI’s own order confirms the investigation was triggered specifically by April 2024 Bloomberg and related media reports about the Millennium lawsuit disclosures. The lawsuit Jane Street filed to protect its strategy became the trigger for the regulatory scrutiny that cost approximately $564 million (₹4,843.57 crore) and India market access. Litigation in one jurisdiction created exposure in the jurisdiction where the strategy operated.

The Scale Arithmetic: What $22.5 Billion in One Quarter Implies

Hudson River Trading generated $6.4 billion in Q1 2026 revenue with approximately 1,000 employees, while Jane Street generated $16.1 billion with approximately 3,500 people — annualizing to approximately $25.6 million per head for HRT and approximately $18.4 million per head for Jane Street, compared to approximately $6 million per head at major investment bank front offices. The bank figure covers only revenue-generating staff. The HRT and Jane Street figures cover the entire firm including operations, compliance, and infrastructure.

HRT carries the higher per-head figure because it generates roughly 40% of Jane Street’s revenue with under a third of the headcount: $6.4 billion across about 1,000 people, against $16.1 billion across about 3,500.

In 2025, Jane Street, Hudson River, and Citadel Securities collectively generated more than $60 billion in trading revenues, per the Financial Times. The structural explanation for these margins is fixed-cost leverage: the marginal cost of additional revenue in this model is close to zero. Better models on the same infrastructure generate more without proportional headcount or capital additions. Banks cannot replicate this because their trading desks are embedded in institutions with fragmented legacy systems, risk committee overhead, and capital allocation tied to loan book requirements.

The Jane Street Options LLC annual financial statement (Form X-17A-5, 2024) shows the entity operating under SEC Rule 15c3-1(b)(1) — the net capital exemption available exclusively to registered market makers — with total assets of $16.4 billion and a revolving credit facility from the parent of up to $6.5 billion. This is the capital structure of a business that carries enormous notional exposure with a relatively compact equity base, justified by the speed and accuracy of its risk management.

Three Structural Risk Vectors That Do Not Appear in the P&L

Volatility normalization. Disruption Banking’s December 2025 analysis of HRT noted peer-leading profit margins of around 59% in Q3 2025, far outpacing the industry average. Q1 2026’s implied margin was substantially higher. Technology infrastructure costs of approximately $1 billion annually for HRT alone, calibrated to an elevated vol regime, compress significantly if geopolitical resolution returns markets to pre-2025 norms.

Regulatory extraterritoriality. Business Standard reported in June 2025 that SEBI had shared details of its investigation of Jane Street with the US SEC, establishing cross-jurisdictional precedent for review of expiry-day HFT mechanics. The India case demonstrated a compounding litigation risk: the lawsuit Jane Street initiated to protect its strategy became the trigger for regulatory scrutiny that cost approximately $564 million (₹4,843.57 crore) and market access, as confirmed by Business Standard’s July 14, 2025 deposit announcement. As these firms expand into Brazil, Taiwan, and other markets with growing derivatives ecosystems, the regulatory risk profile scales with each new geography — and the discovery footprint of any future litigation could again invite scrutiny.

AI homogenization and alpha decay. A 2026 arXiv paper by Chen and Meng, “AI-Driven Alpha Decay: Algorithmic Homogenization, Reflexive Signal Erosion, and the Paradox of Intelligent Markets,” validates empirically using SEC 13F holdings data across 99.5 million positions from 2013-2024 that simulated institutional portfolio convergence increased approximately 42% over the sample period as AI adoption increased. For market makers, the analog is model convergence: as more firms build similar neural network architectures on overlapping market data, the predictive edge degrades. The firms that currently dominate benefit from a first-mover advantage in compute infrastructure and training data accumulation — but that advantage erodes.

Conclusion: Prediction Is the Commodity, Scale Is the Moat

The alpha HRT and Jane Street are generating in Q1 2026 is not mysterious. It is the product of accurate short-term price prediction, applied at a scale that converts a sub-percentage-point statistical edge into tens of billions of annual revenue through compounding across millions of daily transactions. The EIA confirms the Iran war created the precise structural environment these models were built to exploit: the largest quarterly oil price increase on an inflation-adjusted basis in data going back to 1988, sustained for weeks, with cross-asset dislocations that required hours to reprice rather than milliseconds.

Bloomberg confirmed HRT’s Q1 2026 revenue was itself a record across any quarter in the firm’s documented history. Bloomberg separately confirmed Jane Street’s revenue more than doubled from Q1 2025 — a quarter that itself was not a low-base comparison. CoreWeave’s SEC filings confirm both firms are simultaneously investing the resulting capital into the next generation of training infrastructure. The feedback loop is self-reinforcing: trading revenue funds compute, compute improves models, better models improve trading revenue.

The Q1 2026 numbers are not a windfall. They are the compounding return on a decade of technology investment, harvested in the exact market regime those investments were designed to exploit.

📊 Want Deeper Quantitative Analysis?

This research took many hours of data collection, verification, and analysis. If you found value in this deep-dive, I publish exclusive quantitative research, trading strategies, and institutional-grade analysis on Patreon.

The practitioner’s companion note to this piece is already up — the core mechanism distilled, plus a complete worked example carrying the January 17 Bank Nifty trade end to end: the phase-by-phase P&L decomposition and the 91.6% efficiency ratio the Avellaneda-Stoikov framework predicts.

→ Read the companion research note on Patreon

→ Join the Patreon community here

By joining, you’ll be supporting my work and motivating me to publish more content like this.

Navnoor Bawa is a quantitative researcher and financial journalist covering derivatives, market microstructure, and systematic trading strategies.

Follow and connect:

• Patreon (exclusive research and trading strategies): patreon.com/NavnoorBawa

• Patreon — this article’s companion note: Market Making With Alpha Signals

• Substack: navnoorbawa.substack.com

• LinkedIn: linkedin.com/in/navnoorbawa

• YouTube: @TheMathematicalTrader

Cover photograph: King of Hearts, CC BY-SA 4.0, via Wikimedia Commons.

🎬 Prefer to watch rather than read? A NotebookLM-generated video overview of this article is available here: Watch the video overview → Full analysis, citations, and data remain in the article below.

Why the Spread Exists: The Glosten-Milgrom Proof

The foundational misconception in most market making analysis is that the bid-ask spread compensates the market maker for bearing inventory risk or processing costs. The 1985 proof by Lawrence Glosten and Paul Milgrom demolished that assumption four decades ago, and almost every subsequent advance in market making practice rests on what they showed.

Bid, Ask, and Transaction Prices in a Specialist Market with Heterogeneously Informed Traders, published in the Journal of Financial Economics, demonstrated that a positive bid-ask spread emerges even when the market maker is risk-neutral and earns zero expected profit. The abstract states the core result plainly: the spread exists because of traders with superior information. A risk-neutral specialist who breaks even in aggregate must still charge a spread — not to profit, but to recover from uninformed traders what is systematically lost to informed ones.

The model divides counterparties into two populations: informed traders who know the asset’s true value and trade profitably against the market maker; and uninformed liquidity traders who trade for exogenous reasons — hedging, rebalancing, cash needs. The market maker cannot distinguish them before a trade. The rational equilibrium spread is the one at which profits from uninformed flow exactly offset expected losses to informed flow. The spread is not a margin — it is a break-even condition under asymmetric information.

The practical implication is precise and actionable: every dollar of market making alpha attributable to information asymmetry must come from one of two sources — either (a) identifying informed flow before it trades, or (b) forecasting fair value with sufficient accuracy that even informed flow does not know more than the model. A third operational dimension — inventory velocity, governing how quickly a taken position is neutralized regardless of how it was priced — operates alongside these two and is separable from them.

These three dimensions are the entire framework. Everything that follows in this article is a real-world implementation of one or more of them, drawn from primary sources that are publicly available and legally binding.

The Virtu Record: What the SEC Filing Actually Says

The most-cited statistic in market making — Virtu Financial’s near-perfect daily trading record — is almost always reported with imprecision. The SEC-filed S-1 prospectus states it exactly: “we had only one losing trading day during the period depicted, a total of 1,238 trading days” from January 1, 2009 through December 31, 2013. That means 1,237 profitable days and one losing day across five years.

But the filing’s deeper content is more instructive than the headline figure. The same document discloses that in 2013, Virtu earned 27% of its Adjusted Net Trading Income from Americas equities, 11% from EMEA equities, 11% from APAC equities, 23% from global commodities, 20% from global currencies, and 9% from options, fixed income, and other securities. No single category exceeded 30% of total income. This diversification is not incidental — it is the structural source of the near-zero loss-day probability. With thousands of uncorrelated microstructure edges running simultaneously, the law of large numbers ensures that no single position-level adverse move can aggregate into a firm-level negative day.

A separate SEC correspondence filing (CORRESP) submitted during the IPO process makes Virtu’s actual trading mechanism explicit. Responding to SEC staff questions, Virtu described three discrete strategies. The first is “single instrument market making”: actively quoting a single instrument “with the intention of profiting by capturing the spread between the bid and offer,” placing bids and offers at or near the inside market simultaneously. The second is “multi-leg market making”: simultaneously quoting related instruments — for instance, an ETF and its underlying basket — and profiting from the relative spread between economically linked products. The third is “hedging” — using related instruments to neutralize price risk on accumulated inventory.

The 2015 S-1 re-filing added a critical disclosure: “for the 252 trading days of 2014, we averaged approximately 5.3 million trades per day globally across all asset classes, and we profitably exited 49% of our overall positions.” Fewer than half of Virtu’s individual trades were profitable. The CORRESP filing confirmed the 2014 Q1 figure: 49.6% of positions were unprofitable. The alpha is not in picking winners trade-by-trade. It is in the mathematical expectation of a 50.4% win rate multiplied 5.3 million times daily, diversified across more than 11,000 securities and 225 unique exchanges in 35 countries.

📊 Want deeper quantitative analysis?

This research required extensive data collection, verification, and analysis. If you found value in this breakdown, I publish exclusive quantitative research, trading strategies, and institutional-grade analysis for serious readers.

By joining, you directly support this work and make more research like this possible.

→ Join the Patreon community here

The Inventory Control Framework: Avellaneda-Stoikov in Practice

The Virtu record establishes that diversification and win-rate mathematics produce the aggregate edge. But that aggregate conceals a granular operational problem: every individual filled order creates a position, and an accumulating inventory is directional exposure by another name. For any market maker holding a position longer than a millisecond, inventory management is not a secondary concern — it is the central technical problem. The canonical academic solution was provided by Marco Avellaneda and Sasha Stoikov in High-Frequency Trading in a Limit Order Book, published in Quantitative Finance (Volume 8, Issue 3, 2008).

The paper formalizes an insight that every experienced market maker knows intuitively: when a market maker’s inventory is long, they should skew their ask tighter and their bid wider — making it cheaper to sell to them and more expensive to buy from them — to attract sellers and reduce the inventory. The model achieves this through a “reservation price,” the market maker’s private valuation of the asset given current inventory:

r = S − q · γ · σ² · (T − t)

where S is the mid-price, q is the inventory position (positive = long, negative = short), γ is the risk aversion parameter, σ² is the asset’s variance, and (T − t) is time remaining in the trading session. When q > 0 (long inventory), the reservation price drops below mid — the market maker’s private valuation reflects their urgency to offload. Bid and ask quotes are then set symmetrically around this private valuation rather than around the market mid-price, skewing quotes to attract the offsetting flow. As the session approaches close, (T − t) → 0 and the reservation price converges back to market mid, forcing inventory to flatten.

The practical importance of this framework is that it shows market making alpha is not purely about prediction accuracy. A firm with a mediocre price forecast but superior inventory management — one that never lets accumulated directional exposure compound — can outperform a firm with excellent forecasts but poor inventory discipline. The two dimensions are separable.

This is Lever 3 — inventory velocity — operating in parallel with the two information-asymmetry levers that Glosten-Milgrom identified. The five case studies that follow are each, at root, implementations of one or more of these three mechanisms. The 2012 Fodra-Labadie extension on arXiv adds directional bets and inventory penalties at horizon, extending the framework to the conditions real desks actually face.

XTX Markets: A Decade of On-Record Evidence

XTX Markets provides the most extensively documented case of what prediction-quality-over-latency market making looks like at scale. The firm was founded in January 2015 by Alexander Gerko, a PhD mathematician from Moscow State University who served as Head of FX Trading at GSA Capital before spinning XTX out. In 2022, XTX reported profits of £1.1 billion, a 64% increase on the prior year, on revenues of £2.5 billion, making Gerko the UK’s largest individual taxpayer in 2023 — paying approximately £664 million in personal tax that year — and culminating in a record £682 million personal distribution from XTX profits in 2024.

What makes XTX documentable is the directness of its senior leadership’s public statements about strategy. In a February 2017 Risk.net interview, co-CEO Zar Amrolia gave the clearest public articulation of the firm’s operational philosophy: “We’re not a latency-sensitive firm. We take risk and we operate a fair-value model.” The same article confirms that XTX holds positions “for more than 10 minutes on average in G10 markets and for 20 minutes in emerging markets” — timeframes that are orders of magnitude longer than pure latency arbitrage, and which require a confident short-term price forecast to be profitable.

The structural implication is important. A firm holding EUR/USD for 10 minutes is not operating as a latency arbitrageur. It is making a conditional prediction about where EUR/USD will trade over the next 10 minutes, updating that prediction continuously as new data arrives, and managing the resulting inventory position while the prediction plays out. If the prediction is wrong more than some threshold percentage of the time, the inventory losses exceed the spread capture and the strategy is unprofitable. XTX’s documented profitability is therefore direct evidence that their multi-asset pricing models are, on average, correct at 10-minute horizons — a statement that would be remarkable about any firm.

A second key disclosure came in December 2017 Risk.net awards coverage, when Amrolia described the firm’s decision to move to zero hold time on its bilateral FX quotes. Traditionally, FX liquidity providers used a “last look” window — a holding period of 50 to 200 milliseconds — during which they could cancel a trade after seeing the client’s order but before execution. This option protects the market maker from being filled on stale quotes during fast-moving markets. XTX eliminated this window entirely. The result: bilateral volumes with clients rose 50%, and the firm’s direct client count grew from 20 to 100 in a single year.

The mechanism the decision reveals: last look is an adverse selection buffer that protects per-trade profitability by allowing the rejection of trades that arrive precisely when the market maker’s quotes have become stale. By removing it, XTX accepted worse per-trade economics in exchange for flow volume growth. This trade-off is only rational if the underlying pricing model is accurate enough that last look provides marginal rather than essential protection — a self-referential claim about model quality that the documented bilateral volume growth corroborates.

Amrolia’s formal statement on the zero-hold-time decision, quoted in the Finance Magnates report at the time of announcement: “With recent improvements in technology and market data, we believe that it is hard to justify the use of a latency buffer in pricing. All direct counterparties of XTX Markets now enter into transactions either on a ‘Non Last Look’ or a ‘Zero Hold Time’ basis with no latency buffer.”

The Adverse Selection Instrument: VPIN at Tudor Investment Corp

The problem every market maker faces — distinguishing informed from uninformed flow in real time — had been formalized academically six years before XTX’s zero-hold-time decision, and the primary author was working inside a live hedge fund when he published it.

Marcos López de Prado’s institutional affiliation on the VPIN Flash Crash paper is significant: he is listed as “head of High Frequency Futures at Tudor Investment Corporation in Greenwich, CT.” This means the Volume-Synchronized Probability of Informed Trading (VPIN) metric was developed inside an active hedge fund, with access to live trading data, by someone with operational responsibility for high-frequency futures strategies. The paper was published in the Journal of Portfolio Management (Vol. 37, No. 2, Winter 2011).

The mechanism: VPIN divides trading volume into equal-sized “buckets” rather than equal time intervals, classifying each trade as buyer- or seller-initiated using bulk volume classification. The VPIN metric for each bucket is the absolute difference between buy volume and sell volume, divided by total volume. High VPIN — approaching 1 — means nearly all volume in a given period is one-directional, which is the statistical signature of informed, directional trading by agents who believe the asset is moving systematically. Low VPIN means flow is balanced, consistent with uninformed noise trading.

The authors explicitly proposed “a ‘VPIN contract’ that would allow liquidity providers to dynamically monitor and manage their risks” — acknowledging that the metric’s primary application is as a market maker’s real-time risk tool, not merely an academic measure. A companion analysis published as an NYU Stern working paper, Flow Toxicity and Liquidity in a High Frequency World, argued that VPIN reached its highest level for the full sample on May 6, 2010 — the day of the Flash Crash — and that the elevated toxicity caused market makers to pull liquidity, converting from liquidity providers to liquidity consumers. This claim has been disputed in peer-reviewed literature: Andersen and Bondarenko, writing in the Journal of Financial Markets (2014), found that VPIN actually peaked after the Flash Crash rather than before or during it. The operational value of VPIN as a real-time toxicity signal is documented regardless of the dispute over its precise Flash Crash timing.

The operational inference for any firm running a VPIN-type system in live markets: when VPIN exceeds a threshold, widen quotes or pull them entirely; when VPIN normalizes, return to standard spreads. This is the real-time implementation of Glosten-Milgrom’s equilibrium condition — adjusting the spread to maintain expected break-even profitability as the proportion of informed traders in the flow changes dynamically.

Citadel Securities: The SEC Administrative Record on Information Asymmetry

The most explicit primary documentation of how a market maker monetizes an information advantage comes not from any firm’s voluntary disclosures but from an SEC enforcement action.

On January 13, 2017, the SEC issued Administrative Proceeding Release No. 33-10280 against Citadel Securities LLC. The order found that two algorithms — FastFill and SmartProvide — were activated when Citadel’s systems detected discrepancies between the SIP (Securities Information Processor) feed and direct exchange data feeds. The SIP consolidates best bid and offer data from all exchanges; direct feeds deliver the same data faster, before the SIP has published the update. The latency gap between the two is typically measured in microseconds to milliseconds.

The SEC’s order is precise about what each algorithm did with this information gap. FastFill immediately internalized a marketable order at the SIP national best bid or offer — the stale price — rather than the better price already visible on the direct feed. SmartProvide did not internalize at the SIP price, nor did it seek to obtain an execution at the best available price by routing to the market. Instead it routed a non-marketable order — one not priced to execute immediately — bypassing the better price the direct feed was already showing.

The SEC press release quoted then-Acting Enforcement Director Stephanie Avakian: “Citadel Securities made misleading statements suggesting that it would provide or try to get the best prices it saw for retail orders routed by other broker-dealers.” Citadel settled for $22.6 million without admitting or denying the findings.

The enforcement document also quantifies Citadel’s market position at the time: the SEC order states that Citadel’s processing of retail equity orders “accounts on average for approximately 35% of the average daily volume of retail equity shares traded in the U.S. markets.” Roughly one-third of all US retail equity flow was running through algorithms that were using a real-time information advantage over the customers whose orders they were processing. This is the regulatory record of what market making alpha from information asymmetry looks like when it crosses compliance boundaries.

The Jane Street Trade Secret Litigation: A Billion-Dollar Strategy Made Public

Regulatory enforcement is one channel through which market making mechanics enter the public record. Civil litigation is another — and in 2024 it produced the most consequential accidental disclosure in recent market making history.

In April 2024, Jane Street Group filed suit in the United States District Court for the Southern District of New York, Case 1:24-cv-02783-PAE, against Millennium Management and two former traders, Douglas Schadewald and Daniel Spottiswood. The amended complaint alleged that the two traders, who had been “intimately” involved in developing a proprietary India options strategy at Jane Street, carried that strategy to Millennium upon departing, enabling Millennium to replicate it.

The complaint’s financial disclosures were extraordinary. BNN Bloomberg’s reporting on the court hearing confirmed that the India options strategy was revealed inadvertently by Millennium’s own counsel — who apologized upon doing so — and that a Jane Street attorney stated the strategy “was one of the firm’s most lucrative.” Judge Paul Engelmayer confirmed on record that Jane Street claimed the strategy generated approximately $1 billion in 2023 profits.

Filings reviewed by BNN Bloomberg confirmed additional details: Jane Street claimed its profits from the strategy fell by more than 50% in March 2024, which it attributed exclusively to Millennium beginning to deploy the same approach. Jane Street also stated it had spent “tens of millions of dollars” over multiple years developing the strategy, and that the firm relies on confidentiality agreements rather than non-compete clauses to protect its intellectual property. A key passage from the amended complaint states the strategy was “counterintuitive” and “built upon the years-long effort and findings of the algorithmic team.”

The strategy’s geography — Indian equity index options, primarily BANKNIFTY and Nifty 50 contracts on the National Stock Exchange of India — made it visible to Indian regulators. As documented in public filings, the case revealed the extraordinary scale and concentration of Jane Street’s derivatives trading in India, attracting the attention of SEBI. In July 2025, SEBI formally accused Jane Street of manipulating the BANKNIFTY and Nifty 50 index settlement prices and ordered ₹4,843 crore — approximately $566 million — in alleged profits placed into escrow, as confirmed by Reuters.

The litigation’s deepest relevance for market making analysis is structural: it publicly established that a single options strategy — concentrated in a single national derivatives market — generated $1 billion in profits in one year for a firm that makes markets across thousands of instruments globally. The case settled before its mechanics were fully disclosed. The docket remains Jane Street Group LLC v. Millennium Management LLC, 24 cv 02783, SDNY.

The Millennium settlement did not close the legal record on Jane Street’s market making operations. In February 2026, the court-appointed administrator winding down Terraform Labs filed suit against Jane Street in Manhattan federal court (Case No. 1:26-cv-1504), alleging that in May 2022, a wallet linked to Jane Street withdrew $85 million in TerraUSD from a decentralized liquidity pool within ten minutes of Terraform quietly withdrawing $150 million from the same pool — before any public announcement — and that this action contributed to TerraUSD losing its dollar peg and triggering the $40 billion Terra/Luna collapse. On April 23, 2026, Jane Street filed a motion to dismiss, calling the lawsuit “a transparent attempt to extract cash” and arguing that its largest trades occurred after damaging information about TerraUSD was already public. Jane Street denies all allegations. The case is active. Its relevance here is not the outcome — which is not yet determined — but the pattern: for the third time in two years, a federal court is being asked to determine exactly where Jane Street’s information advantage in financial markets ends and unlawful conduct begins. That question is the same question Glosten-Milgrom formalized in 1985. It has not been answered definitively in law.

Optiver’s CFTC Record: Where Market Making Mechanics Become Settlement Manipulation

The Jane Street case drew that regulatory line in options markets. The clearest documentation of where market makers have historically crossed it in futures markets — specifically by exploiting the mechanics of daily price settlement — is the Optiver CFTC enforcement record.

The CFTC press release dated April 20, 2012, covering a consent order entered by Chief Judge Loretta A. Preska of the SDNY, documented that Optiver and its traders engaged in 19 separate instances of manipulation over 11 days in March 2007, targeting NYMEX crude oil, heating oil, and gasoline futures. The mechanism: Optiver deployed a rapid-fire execution tool internally nicknamed “The Hammer” to execute large volumes of orders during the final minutes of trading, moving the settlement price — the price at which futures contracts are marked for daily P&L and final delivery — in the direction of pre-existing positions. The settlement included a $13 million civil monetary penalty plus $1 million in disgorgement, totaling $14 million.

The structural lesson is not primarily about the manipulation itself but about what it reveals regarding the mechanics of settlement-adjacent trading. Futures settlement prices are calculated from actual trades in a defined closing window. A market maker or principal trader with substantial derivatives positions has an inherent financial interest in where settlement falls. When the aggregate position is large enough and the settlement window is thin enough, the economics can favor trading in the underlying to move settlement — even at a loss on those trades — because the gain on the derivatives book outweighs the cost.

The CFTC’s Director of Enforcement, David Meister, stated: “Manipulative schemes like ‘banging the close’ harm market integrity.” The regulatory line between aggressive market making near settlement — legal — and manipulation of the settlement calculation itself — illegal — is precisely the line that the Jane Street-SEBI dispute re-opened in India in 2025, and that the Terraform Labs lawsuit re-opens in a federal court in New York in 2026, at scales that Optiver’s 2007 manipulation case did not approach.

The Three Structural Levers: A Unified Framework

The enforcement cases — Citadel’s SIP arbitrage, Optiver’s banging the close, Jane Street’s India options concentration — collectively define the outer boundary of permissible market making. What the voluntary disclosures and academic record show is the mechanism operating well inside that boundary.

Every documented market making P&L — Virtu’s 1,237-out-of-1,238-day record, XTX’s £1.1 billion profit year, Jane Street’s India billion, Citadel’s 35% retail flow dominance — decomposes into combinations of three structural levers.

Lever 1: Fair-Value Model Accuracy

The primary alpha source in prediction-quality market making is the accuracy of the firm’s estimate of where the asset will trade over its intended hold horizon. As XTX’s own public client documentation states, the firm’s pricing edge is “cross-asset pricing alpha — the ability to successfully forecast short-term price movements.” This is not vague marketing language. It is a direct statement that the bid and ask prices XTX quotes are functions of a multi-asset model that predicts near-term price direction, and that the profitability of the strategy depends on whether those predictions are correct.

Lever 2: Adverse Selection Identification

The second lever is the identification and segregation of informed flow. Every disclosed market making operation operates some version of a toxicity filter. The Virtu CORRESP filing describes their real-time risk system: “if our risk management system detects a trading strategy generating revenues outside of our preset limits, it will freeze, or lockdown that strategy and alert management.” At a trade-population level, this is automated detection of strategy-level adverse selection — when a quoting strategy begins generating unexpected losses, the system interprets this as evidence of informed counterparties and shuts the strategy down. Clients whose flow is systematically toxic are offered different, wider pricing; those with balanced, uninformed flow receive tighter spreads.

Lever 3: Inventory Velocity and Hedging

The third lever is how quickly accumulated inventory is neutralized. The Avellaneda-Stoikov framework handles this via quote skewing. In practice, inventory neutralization takes two forms: passive — waiting for the other side of the trade to arrive at skewed prices — and active — hedging in a related instrument. The Virtu CORRESP filing describes the active form precisely: “Virtu hedges with related securities when they accumulate inventory so that they would end the day flat in terms of risk but not necessarily in terms of their position.” The distinction between risk neutrality — no net Greeks or delta exposure — and position neutrality — flat shares — is the operative difference between market making and directional trading.

📌 If this analysis interests you, the full deep-dive is on Patreon: Virtu Made Money 1,237 Out of 1,238 Trading Days. Here Is Exactly How — Documented in SEC Filings, Federal Court Records, and CFTC Orders Filed by Virtu, XTX, Jane Street, Citadel, and Optiver Themselves This piece documents how each of the three levers is implemented in practice — with exact quotes from Virtu’s SEC CORRESP filing, Zar Amrolia’s on-record Risk.net interviews, and the Avellaneda-Stoikov reservation price derivation sourced to the original Quantitative Finance paper.

The Capacity Ceiling: Primary Evidence That Edges Erode

Market making alpha compresses structurally as competition intensifies, and the public record provides specific evidence of this compression.

Virtu’s own SEC filings trace the compression trajectory. The 2017 8-K earnings filing shows that average daily Adjusted Net Trading Income fell to approximately $1.175 million per day in Q2 2017, down from $1.599 million in Q2 2016 — a 26.5% decline in revenue per day despite trading in more than 12,000 financial instruments across 235 venues in 36 countries. Revenue declined across all categories: Americas equities fell 18.5%, global commodities fell 28.8%, global currencies fell 29.0%. This is spread compression at global scale, documented quarterly in SEC filings.

XTX’s response to capacity pressure was geographic and product expansion combined with the zero-hold-time strategic pivot. After moving to zero hold time, XTX grew its direct client base from 20 to 100 in 12 months — trading per-trade margin compression for volume scale. Jeremy Smart, XTX’s head of distribution, is quoted directly in the December 2017 Risk.net awards profile: “The jump in volume and clients has got a lot to do with people realising that having a zero hold time shows a clear demonstration of our commitment to transparency and fairness, which gives us extra weighting.” The strategic logic: when spread per trade falls, the competitive response is to win more trades — a volume-scale trade-off that only works if operating costs do not scale linearly with volume, which they do not in an automated market making operation.

The regulatory dimension of capacity compression is the most structurally uncertain. Payment for order flow was banned by the EU under MiFIR amendments that entered into force on March 28, 2024, subject to a transitional phase-out period extending to June 30, 2026 for certain member states. India’s SEBI actions against Jane Street in 2025 introduced regulatory uncertainty around structured expiry-day options positioning in developing derivatives markets. The CFTC’s FY2023 enforcement results documented 96 enforcement actions and $4.3 billion in penalties — evidence that the regulatory cost of running near the line of permissible market making conduct has become materially non-zero.

What the Record Shows and What It Doesn’t

The primary sources reviewed here collectively establish six claims with documentary specificity:

1. Virtu ran 1,238 trading days with one loss — meaning 1,237 profitable days — winning just 50.4% of individual trades, through diversification across 11,000+ instruments in 35 countries. Source: Virtu S-1 (2014) and CORRESP filing, SEC.

2. XTX holds positions 10–20 minutes on average in FX, explicitly choosing prediction quality over latency, and documented 50% bilateral volume growth and a client base expansion from 20 to 100 by removing last-look protection. Source: Risk.net, on-record interviews, 2017.

3. Citadel’s FastFill and SmartProvide used a SIP-vs-direct-feed latency gap to internalize retail orders at prices better for Citadel than its stated commitment required, while processing approximately 35% of average daily US retail equity volume. Citadel settled for $22.6 million. Source: SEC Administrative Proceeding Release No. 33-10280.

4. Jane Street ran a single India options strategy that generated approximately $1 billion in 2023 profits, experienced a greater than 50% profit decline when a competitor replicated the strategy, and spent tens of millions of dollars developing it. Source: SDNY Case 1:24-cv-02783-PAE, confirmed in open court.

5. VPIN, the primary real-time adverse selection metric, was developed by a practitioner at Tudor Investment Corp and explicitly proposed as a live market maker risk management tool. Source: Journal of Portfolio Management, Vol. 37, No. 2, 2011.

6. Optiver’s “banging the close” scheme involved 19 manipulation instances across 11 days in March 2007, resulting in a $14 million total settlement — $13 million civil monetary penalty plus $1 million disgorgement. Source: CFTC consent order, April 2012.

What the record does not show is the specific algorithmic content of any of these strategies. The Jane Street complaint’s key sections are redacted. The Citadel order describes behavior but not model architecture. XTX’s volatility surface parameterization, cross-asset signal weights, and toxicity classification thresholds remain proprietary. The public record establishes what these firms do and what it produces. How they do it at the model level is the trade secret the litigation was trying to protect.

For a quant PM evaluating the category: the documented evidence is sufficient to conclude that market making alpha is real, demonstrable over multi-year periods in public filings, and decomposable into three identifiable technical dimensions. The hard part is that the infrastructure requirements — co-location across 35+ countries, sub-millisecond risk monitoring, real-time cross-asset pricing models for 50,000+ instruments — represent capital and engineering barriers that compound with time. The firms whose records are documented here have been building those barriers since 2009. The public filings show the results. The source code is not in any of them.

📌 If this analysis interests you, the full deep-dive is on Patreon: Virtu Made Money 1,237 Out of 1,238 Trading Days. Here Is Exactly How — Documented in SEC Filings, Federal Court Records, and CFTC Orders Filed by Virtu, XTX, Jane Street, Citadel, and Optiver Themselves This piece contains every primary source linked inline at the point of the claim — SEC filings, CFTC consent orders, federal court complaints, and peer-reviewed academic papers — so every number is traceable to its named document without leaving the page.

Primary Sources

Virtu Financial — SEC Filings

• Virtu Financial S-1 (2014) — SEC EDGAR

• Virtu Financial S-1 (2015) — SEC EDGAR

• Virtu Financial CORRESP Filing — SEC EDGAR

• Virtu Financial 8-K Q2 2017 — SEC EDGAR

Academic Papers

• Glosten & Milgrom (1985), Bid, Ask, and Transaction Prices in a Specialist Market with Heterogeneously Informed Traders — Journal of Financial Economics, ScienceDirect

• Easley, López de Prado & O’Hara, VPIN Flash Crash Paper — SSRN

• VPIN, The Microstructure of the Flash Crash — Journal of Portfolio Management, Vol. 37, No. 2, Winter 2011

• Flow Toxicity and Liquidity in a High Frequency World — NYU Stern Working Paper

• Avellaneda & Stoikov (2008), High-Frequency Trading in a Limit Order Book — Quantitative Finance, Cornell PDF

• Fodra & Labadie, Inventory Extension to Avellaneda-Stoikov (2012) — arXiv

XTX Markets

• XTX Markets — Wikipedia

• Alex Gerko — MarketsWiki

• Zar Amrolia Interview, Currencies Flow Market-Maker of the Year 2017 — Risk.net

• XTX Streaming Liquidity Provider of the Year 2017 — Risk.net

• XTX Zero Hold Time, Streaming Liquidity Provider Awards 2017 — Risk.net

• XTX Zero Hold Time Announcement — Finance Magnates

• XTX FX Execution Algorithm Launch — Traders Magazine

Citadel Securities — SEC Enforcement

• Citadel Securities Administrative Proceeding Release No. 33-10280 — SEC PDF

• Citadel Securities Press Release 2017-11 — SEC Newsroom

Jane Street — Federal Court Litigation

• Jane Street Amended Complaint, Case 1:24-cv-02783-PAE, SDNY — Seward & Kissel PDF

• Jane Street India Strategy Revealed at Court Hearing — BNN Bloomberg

• Jane Street Record India Profits Complicate Millennium Fight — BNN Bloomberg

• Jane Street and Millennium Settle India Options Trade Secrets Case — BNN Bloomberg

• Jane Street Deposits $567 Million in Escrow Per India Regulatory Directives — Reuters

• Jane Street Hit with Terraform Labs Insider Trading Lawsuit — Disruption Banking

• Jane Street Seeks Dismissal of Terraform Insider-Trading Lawsuit — Bloomberg

• The $1 Billion Jane Street Clash and the High Stakes of Intangible Assets — The Fashion Law

• NPR Planet Money — Jane Street Episode Transcript

Optiver — CFTC Enforcement

• Optiver CFTC Consent Order, April 20, 2012 — CFTC.gov

• High-Frequency Trader Optiver Pays $14 Million in Oil Manipulation Case — Nasdaq

Regulatory and Policy

• CFTC FY2023 Enforcement Results — CFTC.gov

• Article 39a — Prohibition of Receiving Payment for Order Flow, MiFIR — ESMA Interactive Single Rulebook

🎥 Follow the analysis on YouTube

If you found this breakdown useful, I publish video-format quantitative research on my channel.

Subscribe on YouTube → https://www.youtube.com/@TheMathematicalTrader

🔗 Connect on LinkedIn

For institutional research discussion and professional updates:

Connect on LinkedIn → https://www.linkedin.com/in/navnoorbawa/

📊 Want deeper quantitative analysis?

This research required extensive data collection, verification, and analysis. If you found value in this breakdown, I publish exclusive quantitative research, trading strategies, and institutional-grade analysis for serious readers.

By joining, you directly support this work and make more research like this possible.

→ Join the Patreon community here

Cover photograph: Ken Lund from Reno, Nevada, USA, CC BY-SA 2.0, via Wikimedia Commons.

Cover photograph: Ken Lund from Reno, Nevada, USA, CC BY-SA 2.0, via Wikimedia Commons.

Ken Garschina and Mike Martino founded Mason Capital Management in July 2000 with $50 million, building it to approximately $13 billion by the mid-2010s before redemptions reduced AUM to $1.47 billion as of April 2025. The firm has engineered a repeatable process for extracting returns from situations most institutional investors cannot or will not touch: prepackaged bankruptcies, conflicted private equity continuation vehicles, cross-border governance disputes, and multi-year activist campaigns. The mechanics are visible in real time through Mason’s current battle against Energy & Minerals Group’s $5.5 billion continuation fund at Ascent Resources — a company Mason first acquired through distressed debt conversion during its 2018 bankruptcy, held through seven years of operational improvement, and now seeks to monetize by challenging EMG’s conflicted transaction at what Mason contends is a suppressed valuation.

The Distressed Debt Conversion: Ascent’s $1.2 Billion Restructuring

Mason’s Ascent position originated from Delaware bankruptcy case 18–10265-LSS filed February 6, 2018. The capital structure consisted of a $750 million first lien credit facility and $450 million second lien term loan, both administered by Cortland Capital Market Services. Ascent Resources Marcellus Holdings operated over 43,000 acres in West Virginia’s Marcellus Shale but faced overleveraged balance sheet as natural gas prices declined.

The prepackaged restructuring eliminated execution uncertainty. 78% of first lien holders and 79% of second lien holders voted to approve the plan before filing, converting all $1.2 billion in secured debt to equity in the reorganized entity. Mason, which had accumulated positions in the distressed loans during pre-bankruptcy trading, emerged 45 days later on March 30, 2018 as an equity stakeholder in a deleveraged company.

The trade’s elegance lies in its risk-adjusted structure. Senior secured creditors purchased loans trading at distressed levels — potentially 50–70 cents on the dollar — and received par value through equity conversion. If $1.2 billion of secured debt were acquired at an average of 60 cents on the dollar (cost $720 million) and the reorganized equity is now valued at $5.5–6 billion, this represents an investment multiple of approximately 7.6–8.3x, or gross returns of roughly 664–733% (~34–35% annualized over seven years). Actual realized returns depend on Mason’s specific position size, distributions, and fees.

The Continuation Fund Battle: Exploiting GP-LP Conflicts

Seven years after emerging from bankruptcy, Ascent Resources — now producing 2.2 billion cubic feet equivalent per day as Ohio’s largest natural gas producer — became the focal point of a high-stakes continuation fund battle. Mason Capital’s 2018 distressed debt position had positioned the firm as a significant stakeholder, setting the stage for its most visible activist campaign.

On November 26, 2025, Abu Dhabi Investment Council filed Case №2025–1389-NAC in Delaware Court of Chancery against EMG. The verified complaint alleges EMG engineered a conflicted sale of its Ascent stake into a continuation vehicle at $23.87 per share — an implied $5.5 billion valuation — while simultaneously serving as fiduciary to both buyer and seller. ADIC claims EMG provided materially different information to existing investors versus prospective CV investors, refused in-camera LPAC sessions, and based fairness opinions on inaccurate assumptions.

The economic structure compounds conflicts. Rollover investors must pay new management fees and carried interest rather than maintaining existing terms — transferring value from selling LPs to EMG. The continuation vehicle resets EMG’s carry hurdles, allowing performance fees on future appreciation that wouldn’t generate carry under existing fund structure. 80–90% of LPs typically cash out of continuation vehicles rather than rolling forward — a striking vote of no confidence in GP-proposed valuations.

Mason’s December 12, 2025 letter to Ascent’s board corroborated ADIC’s allegations and offered superior alternative. The hedge fund announced preparedness to deliver a fully financed, all-cash proposal at a price above EMG’s $5.5 billion valuation. Four days later, Kimmeridge Energy Management submitted a $6 billion bid — approximately 10% premium — validating claims that EMG had undervalued the asset. The competing bid’s emergence within days demonstrates how activist disclosure catalyzes price discovery in opaque private transactions.

In January 2026, Mason challenged Kirkland & Ellis’s role as counsel to both Ascent’s board and EMG, arguing the law firm faced disqualifying conflicts representing directors while advising the PE sponsor executing a conflicted transaction. EMG and ADIC subsequently agreed to arbitration per limited partnership agreements, with the continuation fund transaction on hold pending resolution.

The Samsung Arbitration: Treaty-Based Dispute Resolution

While the Ascent campaign demonstrates Mason’s ability to challenge private equity conflicts in U.S. courts, the firm’s Samsung C&T arbitration showcases its willingness to deploy cross-border legal mechanisms for value realization. Mason’s most quantifiable realized return came through international investment arbitration against South Korea. The firm accumulated a 2.18% stake in Samsung C&T before the company’s July 2015 merger with Cheil Industries. Mason alleged government improperly pressured the National Pension Service — Samsung C&T’s largest shareholder with 11.21% — to approve the merger benefiting Samsung’s founding Lee family at minority shareholders’ expense. The merger succeeded by only 0.54% margin.

Mason filed investor-state dispute proceedings under the Korea-US Free Trade Agreement in 2018 seeking $200 million in damages. The case proceeded under Permanent Court of Arbitration rules in Singapore. After six years of litigation, arbitrators awarded Mason $32 million in principal damages in April 2024. South Korea challenged the award in Singapore courts but lost. The government ultimately paid Mason $54 million including accrued interest in July 2025 and declined to appeal further in April 2025.

The arbitration demonstrates sophisticated legal strategy unavailable to most investors. Investment treaties create parallel dispute resolution mechanisms outside domestic courts, allowing foreign investors to sue sovereign governments for treaty violations. Mason deployed FTA protections to convert an adverse corporate governance outcome into a $54 million realized gain — representing potential 100–150% returns if the original Samsung C&T stake cost $36–54 million at 2015 market prices.

The Grifols Campaign: Forensic Analysis as Activist Weapon

Mason’s ongoing activism at Spanish pharmaceutical company Grifols demonstrates how the firm weaponizes forensic analysis to challenge entrenched governance failures. Unlike the binary outcomes of bankruptcy restructurings or treaty arbitrations, the Grifols campaign illustrates Mason’s capacity for sustained, multi-year pressure campaigns backed by granular financial analysis.

Holding approximately 2.1% of Grifols class A shares, Mason documented systematic value destruction through conflicted acquisitions totaling over €4.5 billion since 2014.

The November 8, 2024 letter to Grifols’ board quantified performance post-acquisition for major transactions: Novartis Diagnostics, Hologic NAT, and Biotest. Mason calculated that Diagnostics EBITDA declined 44% on average from 2017–2021 to 2022–2024. Biotest’s reported €129.5 million LTM EBITDA included €123.8 million from intercompany technology services for Grifols — circular accounting inflating acquisition performance metrics.

Every large transaction exceeding €1 billion since 2014 destroyed shareholder value while generating advisory fees for Osborne Clarke Spain, where board member Tomas Daga served as founding partner. Mason identified Daga as having led the Diagnostics and Biotest acquisitions according to the board’s own admissions. The firm calculated that Osborne Clarke Spain advised on approximately 16 M&A transactions worth €8.5 billion — creating obvious conflicts when a board member personally benefits from deals harming shareholders.

These debt-financed acquisitions elevated Grifols’ consolidated net leverage to 5.1x. Mason argued that without these value-destroying transactions, leverage would stand at a prudent 3.6x. The elevated debt load left Grifols vulnerable when Gotham City Research published a short report in January 2024 questioning accounting practices. Grifols lost approximately 33% of market value within hours.

Mason’s November 19, 2024 follow-up letter demanded the board accept Daga’s voluntary resignation and appoint Paul Herendeen as independent director. When Brookfield Asset Management and the Grifols family reportedly proposed a €10.50 per share take-private, Mason characterized the bid as substantially undervaluing the company and resulting directly from governance failures that depressed the share price. The proposed transaction collapsed in November 2025 due to valuation disagreements.

In January 2025, Mason escalated by filing a complaint with Spain’s securities regulator CNMV, alleging failure of internal controls, conflicts of interest in related-party transactions, and a bond issuance clause benefiting Brookfield at other shareholders’ expense.

Return Drivers: Deconstructing Alpha Generation

The Ascent, Samsung, and Grifols campaigns reveal a coherent strategic framework rather than opportunistic case-by-case investing. Mason Capital’s returns derive from exploiting four interconnected strategies most institutional investors cannot execute:

Distressed Debt Conversion with Loan-to-Own Execution: Purchase senior secured debt of overleveraged companies at 50–80 cents on the dollar during pre-bankruptcy distress. Participate actively in creditor committee negotiations to shape reorganization plans favorable to senior creditors. Convert debt claims to equity at par value in the reorganized entity. If the deleveraged company’s enterprise value exceeds face amount of converted debt, capture 100–200%+ returns. Ascent’s $1.2 billion debt converted to 100% equity ownership in an asset now valued at $5.5–6 billion represents approximately 7.6–8.3x multiple (664–733% gross returns) over seven years, assuming average 60 cent purchase prices.

Event-Driven Catalysts with Hard Deadlines: Focus exclusively on situations featuring court-imposed deadlines, regulatory approval timelines, or contractual milestone dates. These catalysts force price discovery and eliminate indefinite holding periods. Ascent’s prepackaged bankruptcy required court approval within 45–60 days of filing. Samsung merger vote occurred on specific scheduled date. Continuation fund transactions trigger LP election deadlines.

Activist Intervention to Force Competitive Processes: When accumulating meaningful minority stakes, publicly challenge transactions transferring value to conflicted parties through detailed letters exposing governance failures, financial misrepresentations, and procedural irregularities. Force competitive sale processes by offering superior bids or attracting alternative bidders. Ascent campaign attracted Kimmeridge’s $6 billion bid within days of Mason’s public disclosure. Samsung arbitration converted governance failure into $54 million realized proceeds.

Legal and Regulatory Arbitrage Across Jurisdictions: Deploy investor-state dispute mechanisms, bankruptcy court procedures, Delaware corporate law, and securities regulation to protect investment positions. Samsung arbitration utilized Korea-US FTA investor protections unavailable to Korean domestic investors. Ascent litigation invoked Delaware Chancery Court jurisdiction over limited partnership agreement disputes. Grifols activism engaged Spanish securities regulator CNMV oversight.

Information Asymmetry as Competitive Moat

Mason’s edge derives from superior information processing in genuinely opaque situations combined with operational capacity most institutional investors lack.

Opaque Transaction Analysis: Continuation funds operate within what Harvard Law School researchers describe as a “black box” — limited partnership agreements and valuations rarely reach public markets. Mason’s ability to challenge EMG’s Ascent continuation fund valuation required analyzing confidential offering materials, comparing representations made to different investor classes, and identifying discrepancies in operational metrics used for fairness opinions. This forensic accounting demands specialized expertise unavailable to passive limited partners.

Bankruptcy Court Participation: Distressed debt investing requires active engagement in creditor committee meetings, plan of reorganization negotiations, and confirmation proceedings. Mason participated in Ascent’s 45-day prepackaged bankruptcy as a creditor voting on the restructuring plan. This operational capacity — maintaining professionals who understand bankruptcy procedures and can negotiate with other creditor constituencies — creates barriers to entry for generalist credit investors.

Cross-Border Legal Sophistication: The Samsung arbitration required filing under Korea-US FTA provisions, navigating Permanent Court of Arbitration procedures, defending the award in Singapore courts, and collecting from a sovereign defendant. Mason’s six-year commitment to international arbitration required expertise most equity-focused hedge funds lack. The $54 million realized return validated this approach but demanded legal sophistication in investment treaty law.

Concentrated Portfolio Construction: With only six disclosed public equity positions totaling $491 million as of Q3 2025, Mason operates with extreme concentration enabling deep fundamental analysis impossible for diversified portfolios. This concentration is strategic, not accidental — each position represents months of research, legal preparation, and engagement before capital deployment. The firm’s Form ADV indicates $1.47 billion in total AUM as of April 2025, suggesting substantial capital deployed in private companies like Ascent, distressed debt not requiring public disclosure, and special situations outside 13F reporting requirements. This structure allows 10–25% position sizing that would constitute career risk for institutional portfolio managers operating within diversified mandates.

📊 Want Deeper Quantitative Analysis?

This research took very long time of data collection, verification, and analysis. If you found value in this deep-dive, I publish exclusive quantitative research, trading strategies, and institutional-grade analysis on Patreon.

By joining, you’ll be supporting my work and motivating me to publish more content like this.

→ Join the Patreon community here

The Continuation Fund Reckoning

The Ascent continuation fund battle arrives at a critical inflection point for private equity. Mason’s campaign arrives as private equity confronts structural reckoning over continuation fund practices. Continuation vehicles accounted for approximately 19% of all private equity asset sales in H1 2025, a 60% increase over H1 2024 — representing nearly 10% of all PE distributions to LPs. This volume growth occurs despite mounting evidence of systematic conflicts.

The SEC’s Private Fund Adviser Rules adopted August 2023 mandated quarterly performance reporting, detailed fee disclosure, and fairness opinions for conflicted transactions — regulations born directly from continuation fund controversies. Although the Fifth Circuit vacated these rules in June 2024, underlying market dynamics remain unchanged and regulatory attention continues escalating.

For Mason specifically, Ascent represents ideal convergence of all core strategies: distressed debt position acquired through bankruptcy at depressed valuations, held through seven-year operational improvement phase, then monetized when controlling sponsor attempts conflicted transaction that activist intervention disrupts. Whether through Mason’s acquisition, Kimmeridge’s $6 billion bid, negotiated settlement at higher valuations, or arbitration award to ADIC, minority stakeholders achieved primary objective — competitive price discovery replacing GP-controlled valuations.

The broader lesson extends beyond any single campaign. Mason Capital has engineered a repeatable formula for extracting returns from corporate distress, governance failures, and structural conflicts — situations most institutional investors cannot or will not touch. From bankruptcy court creditor committees to international arbitration tribunals to public activist campaigns, the firm deploys legal sophistication, forensic analysis, and concentrated capital across multiple jurisdictions and asset classes. The playbook is replicable but requires operational capabilities few possess: willingness to hold illiquid positions for years, expertise across bankruptcy law and investment treaties, and tolerance for public confrontation with entrenched control shareholders.

As continuation vehicles represent nearly 10% of PE distributions and continue growing, the template Mason has documented — minority stakeholder opposition backed by credible alternatives, sophisticated legal strategy, and public pressure — will attract replication from other event-driven managers with requisite capabilities. Private markets, despite opacity and structural advantages favoring sponsors, are not immune to activist scrutiny when meaningful capital and expertise challenge conflicted transactions.

Mason Capital Management operates with approximately 18 investment professionals managing concentrated positions across distressed debt, bankruptcy situations, activist campaigns, and special situations. The firm’s reported AUM of $1.47 billion as of April 2025 reflects significant decline from the approximate $13 billion peak in the mid-2010s, though exact performance attribution remains undisclosed. Current disclosed equity positions represent 100% concentration in top holdings, suggesting substantial capital deployed in non-public situations.

📺 Connect & Subscribe

YouTube: Deep-dive quantitative analysis and trading strategies
→ Subscribe to The Mathematical Trader

LinkedIn: Professional insights and research updates
→ Connect with Navnoor Bawa

Patreon: Exclusive institutional-grade research and analysis
→ Join the community

Cover photograph: Meredithw at English Wikipedia, CC BY-SA 3.0, via Wikimedia Commons.

Complete analysis of Da Vinci Trading’s employee-owned volatility arbitrage operation spanning crypto market making (MEXC, Kraken), traditional derivatives (Euronext), token lending (Skynet acquisition), and multi-market alpha generation strategies. Includes operational analysis, employee compensation data, technology infrastructure, and competitive positioning.

How the employee-owned prop firm systematically extracts alpha through delta-neutral arbitrage spanning crypto market making, TradFi derivatives, and strategic acquisitions

When MEXC announced Da Vinci Trading as a designated market maker in September 2025, it marked the latest evolution in a systematic multi-market expansion strategy. For institutional observers, this partnership represents more than a routine exchange-MM relationship. It demonstrates how a pure proprietary trading firm specializing in “trading volatility across multiple instruments” methodically transfers traditional derivatives expertise across fragmented crypto liquidity, traditional markets, and strategic acquisitions.

Da Vinci, founded in 2015, operates across Amsterdam (headquarters), Miami, Mumbai, and Hong Kong. The firm’s competitive advantage stems from quantitative sophistication applied systematically across traditional derivatives, crypto spot markets, crypto derivatives, and lending infrastructure.

This analysis deconstructs Da Vinci’s complete alpha generation mechanism across all markets and strategic initiatives.

Analysis roadmap: We examine the firm’s operational architecture and culture (employee ownership, compensation, geographic expansion), core volatility arbitrage mechanics across traditional and crypto derivatives, the strategic rationale behind the MEXC partnership and Skynet acquisition, quantitative infrastructure and technology stack, competitive positioning, and economic performance attribution. The conclusion synthesizes implications for institutional capital allocation and market structure evolution.

Core Operational Architecture

Da Vinci emphasizes it’s “owned and managed by its own employees”, creating direct ownership alignment. Glassdoor data reveals quantitative researchers earn €150K-€300K total compensation (€90K base + €135K bonus), competitive with European tech but below Citadel/Jane Street levels. This suggests selective hiring focused on candidates valuing intellectual challenge and ownership over pure compensation maximization.

Interview testimonials document rigorous selection: “Around 50 tough questions in 10 minutes. Its much tougher than what you might get at Optiver screening... I did around 35 with around 95% perfection.”

Geographic Expansion

Da Vinci’s Mumbai office launched in 2022-2023, with Miami expansion announced in 2024. This geographic footprint reflects operational necessity: latency optimization, regulatory arbitrage, and multi-timezone coverage.

Strategy Deconstruction: Volatility Arbitrage Mechanics

Primary Revenue Engine

Da Vinci’s Crunchbase profile states the firm “specializes in trading volatility across multiple instruments.” A Euronext bell ceremony announcement elaborated: “Da Vinci identifies and trades opportunities in financial derivatives based on in-house developed strategies with a short to medium term horizon focused on volatility.”

The announcement specified instruments: “options and futures on single stock, index and fixed income but also options and futures on volatility indices and dividends.”

Real-Time Model Adaptation

A LinkedIn post revealed operational agility during market dislocations: “Markets can be unpredictable, and every so often you have to think fast and quickly adapt the strategy to keep our edge in the market. For example, during COVID the oil prices went negative, which is something no one has seen before, making the pricing models worthless. It was a challenge, but within hours we already had a working model.”

This demonstrates three characteristics: proprietary pricing infrastructure capable of recalibrating within sub-day timescales, tight quant-trader feedback loops enabling rapid strategy deployment, and mathematical first-principles approach when Black-Scholes fails.

Multi-Venue Operations

On Kraken, Da Vinci ranks “among the top 3 trading firms by volume.” Erik Lubach, Head of Business Development, noted: “Kraken’s robust infrastructure and commitment to market integrity enabled us to refine our trading strategies and maintain a competitive advantage.”

The MEXC Partnership: Spot Market Volatility Harvesting

The MEXC announcement positioned the partnership as enhancing “liquidity infrastructure” with “tighter spreads and more reliable execution” for MEXC’s 40 million users across 170+ countries.

The partnership’s value proposition differs from pure HFT. Spot markets on exchanges like MEXC present structurally different opportunities: wider spreads, lower algorithmic sophistication, and retail-dominated order flow.

Infrastructure Advantages

MEXC provides market makers “enhanced API rate limits, customized interfaces, robust trading limits, and dedicated one-on-one professional support” with “highly competitive maker fee rebates.”

Execution Strategy

Da Vinci’s likely operational framework:

1. Post two-sided quotes in spot markets across major pairs

2. Accumulate inventory imbalances throughout trading sessions

3. Hedge net delta exposure via perpetual futures on high-liquidity venues

4. Manage gamma exposure through options positions on crypto derivatives exchanges

5. Extract spread income while maintaining delta-neutral positioning

This combines spot MM (revenue from bid-ask capture) with derivatives hedging (risk neutralization), creating multi-product volatility arbitrage.

The Skynet Acquisition: Vertical Integration

In May 2025, Da Vinci acquired Skynet Trading’s token loan desk. Areta’s announcement stated: “With the acquisition of Skynet’s token loan desk, Da Vinci takes a natural step towards liquidity provisioning for token projects, complementing its full array of market making and OTC activities in the digital asset space.”

Token lending infrastructure enables three capabilities: short inventory sourcing for delta-neutral positions, financing cost optimization through lending excess long inventory, and operational intelligence from lending data revealing short interest concentrations.

Quantitative Infrastructure

Technology Stack

Job postings emphasize C++ expertise: “Part of Da Vinci’s core is developing highly optimised and innovative solutions to bring our trading to the level it needs to be.” C++ provides memory management control, hardware-level optimization, and low-level network stack access critical for latency optimization.

Interview Process Reveals Strategy

Technical questions test: real-time risk calculation (options Greeks), Bayesian updating, risk-reward optimization (variance minimization vs. expected value maximization), and market making intuition under uncertainty.

Employee Testimonials: Operational Insights

Vincent, Digital Assets trader: “As a trader on Da Vinci’s Digital Assets desk, my job is to market make crypto currencies on multiple platforms across the world. Since I joined the desk, our team has been continuously working on improving the strategies and expanding its presence globally.”

He continued: “The best thing in my opinion about working on the Digital Assets desk is its fast-paced nature. Upon coming up with an idea on how to improve our strategy and discussing it with the team members, the change is often implemented immediately after testing.”

Robi, Head of Digital Assets: “One case that stands out to me is the way we handled a big move in one of the stable coins. I was so impressed with how each part of our team stepped up immediately and managed to solve a before unseen situation in just a couple of hours.”

Competitive Positioning

Da Vinci’s approach differs from pure latency-arbitrage: “We use short to medium-term strategies to trade various financial instruments worldwide.” This suggests strategies with holding periods ranging from minutes to days, not microsecond-level positioning of true HFT, aligning with quantitative model-driven approaches.

The firm states: “Our company is owned and managed by its employees and is highly meritocratic and entrepreneurial by nature.” This creates direct incentive alignment, long-term focus without external LP pressure, and knowledge retention through structural incentives.

Performance Attribution and Economics

Operational Scale

Operating across four global offices (Amsterdam, Miami, Mumbai, Hong Kong), Da Vinci maintains a lean operational structure typical of quantitative proprietary trading firms. The firm’s employee ownership model and proprietary capital structure enable patient positioning without external LP redemption pressures.

Market Making Economics

Market making on centralized exchanges generates revenue through bid-ask spread capture combined with exchange maker rebates. Combined with Kraken (top 3 by volume), traditional derivatives markets via Euronext, and token lending operations, the firm operates diversified revenue streams across multiple asset classes and geographies.

Strategic Implications for Institutional Allocators

Market Structure Insights: What Da Vinci’s Success Reveals

Da Vinci’s success validates three characteristics:

Volatility Mispricing Persists: The firm’s ability to scale operations while maintaining profitability suggests crypto markets continue to misprice volatility, differing from mature TradFi markets where vol arbitrage opportunities compress rapidly.

TradFi Expertise Transfers Profitably: Employee testimonials emphasizing founder involvement suggest competitive advantage stems from accumulated derivatives expertise, not merely technology or capital.

Multi-Venue Operations Create Structural Edge: Trading “on multiple platforms across the world” enables liquidity aggregation, venue arbitrage, and regulatory arbitrage.

Investment Thesis Validation

Delta-Neutral Strategies Provide Uncorrelated Returns: Volatility arbitrage with proper hedging generates returns uncorrelated to directional crypto exposure. Da Vinci’s focus on “trading volatility across multiple instruments” represents systematic extraction of volatility risk premium without directional beta.

Vertical Integration Captures More Value Chain: The Skynet acquisition demonstrates strategic evolution toward integrated infrastructure provision, creating multiple revenue streams, operational leverage, and information advantages.

Proprietary Capital Enables Patient Positioning: Trading “solely with our own money” eliminates external LP pressures, enabling longer holding periods, strategy pivots without redemption risk, and opportunistic capital deployment during dislocations.

Evolution Trajectory

Historical Progression

Phase 1 (2015-2020): Founded in 2015 trading options and futures on European and US exchanges.

Phase 2 (2020-2024): Digital Assets desk launched circa 2021-2022.

Phase 3 (2024-2025): Mumbai office launch, Miami establishment, Skynet acquisition, MEXC partnership.

Phase 4 (2026+): Likely trajectories include structured products distribution, prime brokerage services leveraging lending infrastructure, and proprietary exchange development.

Competitive Moats

Defensive Moats: 10+ years derivatives experience versus crypto-native 3-5 years, geographic and product diversification, token lending infrastructure creating switching costs, and proprietary capital enabling patient positioning.

Threats: TradFi encroachment from Jump Trading and Jane Street, native competitors like Wintermute developing derivatives expertise, and market structure evolution compressing vol premiums toward TradFi levels.

Da Vinci’s response appears to be vertical integration and geographic expansion to access less efficient markets in Asia and South America.

Conclusions: The Multi-Market Volatility Extraction Model

Core Findings

Da Vinci operates an integrated multi-market volatility arbitrage platform spanning four distinct revenue streams:

1. Traditional derivatives (options and futures on equities, indices, fixed income via Euronext)

2. Crypto spot market making (MEXC partnership, Kraken top 3 by volume)

3. Crypto derivatives (for delta hedging spot positions)

4. Token lending infrastructure (via Skynet acquisition)

The competitive advantage stems from four integrated capabilities: quantitative sophistication enabling sub-day model recalibration, operational agility through tight quant-trader feedback loops, multi-venue presence providing geographic and product diversification, and vertical integration of critical infrastructure.

Strategic Implications for Institutional Capital

What the MEXC Partnership Reveals: The September 2025 engagement isn’t isolated crypto speculation. It represents systematic application of TradFi derivatives expertise to fragmented crypto liquidity while maintaining core volatility arbitrage operations across traditional markets. This demonstrates strategy portability and revenue diversification beyond directional volatility bets.

The Skynet Acquisition Thesis: Vertical integration into token lending creates operational leverage (controlling short inventory sourcing), financial optimization (lending yield offsets carry costs), and information advantages (lending data reveals positioning flows). This moves Da Vinci from pure alpha extraction toward integrated infrastructure provision.

Investment Framework: Da Vinci’s employee-owned, proprietary capital structure eliminates external LP redemption risk and quarterly performance pressures. Operating across four global offices, the firm demonstrates capital-efficient operations with patient positioning capability unavailable to traditional fund structures facing quarterly redemption cycles.

Final Assessment

The MEXC partnership exemplifies quantitative precision in pricing complex, multi-dimensional risks that most market participants can’t properly value. That edge—built through 10+ years of traditional derivatives market making, adapted systematically to crypto’s unique characteristics, and executed with disciplined risk management—generates consistent alpha across traditional and digital asset markets simultaneously.

The firm’s multi-market approach, operational agility during crises, and strategic vertical integration position it to capture volatility premiums as crypto markets mature while maintaining traditional derivatives operations as stable revenue foundations.

📊 Want Deeper Quantitative Analysis?

This research took very long time of data collection, verification, and analysis. If you found value in this deep-dive, I publish exclusive quantitative research, trading strategies, and institutional-grade analysis on Patreon.

By joining, you’ll be supporting my work and motivating me to publish more content like this.

→ Join the Patreon community here

Follow me on YouTube for video breakdowns and LinkedIn for more quant research.

Cover photograph: Guilhem Vellut from Annecy, France, CC BY 2.0, via Wikimedia Commons.

Cover photograph: Guilhem Vellut from Annecy, France, CC BY 2.0, via Wikimedia Commons.

Boaz Weinstein’s Saba Capital $2 billion flagship credit fund is down 6.5% in 2025. His $1 billion tail hedge fund has bled 12.7% this year. Yet the firm manages $6 billion in regulatory assets as of August 2025, earned Institutional Investor’s “Activist Hedge Fund Manager of the Year” in both 2023 and 2024, and runs five structurally different profit engines that most observers treat as separate businesses.

This isn’t a contradiction. It’s deliberate architecture. Each engine generates alpha under different market conditions: quantitative credit arbitrage (London Whale), tail-hedged carry trades (COVID), discount-to-NAV capture (SPACs), closed-end fund activism, and investment-grade CDS underwriting. When credit markets are calm — as in 2025 — the flagship struggles. When volatility spikes, it prints money.

This is the forensic breakdown of how each engine generates returns, starting with the foundational strategy that powers all five.

The Structural Edge: Credit vs. Equity Derivatives Mispricing (1998-Present)

Weinstein’s foundational alpha comes from a pricing anomaly that persists across cycles: credit and equity derivatives on the same company price risk differently.

At Deutsche Bank from 1998–2009, where he became Managing Director at 27 while running a $30 billion internal hedge fund, Weinstein pioneered capital structure arbitrage. The strategy: buy out-of-the-money puts on a stock, hedge with the company’s bonds. When spreads converge, extract the differential.

Documented trades:

Household International (2002): Bought bonds trading at 800bps over Treasuries, CDS at 900bps. Partially hedged by shorting equity. HSBC’s acquisition announcement tightened credit spreads violently — bond gains exceeded equity hedge losses.

General Motors (2005): Sold CDS protection on GM debt while shorting the stock as GM faced junk rating but maintained elevated equity prices.

Deutsche Bank P&L:

• 2006: +$900 million

• 2007: +$600 million

• 2008: -$1.8 billion

• January 2009: Recovered ~$600 million

Weinstein left Deutsche Bank in February 2009 with 15 team members, forming Saba Capital in April 2009. The firm launched its flagship Capital Master Fund in August 2009 with $140–160 million in initial capital.

The capital structure arbitrage foundation would power all five engines. But the first major test — and public validation — came from a quantitative mispricing that JPMorgan failed to detect.

Engine 1: The London Whale — Quantitative Index Arbitrage ($200–300M, 2011–2012)

This wasn’t macro intuition. It was model-detected mispricing.

Detection (November 2011): Saba’s systems flagged the Markit CDX North America Investment Grade Series 9 10-Year Index trading significantly cheaper than fair value. Someone was selling enormous CDS protection, pushing the index below model prices.

That seller was Bruno Iksil at JPMorgan’s Chief Investment Office in London. JPMorgan’s CIO portfolio exploded from $51 billion notional at year-end 2011 to $157 billion by March 31, 2012.

Execution: Weinstein began buying CDS protection on CDX IG9 without initially knowing JPMorgan was the counterparty. He was exploiting measurable mispricing between index spread and constituent single-name CDS.

In February 2012, Weinstein publicly recommended the trade at an investment conference, urging hedge funds to buy protection because the large seller persisted.

Initially, the trade went against Saba — one of Weinstein’s funds was down 20% heading into May. Then Europe’s debt crisis intensified, positions reversed, and Weinstein made back all losses “in a matter of weeks”.

CEO Jamie Dimon dismissed concerns as a “tempest in a teapot” in April 2012. JPMorgan disclosed $2 billion in losses in May, ultimately totaling $6.2 billion, plus $920 million in regulatory fines.

Saba’s profit: $200–300 million. AUM peaked at $5.6 billion after the trade.

The London Whale validated quantitative credit arbitrage at scale. Eight years later, the same model-driven scanning system would detect an even larger mispricing — one driven not by a single trader, but by structural distortions across the entire credit curve.

Engine 2: The Pandemic Trade — Zero-Cost Tail Protection (+99% in March 2020)

Late 2019: Saba’s quantitative systems — which continuously scan approximately 18,000 bonds and 800 different CDS contracts — identified unprecedented mispricing. High-yield credits like cruise lines and airlines traded at similar CDS spreads to investment-grade names like AT&T and IBM.

The Setup: As Weinstein told Risk.net, banks buying CDS protection on companies they lent to (like IBM for its $34 billion Red Hat acquisition) had driven up investment-grade protection costs. Simultaneously, traders arbitraging the HY CDX index fed markets cheap CDS on high-yield names by selling protection on single names.

Saba built long CDS positions on Royal Caribbean, United Airlines, Sabre Corp., and Vue International, while selling protection on AT&T and IBM. Positions were established at “basically zero cost” — structured as carry-neutral relative-value trades.

Critical innovation: Traditional tail hedges (S&P 500 puts, VIX calls) bleed premium daily. Saba’s credit-based tail hedge was carry-neutral with “no negative carry or bleed”.

Execution (January-March 2020): As pandemic news intensified, Saba added $1 billion to its curve-flattening trade and approximately $500 million to its short CDS portfolio.

On February 21, 2020, as Italy reported COVID clusters, five-year Sabre CDS closed at 30 basis points. By March, Sabre traded at 350bps while McDonald’s remained at 25bps.

Saba held “tens of billions of notional CDS” during the crash.

Performance:

• Tail Fund: +99% in March 2020

• Flagship Fund: +33% in March 2020

• Full Year 2020: Flagship +33%, Tail +99%

The crisis attracted $1.7 billion in inflows. Risk.net named Saba “Hedge Fund of the Year” in 2021.

Exit mechanics: Mass selling created negative basis between bonds and CDS. Saba bought crashed cash bonds against short CDS positions to capture convergence. The opportunity was so large Saba launched a dedicated “Basis Opportunities Fund” in May 2020.

The pandemic trade proved tail protection didn’t require premium bleeding. While credit mispricing drove Engines 1 and 2, a parallel opportunity emerged in equity-linked structures trading below intrinsic value — requiring zero credit analysis.

Engine 3: SPAC Arbitrage — “Bonds with an Equity Call Option” ($6.7B Peak, 2021–2022)

SPACs provided asymmetric downside protection with upside optionality.

Mechanics (as explained to Forbes):

1. Buy SPAC shares below $10 trust value (typically $9.70-$9.80)

2. Collect spread — SPAC trusts invest in U.S. Treasuries

3. Downside capped — shareholders can redeem at ~$10 + accrued interest

4. Upside uncapped — popular mergers spike shares above $10

Weinstein called this a “call option on euphoria”. Critically, he used SPAC arbitrage yields to finance the cost of credit protection — creating a self-funding tail hedge.

Scale: By mid-2022, Saba held approximately $6.7 billion in SPACs, among the largest SPAC investors globally.

Notable exit: Saba sold its stake in the SPAC that took Trump’s social media platform public immediately upon target announcement.

By end-2021, Saba was the fourth-biggest SPAC hedge fund investor with $4.26 billion invested.

SPACs exploited structural mispricing in equity-linked vehicles. But another discount-to-value opportunity existed at even larger scale — one that required activist pressure, not passive arbitrage.

Engine 4: Closed-End Fund Activism — Buying “Dollars for 80 Cents” ($3.66B, Ongoing)

Saba’s highest-returning strategy today targets structural discounts in closed-end funds.

The trade: Buy CEFs trading at 10–20% discounts to NAV, pressure boards through activism to close the gap via tender offers, conversions to open-end funds, or liquidations.

As of Q3 2025, Saba held 329 CEF positions worth $3.66 billion.

BlackRock Campaign (2023–2025): Saba waged an 18-month battle against BlackRock. In December 2023, Saba won a court ruling invalidating shareholder voting restrictions.

Settlement announced January 21, 2025:

• Tender offers: BIGZ (50% of shares), BMEZ (40% of shares) at 99.5% of NAV

• BlackRock to repurchase approximately $1.6 billion in shares

• 3-year standstill agreements through 2027 proxy season

UK Investment Trust Campaign (2024–2026):

December 2024: Saba targeted seven UK investment trusts (Baillie Gifford US Growth, CQS Natural Resources, Edinburgh Worldwide, European Smaller Companies, Henderson Opportunities, Herald Investment Trust, Keystone Positive Change).

February 2025: Six of seven trusts rejected Saba’s proposals. Excluding Saba’s votes, 98.4% of Edinburgh Worldwide shareholders voted against.

May 2025: CQS Natural Resources agreed to 100% tender offer at NAV, plus ~8% annual dividend and 20bps fee reduction. 46% of shareholders tendered, including Saba’s full holding.

January 2026: Edinburgh Worldwide shareholders rejected Saba’s second attempt — 92.7% voted against (excluding Saba’s shares).

SpaceX Controversy: Saba accused Edinburgh Worldwide’s manager Baillie Gifford of selling down SpaceX stake in October 2025 at inopportune moment, estimating £37 million in shareholder losses — two months before revaluation increased SpaceX’s worth.

Performance:

• Closed-End Opportunities Fund 1 ($240M): +17.7% (2023), +20.2% (2024)

• CEFS ETF (NAV return): +23.48% (2024), +15.53% (2025)

CEF activism delivered double-digit returns while the flagship struggled. Meanwhile, Saba quietly launched its fifth engine — returning to Weinstein’s original expertise in credit derivatives, but this time as a premium seller, not buyer.

Engine 5: Selling CDS on Big Tech AI Debt (November 2025-Present)

Saba’s newest revenue stream: acting as insurance underwriter for AI-driven debt accumulation.

November 2025: Saba began selling CDS protection on Oracle, Microsoft, Meta, and Amazon to banks worried about AI infrastructure debt. This was the first time Saba had provided such hedging for some of these companies.

Context:

• Meta plans up to $600 billion in AI infrastructure through 2028

• Oracle sold $18 billion in investment-grade bonds in September 2025

• CDS spreads on Oracle and Alphabet reached two-year highs

Weinstein is earning premium by selling protection on companies he judges creditworthy despite explosive debt growth.

Five engines running simultaneously — each designed for different market regimes. Which explains why the flagship struggles in 2025 while CEF activism thrives. The architecture isn’t broken. It’s working exactly as designed.

The Flagship’s Structural Problem — By Design

Annual Performance (~$2 billion flagship fund):

• 2020: +33% (COVID crash)

• 2022: +22% (bond/equity sell-off)

• 2023: -16.9% (credit spreads compressed)

• 2024: +3.4% (spreads remained tight)

• 2025 (to Nov 21): -6.5% (markets rebounded from tariff shock)

$1 billion tail fund:

• 2023: -16.7%

• 2024: -2.7%

• 2025: -12.7%

Market warnings:

April 2023: Warned “There’s going to be a huge problem in private credit,” predicting stress for insurers in private credit and commercial real estate.

April 2025: Warned of credit “avalanche,” predicting tariff-driven inflation would limit the Fed and bankruptcies would spike “much faster than in other crises.”

The flagship’s struggles reflect deliberate positioning for credit dislocations that haven’t materialized at scale since 2022. The same mechanics that produced +33% in 2020 and +22% in 2022 remain primed to trigger when spreads blow out.

This is why Weinstein runs five strategies instead of one. Each engine targets different market conditions. When volatility is suppressed and credit spreads are tight — as in 2025 — Engines 1 and 2 (credit arbitrage and tail hedging) underperform. But Engines 4 and 5 (CEF activism and tech CDS) print money. Here’s the current state of all five:

The Portfolio View: Five Engines Running Simultaneously

As of August 2025, Saba manages $6 billion in regulatory assets. The firm’s 13F shows 329 CEF positions worth $3.66 billion as of Q3 2025.

📊 Want Deeper Quantitative Analysis?

This research took very long time of data collection, verification, and analysis. If you found value in this deep-dive, I publish exclusive quantitative research, trading strategies, and institutional-grade analysis on Patreon.

By joining, you’ll be supporting my work and motivating me to publish more content like this.

→ Join the Patreon community here

The opening paradox resolves itself: a -6.5% flagship fund and “Activist Hedge Fund Manager of the Year” awards aren’t contradictory. They’re evidence of successful architectural design. When one engine struggles, another accelerates.

The chess master who learned the game at five, became a U.S. Chess Federation Life Master at 16, and was banned from casinos for card counting now runs five structurally independent arbitrage engines with $6 billion across markets most traders treat as separate asset classes.

The flagship bleeds by design in low-volatility regimes. The CEF strategy compounds at 20%+ regardless of credit conditions. The tail fund costs money until it doesn’t. The core bet: credit dislocations will arrive before tail hedge costs overwhelm activism profits.

As Weinstein told Investment Week: “I am a hedge fund manager trying to make money for my investors.”

About the Author

Navnoor Bawa is a quantitative finance researcher specializing in systematic trading strategies and hedge fund analysis. He publishes technical deep-dives on institutional strategies and quantitative trading systems.

• 📺 YouTube: The Mathematical Trader — Quantitative finance analysis and trading strategies

• 💼 LinkedIn: Navnoor Bawa — Connect for quant finance insights

• 📊 Patreon: Exclusive Research & Strategies — Support deep-dive research like this

If you found this analysis valuable, consider subscribing to the YouTube channel and connecting on LinkedIn.

Cover photograph: Kidfly182, CC BY 4.0, via Wikimedia Commons.

Deutsche Bank’s decline embodies a brutal paradox: the trades that made the bank billions taught hedge funds how to destroy it. Greg Lippmann generated $1.5 billion shorting subprime FOR Deutsche Bank — while distributing his strategy deck to 50+ external hedge funds. When Boaz Weinstein’s prop desk lost $1.8 billion the following year, those same funds recognized what Lippmann’s success had obscured: Deutsche Bank couldn’t manage risk. By 2016, external players weaponized the bank’s fragility through coordinated attacks — equity shorts, prime brokerage exodus, distressed bond buying — extracting hundreds of millions using the exact playbook Deutsche Bank’s own traders had pioneered.

This analysis documents the complete arc through verified Senate investigations, regulatory filings, and company disclosures: internal trades ($3.5B+ in profits and losses), the catastrophic management failures that exposed vulnerability, and the systematic external exploitation campaigns ($200M+ documented extraction) that followed.

Part I: The Warning Ignored (2005–2008)

The Lippmann Subprime Short: $1.5B–$2B Internal Profit (2007–2008)

Position Architecture

Greg Lippmann, Deutsche Bank’s Global Head of ABS/CDO Trading, built a $5 billion short position in single-name credit default swaps referencing BBB/BBB- mezzanine tranches from 2005–2007. His February 2007 presentation “Shorting Home Equity Mezzanine Tranches” detailed the exact structure:

• Target: BBB/BBB- rated RMBS tranches with 2005–2007 vintage loans

• Instrument: Single-name CDS (buying protection)

• Payout ratio: 10:1 asymmetry — 50–90 points gain vs. 4–8 points annual premium

• Thesis: Rising delinquencies would trigger defaults even without home price declines

Verified Results

Deutsche Bank generated $1.5 billion to nearly $2 billion in 2007 from the position, the bank’s largest single-trade profit at that time. Lippmann personally received $47 million in 2007. The U.S. Senate investigation revealed internal emails where Lippmann called CDOs “crap” and “pigs” while Deutsche Bank simultaneously assembled the $1.1 billion Gemstone 7 CDO.

Execution Insight

Lippmann evangelized the short thesis aggressively, distributing pitch decks to at least 50 hedge funds and brokering CDS trades for external clients while simultaneously building the bank’s proprietary position. The multi-year position required overcoming internal resistance as Deutsche Bank maintained $128 billion in long mortgage exposure.

The Fatal Paradox: Lippmann was a Deutsche Bank employee generating profits FOR the bank — but his pitch deck became the instruction manual for shorting Deutsche Bank itself. Those 50+ hedge funds learned:

• How to identify toxic structured products

• How to use CDS for asymmetric payoffs

• How to position ahead of credit deterioration

• That Deutsche Bank management wouldn’t hedge despite warnings

When Deutsche Bank’s mortgage book imploded six months after Lippmann’s profit, external funds understood: this institution had contrarians who saw disasters coming — and management that ignored them. The target was identified.

The Critical Failure: Lippmann’s $2 billion profit proved his thesis — but Deutsche Bank’s senior management refused to hedge the bank’s $128 billion long mortgage position. While one desk made billions shorting subprime, the rest of the bank remained massively exposed. This institutional schizophrenia — profitable contrarian trader versus overconfident management — would define the next 15 years and enable every subsequent external attack.

Part II: The Vulnerability Exposed (2008)

Weinstein’s Prop Desk: $1.8B Internal Loss

The Setup

Boaz Weinstein established the “Saba” proprietary trading group at Deutsche Bank, scaling to $30 billion in positions and $10 billion in capital by early 2008. As Co-Head of Global Credit Trading, Weinstein deployed capital structure arbitrage and credit default swap strategies.

The Loss

In 2008, Saba lost $1.8 billion, erasing two years of gains. Deutsche Bank reported a $5 billion loss that year — its first in five decades. Weinstein departed in early 2009, taking 15 team members to launch Saba Capital Management with Deutsche Bank as his primary broker.

The Recognition Moment: The $1.8 billion Weinstein loss combined with management’s refusal to act on Lippmann’s warnings revealed a fatal pattern to external observers:

1. Lippmann proved: Contrarian analysis works, Deutsche Bank has structural blind spots

2. Weinstein proved: Risk management is broken, catastrophic losses are possible

3. Management proved: They won’t learn, they won’t adapt, they’ll repeat mistakes

For sophisticated hedge funds who’d received Lippmann’s pitch deck, this was the signal: Deutsche Bank isn’t just vulnerable — it’s systematically exploitable. The question shifted from “Will this bank have problems?” to “How do we position to extract maximum value when it does?”

The eight-year gap between Weinstein’s 2008 loss and the 2016 coordinated attacks wasn’t inaction — it was patient capital positioning, waiting for the catalyst that would let them apply Lippmann’s CDS playbook to Deutsche Bank’s own balance sheet.

Part III: The Playbook Applied — External Extraction (2016)

The Coordinated September 2016 Attack

The 2016 attack executed Lippmann’s playbook against Deutsche Bank itself. Where Lippmann had taught hedge funds to use CDS for asymmetric shorting of structured products, they now applied the same strategy to the bank: identify fundamental weakness (regulatory fines), use derivatives for leverage (equity shorts + CDS), position ahead of catalysts (DOJ announcement), exploit illiquid markets (CoCo bonds).

Three coordinated moves — equity shorts, prime brokerage exodus, and distressed bond buying — occurred within months. Each amplified the others. This wasn’t opportunistic; this was the strategy Deutsche Bank’s own trader had pioneered, turned against the institution.

Catalyst: DOJ $14B Demand (September 15, 2016)

On September 15, 2016, the U.S. Department of Justice demanded $14 billion for mortgage-backed securities fraud. This single announcement triggered a cascade of coordinated external attacks. Deutsche Bank shares fell over 50% in 2016, reaching a 30-year low below €10.

Phase 1: Soros Brexit Short (June 24, 2016)

George Soros had positioned before the DOJ announcement. He established a €100 million short position (7 million shares, 0.51% of capital) on June 24, 2016, the day of the Brexit vote. Deutsche Bank dropped around 14% that day. Soros had called Deutsche Bank “clearly the most vulnerable commercial bank” in July 2016.

The timing was surgical: short European banks on Brexit uncertainty, especially the most leveraged one. Soros simultaneously held long positions in gold as a portfolio hedge.

Phase 2: Prime Brokerage Exodus (September 2016)

Bloomberg reported approximately ten hedge funds withdrew listed derivatives clearing from Deutsche Bank’s prime brokerage in September 2016, including:

• Millennium Management (Israel Englander)

• Rokos Capital Management (Chris Rokos)

• Capula Investment Management (Yan Huo)

This withdrawal created a death spiral: fund exits signal distrust → stock falls → more exits → larger fall. CEO John Cryan’s September 30 memo claimed “no point in the last two decades has the balance sheet been as stable.” Deutsche Bank’s 2016 Annual Report later confirmed: “Prime Finance revenues were impacted by higher funding costs due to widening of our own credit spreads.”

The exodus amplified the equity short — as prime brokers withdraw, it signals sophisticated insiders no longer trust the counterparty, accelerating the stock decline and validating the short thesis.

Phase 3: CoCo Bond Distressed Buying (February-September 2016)

Deutsche Bank’s AT1 contingent convertible bonds crashed to 70% of face value in February 2016, with yields spiking from 6–7.5% to 10–11.5%. Following the DOJ announcement, CoCos dropped further. Deutsche Bank announced a $5.4 billion tender offer (€3B euro bonds + $2B dollar bonds) on February 12, 2016 to stabilize prices, creating opportunities for traders who bought the February lows.

Settlement Resolution

Deutsche Bank finalized a $7.2 billion settlement in January 2017 — $3.1 billion civil penalty plus $4.1 billion consumer relief.

The 2016 Pattern: Three coordinated external attacks — equity short (Soros), prime exodus (Millennium/Rokos/Capula), and distressed debt buying (CoCo traders) — each amplified the others. Total external extraction in 2016 exceeded $200 million in documented profits, with significantly more in unreported gains.

Part IV: Internal Talent Cannot Save Broken Culture (2016–2021)

Spehn’s Zim Distressed Bet: $1B Internal Profit

While external funds coordinated attacks on Deutsche Bank in 2016, one internal trader executed a contrarian distressed debt trade that would generate $1 billion by 2021. Mark Spehn’s success proved the same point Lippmann had demonstrated: Deutsche Bank still employed talented traders capable of billion-dollar wins.

But this success revealed the deeper tragedy: individual trading talent was irrelevant to institutional survival. Lippmann made $2B (2007), management learned nothing. Spehn made $1B (2021), management had already surrendered (2019 capitulation). The pattern was clear — profitable trades couldn’t overcome systematic cultural dysfunction.

Why This Trade Matters: Spehn’s $1B profit demonstrates that Deutsche Bank’s failure wasn’t about lacking trading talent. It was about institutional culture that couldn’t protect or scale success. While Spehn executed brilliant distressed analysis, the bank around him was collapsing from regulatory fines ($10.5B), reputational damage, and prime broker defections. One trader’s win couldn’t offset systemic institutional rot.

Position Architecture

Mark Spehn, a distressed debt trader in Deutsche Bank’s London office, invested less than $100 million starting in 2016 in Zim Integrated Shipping Services bonds, bank loans, and equity following the company’s 2014 debt restructuring.

Investment Thesis

• Industry consolidation reducing overcapacity

• Support from controlling shareholder Idan Ofer

• Zim’s digitization and strategic alliance initiatives

• Maritime transport rates trading at cyclical lows

COVID Catalyst

Pandemic-driven supply chain bottlenecks caused freight rates to surge dramatically — with some routes like Shanghai-Europe jumping from under $1,000 to $7,395 per TEU (640%+ increase) and China-South America spiking 443%. Zim’s January 2021 IPO at $15/share valued the company at $1.75 billion, subsequently nearly tripling in value.

Verified Profit

Deutsche Bank’s position generated approximately $1 billion — roughly 25% of the bank’s 2020 investment banking profits. The trade was mentioned in analyst earnings calls and helped Deutsche Bank return to profitability in 2021. Spehn later sued the bank in 2024 for disability discrimination.

The Lippmann-Spehn Echo: Both Lippmann (2007) and Spehn (2021) proved Deutsche Bank employed world-class traders. Both generated billion-dollar profits through sophisticated analysis. Both were right when management was wrong.

And both successes changed nothing about institutional trajectory.

After Lippmann’s $2B: Management maintained $128B long mortgage exposure, leading to crisis losses.
After Spehn’s $1B: Management had already announced (2019) 18,000 job cuts, equities exit, €74B bad bank.

The message to external hedge funds was crystal clear: internal talent cannot fix institutional culture. No matter how much money individual traders made, the organization would continue making exploitable mistakes. The extraction opportunity remained viable.

Part V: The Final Evolution — Market Structure as Weapon (2023)

The March 2023 CDS Cascade: Beyond the Lippmann Playbook

The 2023 attack represented something Lippmann never imagined: weaponizing market illiquidity itself. No multi-year CDS position needed. No billion-dollar capital commitment. Just €5 million and knowledge of where the system breaks — a single trade that triggered billions in contagion losses.

This was the ultimate evolution of the strategy Lippmann taught: find structural vulnerability, use derivatives for leverage, exploit illiquid markets. But refined to its essence — pure market mechanics, minimal capital, maximum impact.

The Trade

On March 24, 2023, a single €5 million ($5.4 million) trade in Deutsche Bank junior debt CDS triggered global contagion. Regulators identified the transaction as the catalyst for market-wide selling.

Market Impact

• CDS spreads: Surged from 142 bps to over 200 bps on March 23–24 (with reports varying by measurement window and contract type — some sources cite an 85 bp move, others cite the absolute level change from 142→173+ bps)

• Stock: 14% intraday decline, 8.5% close

• European banking index: €30+ billion market value loss

Andrea Enria, ECB Supervisory Board Chair, described single-name CDS as “very opaque, very shallow and very illiquid,” noting “with a few millions, you can move the CDS spreads” of a major bank. ESMA initiated an investigation into CDS market transparency following the volatility.

The Complete Pattern: How Internal Success Enabled External Destruction

The Deutsche Bank collapse follows a causal chain where each successful internal trade paradoxically made the bank MORE vulnerable to external exploitation:

Phase 1: Internal Profit Creates External Knowledge Transfer (2005–2008)

Lippmann generates $2B FOR Deutsche Bank by shorting subprime. Success requires evangelizing to 50+ hedge funds — teaching them:

• How to identify toxic structured products

• How to use CDS for asymmetric leverage

• How to position ahead of credit crises

• That Deutsche Bank management won’t hedge despite warnings

Causal Link: Lippmann’s pitch deck becomes the instruction manual. When his profit proves the thesis while DB’s mortgage book collapses, external funds learn: “This bank has contrarians who are right, and management that ignores them.”

Phase 2: Internal Loss Proves Systemic Risk Management Failure (2008)

Weinstein’s $1.8B loss proves Deutsche Bank cannot manage risk despite Lippmann’s warnings. DB posts first loss in 50 years.

Causal Link: External funds now have complete intelligence — talented internal contrarians (Lippmann) + broken risk management (Weinstein) + management that won’t learn = systematically exploitable target. The 8-year wait (2008–2016) isn’t hesitation; it’s patient positioning for the right catalyst.

Phase 3: External Funds Apply the Playbook to DB Itself (2016)

DOJ $14B demand provides the catalyst. External players execute Lippmann’s strategy against Deutsche Bank:

• Identify fundamental weakness: Regulatory vulnerability (like Lippmann identified subprime)

• Use derivatives for leverage: Equity shorts + CDS (like Lippmann used CDS)

• Position ahead of events: Brexit vote, DOJ announcement (like Lippmann positioned pre-crisis)

• Exploit illiquid markets: CoCo bonds panic (like Lippmann exploited illiquid CDO market)

Soros equity short + Millennium/Rokos prime exodus + CoCo distressed buying = coordinated multi-asset attack using DB’s own playbook.

Causal Link: Three simultaneous attacks amplify each other — prime exodus signals distrust → equity falls → CoCo bonds panic → more prime exits. Each validates the other. External extraction: $200M+ documented.

Phase 4: Internal Talent Cannot Offset Institutional Rot (2016–2021)

Spehn makes $1B proving talent still exists. But management has already capitulated (2019): 18,000 jobs cut, equities exit, €74B bad bank.

Causal Link: Second billion-dollar internal profit changes nothing, proving to external funds that no amount of trading talent can fix the culture. Exploitation opportunity remains permanently viable.

Phase 5: Pure Market Structure Exploitation (2023)

€5M CDS trade triggers billions in contagion. Evolution beyond Lippmann’s capital-intensive strategy — this is weaponized illiquidity. Minimal capital, maximum impact.

Causal Link: After teaching external funds CDS mechanics (Lippmann) and proving management won’t fix culture (Weinstein, Spehn), the final stage requires no large position — just knowledge of where the system breaks.

Common Execution Elements Across All Five Trades

Instrument Selection

CDS dominated equity shorts for three reasons:

1. Better liquidity in stressed periods

2. Defined maximum loss (premium paid)

3. Leveraged exposure without margin calls

Event-Driven Timing

Major catalysts included:

• DOJ settlement announcements (September 2016, January 2017)

• Regulatory fine deadlines (Libor 2015, FX manipulation 2018)

• Prime brokerage relationship changes (visible through SEC filings)

Multi-Asset Convergence

Sophisticated traders pressured multiple instruments simultaneously:

• Equity shorts

• AT1 CoCo bonds

• Senior/junior debt CDS

• Prime brokerage withdrawal (indirect pressure)

Intelligence Gathering

Prime brokerage withdrawals preceded major moves by weeks, providing early-warning signals to connected market participants.

Patience Requirement

• Lippmann: 2-year position building (2005–2007)

• Spehn: 5-year hold (2016–2021)

• Soros: Multi-month staging ahead of Brexit

These holding periods distinguish institutional distress trades from typical short-term speculation.

The Regulatory Backdrop: Why External Attacks Had Ammunition

Deutsche Bank’s vulnerability wasn’t just internal dysfunction — it was systematic regulatory violations that signaled cultural breakdown:

Libor Rigging (2015)
$2.5 billion fine — record at the time. 29 individuals involved, bank pleaded guilty to wire fraud.

Russian Mirror Trades (2017)
$630 million fine for facilitating $10 billion in illicit transfers (2011–2015). NY DFS assessed $425 million, UK FCA £163 million.

FX Manipulation (2018)
$205 million fine for FX trading misconduct (2007–2013). Traders used chat rooms named “Butter the Comedian” and “The Mafia” to coordinate.

Total regulatory fines (2015–2018): $3.3 billion — before the $7.2 billion DOJ settlement. These weren’t one-off failures; they revealed systematic control breakdowns across divisions.

The 2019 Capitulation

On July 7, 2019, Deutsche Bank announced:

• 18,000 job cuts (20% of workforce)

• Complete exit from Equities Sales & Trading

• €74 billion Capital Release Unit (“bad bank”)

• Prime brokerage transfer to BNP Paribas (~$200 billion assets, 900 employees)

The restructuring eliminated the proprietary trading infrastructure that had historically generated both massive profits (Lippmann, Spehn) and catastrophic losses (Weinstein). But closing the prop desk couldn’t fix the cultural rot that enabled systematic fraud.

Complete Trade Summary

Internal trades: +$1.7B net (Lippmann +$1.5–2B, Weinstein -$1.8B, Spehn +$1B)
External extraction: $200M+ documented (Soros €100M+, CoCo traders undisclosed, 2023 CDS unknown)
Regulatory fines: $10.5B+ ($3.3B + $7.2B DOJ settlement)

Conclusion: The Paradox That Destroyed Deutsche Bank

Deutsche Bank’s collapse reveals a brutal paradox: the trades that made the bank billions taught hedge funds how to destroy it.

The Causal Chain

Step 1: Internal Profit as External Education
Lippmann generates $2B FOR Deutsche Bank (2007) while distributing his CDS strategy to 50+ hedge funds. He teaches them how to identify toxic assets, use derivatives for leverage, and position ahead of crises. His profit proves the strategy works.

Step 2: Management Ignores the Lesson
Despite Lippmann’s $2B profit validating his warnings, Deutsche Bank maintains $128B long mortgage exposure. The institution proves it won’t learn from successful contrarians.

Step 3: Internal Loss Confirms Vulnerability
Weinstein’s $1.8B loss (2008) proves risk management is broken. External funds now have complete intelligence: talented contrarians (who are right) + broken risk controls (catastrophic losses) + management that won’t adapt = systematically exploitable target.

Step 4: External Funds Apply the Playbook to DB
2016 attacks execute Lippmann’s strategy against Deutsche Bank itself:

• Identify weakness → Regulatory fines (like Lippmann identified subprime)

• Use derivatives → CDS + equity shorts (Lippmann’s method)

• Exploit illiquidity → CoCo panic (Lippmann’s insight)

• Coordinate timing → DOJ catalyst (Lippmann’s event-driven approach)

External extraction: $200M+ documented, using the exact playbook Deutsche Bank’s own trader pioneered.

Step 5: Second Internal Success Changes Nothing
Spehn makes $1B (2021), proving talent still exists. But management had already capitulated (2019): 18,000 job cuts, equities exit, €74B bad bank. Message to external funds: internal talent cannot fix institutional culture. Exploitation remains viable.

Step 6: Evolution to Pure Market Structure Attack
2023: €5M trade triggers billions in contagion. Beyond even Lippmann’s strategy — this is weaponized market mechanics. The final stage of knowledge transfer: from “how to use CDS” to “how to break markets with minimal capital.”

The Practitioner Insight

For institutional investors: The critical signal isn’t the first large loss — it’s management’s response to the profitable contrarian.

When you observe:

• Internal contrarian makes billions being right ✓

• Management ignores the warning ✓

• Catastrophic loss follows ✓

• Management still doesn’t adapt ✓

The institution is in the “extraction phase.” External capital will weaponize the vulnerability. It’s not a question of if, but when.

The Deutsche Bank Legacy

Internal trades: $3.5B+ in profits and losses (Lippmann +$1.5–2B, Weinstein -$1.8B, Spehn +$1B)
External documented extraction: $200M+ (Soros €100M, CoCo traders undisclosed, 2023 CDS unknown)
Regulatory fines: $10.5B+ (systematic fraud across divisions)
Market cap destruction: €40B+ (from €30B peak to crisis lows)

The trades that Deutsche Bank thought were saving them — Lippmann’s $2B, Spehn’s $1B — were actually creating the blueprint for their destruction. Success without institutional learning is worse than failure, because it generates false confidence while teaching competitors your vulnerabilities.

Final lesson: Trading profits mean nothing if management won’t learn. Once external traders recognize that pattern, the institution becomes a perpetual extraction opportunity — until complete restructuring or collapse.

📊 Want Deeper Quantitative Analysis?

This research took weeks of data collection, verification, and analysis. If you found value in this deep-dive, I publish exclusive quantitative research, trading strategies, and institutional-grade analysis on Patreon.

By joining, you’ll be supporting my work and motivating me to publish more content like this.

→ Join the Patreon community here

Methodology & Sources

All facts, figures, and dates in this article were independently verified through web searches conducted January 30, 2026. Primary sources include: U.S. Senate Permanent Subcommittee on Investigations official reports, Bloomberg terminal data and news articles, Reuters financial coverage, CNBC market reporting, official Deutsche Bank press releases and annual reports, Department of Justice press releases, BNP Paribas announcements, and Zim Integrated Shipping investor relations materials. All inline citations were tested for accessibility at time of publication. Where multiple sources reported varying figures (e.g., CDS spread measurements), discrepancies are noted with attribution to specific vendors. Profit figures are from official investigations or company-disclosed earnings data rather than estimates.

Connect With Me

For more deep-dive analysis on quantitative trading strategies, systematic hedge fund approaches, and institutional market dynamics:

📺 YouTube: The Mathematical Trader — I regularly break down complex trading strategies and analyze how the world’s most successful hedge funds generate alpha.

💼 LinkedIn: Navnoor Bawa — Connect with me for insights on quantitative research, systematic trading, and institutional finance.

Cover photograph: Tilman AB, CC0, via Wikimedia Commons.

Most hedge funds claim “hard catalyst” discipline. Kite Lake Capital Management actually enforces it: 95% of positions require a contractual event with a defined timeline. No soft catalysts. No activist situations. No management change speculation.

The London-based event-driven firm manages approximately $2 billion across its flagship Cayman vehicle and UCITS fund. After closing to new capital in October 2024 at $1.4 billion and reopening to new investments on December 1, 2024, the firm posted 6.9% returns through July 2024 and has delivered positive annual returns every year since inception in February 2011.

How that discipline translates into actual deal flow and portfolio construction starts with the screening process.

The Numbers Behind Deal Selection

Kite Lake screens 500-600 announced M&A deals globally per year. From that universe, the team invests in just 100-120 higher-complexity transactions. At any given time, the portfolio holds roughly 30 positions with an average holding period of 90-100 days for merger arbitrage trades, producing approximately 3x annual portfolio turnover.

The firm only operates in markets with “proper legal framework for contractual M&A”: US, Europe, Scandinavia, Brazil, Hong Kong, Australia, Taiwan, Japan, Thailand, and South Africa. Southern Europe gets selective treatment due to less predictable workout processes. The US credit market? Avoided entirely due to lack of competitive advantage.

Strategy #1: Long Merger Arbitrage

Standard merger arb forms the core capital allocation. On stock-for-stock deals, Kite Lake shorts the acquirer’s stock, sometimes using options for more deterministic risk-reward profiles. Dynamic hedging occurs at both position and portfolio level to neutralize equity beta.

Jamie Sherman, Partner and M&A PM, explained the execution philosophy at the 2018 FERI Conference: “Kite Lake’s aim is to be a speedboat not an oil tanker. As we are small and nimble we want to be in the position to change our mind when the facts change, and be buyers when spreads widen for irrational or non-deal related reasons. We never want to be forced buyers or forced sellers.”

Sherman views merger arbitrage as resilient across market conditions. In bull markets, CEO confidence drives deal flow. During downturns, defensive mergers in distressed sectors like generic pharma create opportunities. Complex cross-border structures like Takeda’s all-stock bid for Shire generate unique spreads. Private equity and activists both initiate and sometimes impede deals, creating mispricing windows.

Strategy #2: Reverse Merger Arbitrage

Unlike typical merger arb (long target), Kite Lake takes short positions on targets expected to fail regulatory or shareholder approval.

Recent examples from 13F filings:

Silicon Motion (SIMO): MaxLinear’s $3.8 billion acquisition terminated July 2023 after China antitrust approval. MaxLinear claimed material adverse effect; Silicon Motion disputed and pursued arbitration. Kite Lake held a 37.18% portfolio weight in this position.

Albertsons (ACI): Kroger’s $24.6 billion merger terminated December 11, 2024 after federal judges in Oregon and Washington blocked the deal on antitrust grounds. Albertsons sued Kroger for billions of dollars in damages plus a $600 million termination fee.

EchoStar/DISH: The merger completed December 31, 2023 with Kite Lake holding 52.88% portfolio weight in satellite positions. The firm’s Q3 2025 13F shows EchoStar Corp as the largest holding with 644,452 shares.

While merger arbitrage (long and reverse) dominates equity capital allocation, the credit book provides uncorrelated returns and portfolio stability during equity market dislocations.

Strategy #3: European Credit Shorting

Geography dictates opportunity set. UK, Netherlands, and Germany get primary focus due to predictable legal and workout processes. Southern Europe requires higher return thresholds given less predictable restructuring frameworks, though France has improved.

Execution mechanics: Use CDS for synthetic short positions targeting covenant defaults, profit warnings, and restructuring triggers. Exit catalyst required within 12 months. The team seldom sits on creditor committees (too time-intensive), instead monitoring workouts via specialized advisors and news services. Sector focus includes idiosyncratic situations in oil and oil services.

The credit strategy proved its value during stress: In 2012’s European sovereign crisis, credit investments generated the majority of flagship returns. In 2016, roughly half of returns came from an average 12% allocation to credit.

Capacity discipline remains strict: $1 billion hard cap on credit exposure to preserve alpha and avoid return dilution. The firm is structuring a longer-duration distressed debt vehicle with five-year lock-up to accommodate multi-year turnaround investments.

Beyond the three core strategies (long merger arb, reverse arb, and credit shorts), Kite Lake maintains a fourth bucket for opportunistic non-M&A event trades—but only when catalyst timing is certain.

Strategy #4: Non-M&A Special Situations

Selective deployment only if catalyst timing matches: minority squeeze-outs (not too illiquid or long-dated), major asset divestitures, complex litigation with legal catalysts, share class arbitrage, and holding company discounts trading below NAV.

Explicitly avoided: Soft catalysts (management changes, strategic reviews, rumors) and activist situations. The firm “does not engage in activist situations per se and does not join activists’ groups”.

These four strategies—each with quantified selection criteria, geographic boundaries, and timeline discipline—reflect a deliberate institutional design. That design originates from the founders’ combined 70+ years in event-driven markets.

The Founders’ Institutional Pedigree

Massi Khadjenouri (Co-Founder, CIO & CRO) began as proprietary trader at Paribas, with the team that was then spun out into Centaurus Capital. From 2005-2011, she established the event desk at Cheyne Capital and served as CIO of Cheyne Special Situations Fund, building it to become one of the largest and best-performing event-driven hedge funds in Europe. Featured in “50 Leading Women in Hedge Funds” and “Tomorrow’s Titans 2012”.

Jan Lernout (Co-Founder, Credit PM) holds graduate degree from Katholieke Universiteit Leuven and MBA from University of Chicago Booth. Started as workout specialist in Goldman Sachs European Special Situations Group, specializing in UK power sector and European cable companies distressed workouts. CFA Charterholder active in European leveraged, stressed, and distressed credit since late 1990s.

Jamie Sherman (Partner, M&A PM) previously Co-Head of European Research at PSAM, started career at UBS in 2000. Manages UCITS fund and merger arbitrage strategies. Presented at 7th FERI Hedge Funds Investment Day (September 2018, Bad Homburg).

The team structure: 17 employees as of mid-2024. Half the team worked together before Kite Lake; 70+ years collective event-driven experience. In 2024, the firm hired Ram Krishnan as Head of Credit Trading & Origination from Goldman Sachs, signaling credit expansion.

This institutional pedigree translates into a deliberately anti-hierarchical structure designed to eliminate the internal competition that derails many multi-strategy funds.

Organizational Structure: No Silos, No Star Culture

Khadjenouri on the firm’s approach: “We don’t believe in silos. They can create problems at both an interpersonal level and an investment level through forced investing and internal competition over book sizes and returns. Our PMs are indifferent to the size of their book.”

Compensation based on overall fund performance, not individual sub-strategy returns. The flagship is managed as a single unified portfolio, not competing fiefdoms.

The unified structure enables centralized risk oversight—a design choice informed directly by witnessing 2008’s liquidity crisis.

Risk Management: 2008 Lessons Applied

Khadjenouri on liquidity discipline: “The lesson we all learnt in 2008 is not to position yourself in a situation where you get trapped as liquidity evaporates and creates forced sellers. I am confident that our fund is unlikely to get in that sort of situation because of our structure and strategies.”

Framework execution:

• Khadjenouri holds both CIO and CRO roles (direct accountability)

• Independent non-investment team led by Rupert Haworth-Booth provides risk governance

• Dynamic hedging at position and portfolio level to neutralize equity beta

• Modest leverage with clear exit catalysts required within 6-12 months

• $1B credit capacity hard cap to preserve alpha

This risk discipline translates directly into the fund’s performance characteristics: low volatility relative to returns, minimal correlation to equity markets, and consistent profitability through market cycles.

Performance Metrics

7-year track record through 2018:

• Correlation to MSCI World: 0.18

• Positive in 21 of 33 MSCI down months

• Annualized return: 9% net

• Volatility: 5.3%

• Sharpe Ratio: 1.7

Specific periods:

• 2017: 10.3% net return

• Q1 2018: 5% net

• 7.5 years through Sep 2018: 77.6% cumulative (5% volatility)

• Q1 2024: 5.5% (leading event-driven peers)

• Through July 2024: 6.9% YTD

These returns come through two distinct fund structures designed for different investor constraints: a Cayman vehicle offering the full strategy set, and a UCITS fund limited to merger arbitrage.

Fund Structure & Capacity Management

Legal vehicles:

• KL Special Opportunities Master Fund (Cayman Islands): Full strategy including merger arb, credit, and distressed

• KL Event Driven UCITS Fund (Ireland): Merger arb only, no credit

• Management Company: Waystone Management Company (IE) Limited

Capacity timeline:

• October 1, 2024: Closed flagship to new investments at $1.4B

• December 1, 2024: Reopened to new investments

Sherman on founders’ philosophy: “The founders, coming from larger organizations, sought to work in a smaller place, running limited assets for both personal and professional reasons.”

This capacity restraint isn’t mere positioning—it’s structural to every alpha source the firm exploits.

The Alpha Sources Synthesized

Deal complexity premium: Complex cross-border or regulatory deals offer wider spreads due to participant uncertainty.

Reverse merger arb: Short positions on deals expected to fail (Silicon Motion, Albertsons examples).

Credit event shorting: CDS positions targeting covenant defaults and profit warnings with 12-month exit catalysts.

Capacity discipline: $1B credit hard cap prevents return dilution; firm-wide AUM management prevents strategy drift.

Geographic selectivity: Only jurisdictions with predictable legal outcomes; no US credit exposure.

Timeline discipline: Hard catalyst required within 6-12 months; no soft catalysts, no activist plays.

Team alignment: Unified book structure eliminates internal competition; compensation tied to overall fund performance.

2008 lessons embedded: Avoid liquidity traps, maintain dynamic hedging, never create forced selling scenarios.

📊 Want Deeper Quantitative Analysis?

This research took very long time of data collection, verification, and analysis. If you found value in this deep-dive, I publish exclusive quantitative research, trading strategies, and institutional-grade analysis on Patreon.

By joining, you’ll be supporting my work and motivating me to publish more content like this.

→ Join the Patreon community here

The operational mechanics matter as much as the thesis. Kite Lake incorporated July 15, 2010 as UK Companies House OC356471. SEC registration: File No. 802-78380, CRD No. 168309.

The firm’s Q3 2025 13F filing showed $93 million in disclosed US equities, representing only a fraction of total assets given the focus on non-US markets and credit instruments not captured in 13F reporting requirements.

The hard catalyst discipline referenced in the opening isn’t marketing language—it’s operationalized through quantified deal screening (500-600 to 100-120 to 30 positions), geographic restrictions (only jurisdictions with predictable legal outcomes), timeline requirements (6-12 month catalysts), capacity caps ($1B credit ceiling), and organizational structure (unified book, no internal competition).

In an industry where “event-driven” often means “we hold 50 positions and hope something happens,” Kite Lake maintains 30 positions with 95% hard catalysts, enforces the $1B credit cap despite $2B AUM, and delivered 14 consecutive profitable years. The execution discipline differentiates performance from marketing claims.

Want more institutional-grade hedge fund research and quantitative trading analysis?

Subscribe to my YouTube channel: The Mathematical Trader

Connect on LinkedIn for daily insights on systematic trading and quantitative finance: Navnoor Bawa

Cover photograph: No Swan So Fine, CC BY-SA 4.0, via Wikimedia Commons.