Research Memo
What Michael Burry’s “$1.6 billion short” actually was, and why the same headline keeps coming back
In August 2023, Scion Asset Management’s quarterly 13F filing produced a wave of coverage reporting that Michael Burry had made a $1.6 billion bet against the stock market. The figure was widely repeated as representing something like 93% of his portfolio. Clients noticed, and several asked us what it meant.
The answer was that the filing did not say what the headlines said it said. The position was real, the conviction was real, and the size was a small fraction of what was reported. The gap between those two things is entirely a function of how the SEC requires options to be disclosed, and it is worth understanding, because the same misreading has recurred with the same manager since.
Scion’s second-quarter filing, reporting holdings as of June 30 and released in mid-August, listed put options against the SPDR S&P 500 ETF at approximately $886 million and put options against the Invesco QQQ Trust at approximately $739 million. Together, roughly $1.6 billion.
The firm’s actual assets under management were roughly $230 million.
That discrepancy should have stopped the story before it started. A fund cannot deploy $1.6 billion it does not have. But the filing’s own reported market value had jumped from about $110 million in the first quarter to roughly $1.7 billion in the second, and taking that number at face value is exactly what produced the 93% figure.
The filing lists the SPY position at approximately $886 million against about 2 million shares. Divide one by the other and you get roughly $443 per share, which is almost exactly where SPY was trading when the filing was released.
That is the tell. The filing is reporting the market value of the underlying shares, not what Scion paid.
The share count requires a second adjustment. Option positions are reported in share-equivalents, and a standard listed option contract represents 100 shares. Two million share-equivalents is therefore 20,000 contracts, not two million. The same holds for the QQQ line: 20,000 contracts, 40,000 across both positions.
What Scion actually paid is the premium, which the filing does not disclose. Premium depends on strike, expiry, and implied volatility at the time of purchase, and for out-of-the-money index puts it is a small percentage of notional. Reasonable estimates at the time placed the outlay in the tens of millions against a $230 million fund.
So: a real position, a significant one, and a fraction of the reported figure. Forty thousand contracts is a large trade by any measure. It is not 93% of the portfolio, and it was never capable of being.
Form 13F requires institutional managers to report the value of the securities underlying an option position rather than the premium paid. That is the rule, applied consistently, and it is not an error on the filer’s part.
It also is not a meaningless convention. Notional value describes exposure, the quantity of underlying the position controls, which is the figure a regulator monitoring concentration and systemic exposure would want. For a long stock position, notional and cost are close enough that the distinction rarely matters. For options, they diverge enormously, and the direction of the divergence is always the same: notional overstates outlay, often by one or two orders of magnitude.
The result is a disclosure that is accurate, complete, and routinely misread.
If the 2023 episode were isolated, it would be a curiosity. It is not.
In 2025, Scion’s filings again showed large put positions, roughly $912 million in notional exposure against Palantir and $187 million against Nvidia, together making up close to 80% of the fund’s reported value. The coverage followed the same pattern, describing an enormous bearish bet on the artificial intelligence trade.
This time the manager responded directly. Burry stated publicly that he had spent $9.2 million on the options, against the $912 million figure being reported.
A 99% overstatement, confirmed by the person who placed the trade.
That confirmation is what elevates this from a reading error to a structural problem. The same filing convention produced the same misreading about the same manager two years apart, and the correction only arrived because he chose to volunteer it. In most cases no such correction is available, because the premium simply is not disclosed anywhere.
Shortly afterward, Burry deregistered Scion and stopped filing 13Fs altogether.
The options issue is the most dramatic, but it is not the only reason these filings should be read with care.
A 13F reports holdings as of the last day of the quarter. It says nothing about what was held on any other day. A manager could establish a position days before quarter end and exit it days after, and the filing would report it identically to a position held for years.
The 2023 filing described holdings as of June 30 and became public in mid-August. Six weeks is a long time in an options position with a defined expiry. By the time anyone read it, the position may have been sold, rolled, resized, or expired. The filing is a photograph of a moment that has already passed.
13F reporting covers long positions in listed equities and certain derivatives. Short stock positions are not reported. A filing therefore cannot be read as a complete picture of a manager’s market exposure, and a portfolio that appears directionally long may not be.
A put position may be a directional bet, a hedge against a long book, a financing leg of a spread, or one side of a more complex structure. The filing does not distinguish between them. In the 2023 case, Scion had simultaneously increased consumer equity holdings, which is at least consistent with hedging rather than an outright directional view.
None of this makes 13Fs useless. They are among the few windows into institutional positioning available to the public, and read carefully they are informative.
But “read carefully” means three specific habits. Check option lines against the underlying’s price to see whether the figure is notional. Divide reported share counts by 100 to recover contract counts. And treat any percentage-of-portfolio calculation that includes an options line as unreliable until the notional has been stripped out.
The broader point is one that applies well beyond this filing. A disclosure can be entirely accurate and still support a conclusion that is entirely wrong, when the reader assumes a convention the filer was not using. The work is not in finding hidden information. It is in understanding what the visible information actually measures.
Sources: Scion Asset Management Form 13F-HR filings with the SEC; public statements by Michael Burry; contemporaneous coverage from August 2023 and November 2025. Written for general information and not as investment advice.
CFA® and Chartered Financial Analyst® are registered trademarks owned by CFA Institute.