Michael Burry and the Big Short

Background, Briefly

Burry was born on June 19, 1971, in San Jose, California. At age two, retinoblastoma cost him his left eye; he has worn a prosthetic since childhood. He has discussed, around the time of his Asperger syndrome disclosure, how the prosthesis shaped his social interactions and pushed him toward solitary, analytical work. Treat that as biographical context. It is not a causal explanation. Plenty of people with similar biographies do not end up shorting structured credit.

The formal training pointed at medicine. Undergraduate economics and pre-medicine at UCLA, medical degree at Vanderbilt, neurology residency at Stanford Hospital. During the residency he was running a stock-analysis website on off-hours, applying [[value investing]] reading — Graham, Buffett — to public-equity ideas. The site attracted professional money managers as readers, and several of them ended up seeding his fund. The unusual part of the biography is not the medicine-to-finance pivot; it is that the audience for his off-hours research was institutional before he had a fund to run.

He left clinical work and founded Scion Capital in 2000, with roughly one million dollars from family resources and an inheritance. The fund name is taken from Terry Brooks’s The Scions of Shannara. Scion’s early letters describe a Graham-and-Dodd process: deep reading of primary documents, concentrated long positions in undervalued small- and mid-cap names, selective shorts in businesses he assessed as fundamentally impaired.

Scion’s Pre-Subprime Track Record

From inception in November 2000 through June 2008, Scion returned 489.34 percent net of fees, comfortably ahead of the S&P 500 over the same window. That figure folds in the subprime payoff at the end, which dominates the geometric contribution, but the pre-2005 portion of the track record was already strong on conventional long/short equity. Assets grew from the seed million into the several-hundred-million range over a few years, drawn from high-net-worth individuals, family offices, and a small institutional allocator base.

One quant flag before moving on. Scion ran a high-conviction, low-position-count book. With a portfolio that concentrated, return dispersion across funds with similar nominal style is large, and the realized Sharpe of any single manager carries a wide confidence interval. The 489 percent is real, but it is one realized path. Reading it as a clean indictment of efficient markets overweights the sample. The number says the manager was good, conditional on the path observed. It does not by itself say how good.

The Loan-Tape Research

From 2004 into 2005 Burry redirected research effort onto US residential mortgage-backed securities, specifically the subprime tranches issued during the housing boom. The work was loan-tape work. Not commentary on housing, not macro intuition: actual loan-level data on the mortgages collateralizing specific deals, pulled from prospectuses and trustee reports.

What he found, reduced to its skeleton, was a maturity-and-reset structure that priced as if defaults would stay near baseline but which contained a large concentration of 2/28 and 3/27 adjustable-rate mortgages with introductory teaser rates due to reset in 2007. Reset arithmetic alone implied a payment shock of several hundred dollars per month for borrowers who, in many cases, had qualified only on the teaser-rate payment. To first order, the question was not whether default rates would rise. The question was by how much, and whether the lower-rated tranches of the deals had any cushion against it.

The conclusion, supported by the tapes, was that the cushion was thin or imaginary. The rating-agency models were calibrated on a regime in which house prices did not nationally decline. The mezzanine and BBB tranches of the 2005 and 2006 vintages were structurally short a national housing downturn whether or not they were priced that way. The ABX.HE BBB- subindex, once it began trading in early 2006, traced the same conclusion in a public price series, but Burry had reached it from the underlying tapes a year earlier.

Constructing a Position That Did Not Exist

The thesis is the easy half. The hard half is implementation.

In 2005 there was no liquid public market for shorting individual subprime mortgage bonds. You could short the equity of mortgage originators, but that is a different trade with different risks — primarily, it pays off only if the originator’s equity declines, which can lag the underlying loan performance by quarters or years and is sensitive to refinancing flows you do not control. Equity shorts on originators are a noisy proxy for a subprime credit view. Burry wanted the credit view itself.

The contract he wanted did not exist as a listed product. He approached the major dealers — Goldman Sachs and Deutsche Bank are the two most often named — and persuaded them, over 2005, to write bespoke [[credit default swap]] contracts referencing specific subprime mortgage-backed tranches by CUSIP. Structure: Scion paid periodic insurance-style premiums; in exchange, the dealer paid a contractually defined amount if the referenced bonds suffered defined credit events (write-downs, principal shortfalls). The premia ran in the low hundreds of basis points per year on notional, depending on the tranche.

Mechanically, this is a structured insurance contract. The premium is the carry cost. The payout is the contingent payoff. The position has negative theta until the credit event, and convex positive payoff after. There is a programming-side analogy worth flagging: Burry was not buying a product off a shelf. He was, in effect, specifying an interface and persuading a counterparty to implement it — defining the reference obligation, the credit-event triggers, the settlement mechanics. That kind of contract design is closer to API negotiation than to order entry, and the dealers ultimately had structural incentives to write it (warehouse inventory they wanted hedged, fee revenue, eventual collateralization into ABS CDOs).

Carry, Marks, and the LP Conflict

Mathematically the trade was correct on a horizon of two to three years. Operationally it was almost fatal because of the carry.

For roughly two years, from late 2005 into 2007, the referenced subprime tranches kept performing at par on a [[mark-to-market]] basis. The dealers’ marks on the CDS positions moved against Scion, sometimes substantially, even as the underlying loans were starting to deteriorate but had not yet hit defined credit events. Premium payments were a steady drag on reported NAV. Limited partners saw a fund that had compounded at high rates losing money on a position they did not understand and which the manager refused to unwind.

The redemption pressure was severe. Several investors demanded their money back. Some threatened legal action. Burry invoked side-pocket provisions in the fund’s governing documents, designating the CDS book as illiquid and restricting withdrawals against it. This is the legally clean response and, given the position’s eventual payoff, the correct one. It also destroys most of the goodwill between a hedge fund manager and his investors. Both things are true.

This is the most underdiscussed part of the trade. The popular reading treats Burry’s conviction as the difficult part. From the manager’s seat, the conviction was the easy part — he had read the tapes. The hard part was the institutional structure that had to hold up while the position carried negatively against rising client anger. Plenty of people with the same loan-tape conclusion would have unwound at the bottom of the mark-to-market hole. The Scion subprime position is also, less glamorously, a structural argument for side-pocket provisions in fund documents. Few writeups frame it that way.

Outcome

The thesis was validated as subprime defaults accelerated through 2007 and the broader credit crisis worked through structured-credit balance sheets in 2008. The CDS positions appreciated as referenced tranches were written down. The trade returned approximately 489 percent on the capital allocated to it, generated approximately one hundred million dollars in personal compensation for Burry, and produced approximately seven hundred million dollars in profits for outside investors.

Following the resolution, Burry closed Scion Capital in 2008 and returned outside capital. His stated reasons combined the strain of the 2005–2007 investor conflict, the regulatory environment for hedge fund managers, and a preference for managing primarily his own capital. The closure landed near the trough of the 2008 [[bear market]].

Media Treatment, and One Caveat on the Asperger Disclosure

The trade was documented by Michael Lewis in The Big Short: Inside the Doomsday Machine (2010), which traced several investors who shorted subprime via credit default swaps. The 2015 film adaptation, directed by Adam McKay, portrayed Burry as a character played by Christian Bale; the performance was nominated for an Academy Award for Best Supporting Actor.

Burry publicly disclosed an Asperger syndrome diagnosis around the time of the Lewis book’s publication. He has cited the diagnosis as relevant to his preference for working in isolation and to his tolerance for staying with positions through long stretches of social and mark-to-market pressure. The disclosure is part of the public record. I would resist any reading that uses it to make the trade unrepeatable or supernatural. The loan tapes were available to anyone who would read them. The thing that was scarce in 2005 was not cognition. It was the willingness to do the document-level work and the institutional setup to hold the position.

Scion Asset Management and the 13F Era

In 2013 Burry registered a new vehicle, Scion Asset Management, based in Saratoga, California. The firm files quarterly 13F reports with the SEC disclosing US long-equity positions, and the filings get parsed in detail by financial media and retail investors. The disclosed book has rotated through small-cap equities, distressed names, commodity-linked stocks, and water- and farmland-related investments.

Scion Asset Management has also taken visible short positions through listed [[options]] contracts. Public filings and Burry’s social-media commentary have indicated bearish positions against Tesla, Inc., against the ETFs run by Cathie Wood’s ARK Investment Management, and against the broad US equity market via put options on the SPDR S&P 500 ETF Trust (SPY) and the Invesco QQQ Trust (QQQ).

One technical note that gets repeatedly missed in the coverage. [[13F filing]] disclosures report options at notional value, not at premium paid. A reported notional of one billion dollars in SPY puts is consistent with a premium outlay of roughly twenty to forty million dollars depending on strike, expiry, and underlying σ. The two numbers differ by an order of magnitude. The capital actually at risk on a long-put position is bounded by the premium paid; the notional is a reference quantity for the size of the directional bet, not for the size of the loss if the trade goes wrong. Most of the headline coverage of “Burry’s billion-dollar short” has confused these two. The error is structural and worth flagging because it recurs in 13F coverage every quarter.

Investment Philosophy

Burry’s described approach combines deep fundamental research, contrarian positioning, and willingness to sit through long periods of adverse mark-to-market. His writings emphasize primary documents — loan tapes, prospectuses, regulatory filings, financial statements — over sell-side research or rating-agency outputs. He has cited Graham’s margin-of-safety concept as foundational and has applied the same valuation discipline to shorts, looking for situations where the gap between price and assessed intrinsic value is large and where a catalyst for resolution is identifiable on a defined horizon.

A useful comparison from a different domain. In Go (igo), the late-game phase known as yose is the conversion of accumulated positional influence into actual points on the board. A player can have a clearly winning middle-game shape and still lose if the endgame is mistimed or mishandled — if territory that exists in potential is never converted before the game ends. The Scion subprime trade reads in those terms. The thesis was the middle-game advantage; the loan tapes had already established it. The endgame was the two-year carry, the side-pocket clause, the dealer relationships, the cash buffer. That is where the points were actually counted. Most retail attempts at structurally similar trades fail not on thesis but on yose: on the conversion.

Since Scion Capital’s closure, Burry’s public output has shifted toward macro commentary — on the [[Federal Reserve]], on speculative excess in equities and cryptocurrencies, on demographic and inflationary risks — published in brief social-media posts, with periodic account deletions and reactivations. The macro calls are a different exercise from the loan-tape work. They are not loan-tape work in macro clothing. That distinction has been blurred in the popular reception, partly to his own benefit, partly not.

What the Trade Is a Lesson In

The popular reading treats the Burry subprime trade as a story about an outsider who saw what insiders missed. That reading is not wrong; it is shallow. The trade is more usefully read as a case study in three engineering-style problems. Instrument construction: negotiating a non-existent contract into existence with a counterparty whose incentives had to be aligned with writing it. Carry management: sizing a position whose premium drag is bounded but real, against a fund structure with redemption pressure on the same horizon. Joint-tail conditioning: pricing in the assumption that defaults across geographies were positively correlated in a national downturn, an assumption the rating models did not carry.

Each of those is teachable. None requires the prosthetic eye, the Asperger diagnosis, or the off-hours blog. They require reading the documents, doing the carry math, and structuring the fund vehicle so that you can hold the position long enough for the convex payoff to arrive. That is the transferable content. Modulo the assumption that LP redemption pressure can be legally contained — which depends on fund documents that not every manager has — the trade is reproducible in mechanism if not in scale.

That is a less romantic conclusion than most coverage offers. It is also the useful one.