TL;DR: Archegos used total return swaps to hide a $36 billion bet from its own lenders.
When it broke in March 2021, banks lost more than $10 billion in days.
The collapse of Archegos Capital Management is usually told as a story about greed. It is more useful to tell it as a story about plumbing. In five trading days in March 2021, a family office almost nobody on Wall Street could name erased more than $100 billion in market value from a handful of stocks and inflicted some of the largest counterparty losses since 2008 — without technically owning most of the shares involved. Understanding exactly how that was possible is the entire lesson.
A $36 Billion Portfolio That Owned Almost Nothing
Bill Hwang, a former Tiger Management protégé, ran Archegos as a family office managing his personal fortune. Instead of buying stocks outright, Archegos built most of its book through total return swaps with its prime brokers. The mechanics are simple once stripped of jargon. Archegos told a bank, in effect: “Take a $1 billion position in ViacomCBS on your own balance sheet. Whatever it earns — price gains plus dividends — you pay to me; whatever it loses, I pay you, plus a financing fee.” The bank hedged by actually buying the shares, so the shares sat in the bank’s name. Archegos held only a contract.
That contract delivered the full economic result of ownership for a fraction of the capital, because the bank required only a margin deposit — often 15 to 25 percent of the position’s value. Repeated across multiple banks, this is how, per the SEC’s 2022 complaint, Archegos turned roughly $1.5 billion of capital in March 2020 into a portfolio worth more than $36 billion a year later, with gross market exposure that peaked around $160 billion. The book was ferociously concentrated: ViacomCBS, Discovery, GSX Techedu, Baidu, Vipshop, Tencent Music and a few other names made up nearly all of it. Prosecutors later alleged Hwang’s combined positions gave him economic exposure to more than half of ViacomCBS’s freely trading shares.
Why No 13F Ever Showed Hwang’s Positions
Here is what made the concentration invisible. U.S. securities law requires large investment managers to disclose equity holdings quarterly on Form 13F, and anyone crossing 5 percent beneficial ownership of a company to file a 13D or 13G. Both regimes, as they stood in 2021, keyed off ownership of the shares themselves. Archegos owned swaps, not shares; the shares belonged to its banks, each holding a stake small enough to attract no attention. And as a family office managing only Hwang’s own money, Archegos was not a registered investment adviser and faced none of the reporting that hedge funds with outside clients accept.
The result was a genuine blind spot. A single investor had built enormous exposure to a short list of mid-cap stocks — in some names amounting to a controlling economic interest — and no public filing anywhere recorded it. Investors bidding those stocks up in early 2021 had no way to know that much of the demand traced back to one buyer, whose purchases were themselves pushing the prices that justified more purchases.
Seven Prime Brokers, Each Seeing One Slice
The same opacity applied inside the banks. Archegos ran essentially the same trade — long swaps on the same concentrated names — through Morgan Stanley, Goldman Sachs, Credit Suisse, Nomura, UBS, Deutsche Bank, Mitsubishi UFJ and others. Swap positions are bilateral and confidential, so each prime broker saw only its own slice and could satisfy itself that its exposure, taken alone, was manageable. None could see that the client had mirrored the position six or seven times over — so none could price the true risk: if Archegos ever had to unwind, every bank would be dumping the same illiquid stocks into the same market at the same moment.
According to the DOJ’s April 2022 indictment, Archegos did not leave this ambiguity to chance — its officers allegedly lied outright when banks asked, understating concentration and overstating liquidity to win more swap capacity. The margin mechanics quietly made things worse. Credit Suisse, for instance, margined many Archegos swaps statically: the rate was fixed when the trade was struck and never reset as the position ballooned. As the stocks rose, margin as a share of the position’s value eroded, so the bank’s protection was thinnest exactly when the position was largest. Rising prices also generated paper gains Archegos could withdraw or recycle into new swaps — leverage compounding on leverage, with each bank watching only its own dial.
Five Days in March: From Share Sale to Fire Sale
The unwind needed only an ordinary corporate event. On March 22, 2021, ViacomCBS — whose stock had roughly tripled in months, lifted in part by Archegos’s own buying — announced a $3 billion share offering. The deal priced poorly, the stock slid, and the slide cascaded through a portfolio long the same correlated names everywhere. By March 24, Archegos faced margin calls it could not meet at multiple banks simultaneously. A hastily convened call among the prime brokers on March 25 produced no standstill agreement. Starting Friday, March 26, banks seized collateral and raced one another to liquidate, moving more than $20 billion in block trades in ViacomCBS, Discovery, Baidu, Tencent Music, Vipshop and related names. ViacomCBS and Discovery each fell more than 27 percent in a single session.
Speed determined survival. Goldman Sachs and Morgan Stanley sold first and escaped with minimal or modest damage (Morgan Stanley later disclosed a $911 million hit). Deutsche Bank de-risked quickly. The slower movers absorbed the losses:
- Credit Suisse: roughly $5.5 billion, the largest trading loss in the bank’s history and a major step on its road to the 2023 UBS rescue
- Nomura: ultimately about $2.9 billion, after an initial estimate near $2 billion
- UBS: roughly $860 million, disclosed with its first-quarter results
- Archegos itself: essentially the entire ~$36 billion portfolio, gone in about a week
Inside Credit Suisse: Warnings That Went Nowhere
The most detailed post-mortem of any counterparty is the report Credit Suisse’s board commissioned from the law firm Paul, Weiss, published in July 2021 and filed with the SEC. Its verdict was blunt: the losses reflected “a fundamental failure of management and controls” in the investment bank’s prime services business. Notably, the report found the risks were not hidden from the bank — they were visible, documented and discussed, and still nothing happened.
The particulars read like a checklist of governance decay. Archegos breached its potential-exposure limit at Credit Suisse persistently from 2020 onward — at one point by a factor of more than 25 — and the response was to raise limits or grant exceptions rather than cut exposure. By March 2021 the bank carried around $21 billion in gross Archegos positions against margin that erosion had left dangerously thin. Efforts to move Archegos onto dynamic margining crawled through negotiations for months because staff feared losing the business; the second line of defense deferred to the first; senior risk committees were either not told or did not press. The risk was never mispriced by a model. It was waved through by people.
From Indictment to an 18-Year Sentence
The legal reckoning was unusually severe for a trading blow-up. In April 2022, federal prosecutors in Manhattan charged Hwang and CFO Patrick Halligan with racketeering conspiracy, securities fraud, market manipulation and wire fraud; two other executives pleaded guilty and cooperated. The government’s theory was that the swaps were not merely a leverage tool but a manipulation device — buying to drive prices, funded by capacity obtained through lies to the banks. In July 2024, after a two-month trial, a jury convicted Hwang on 10 of 11 counts and Halligan on all three counts he faced. In November 2024, Hwang was sentenced to 18 years in prison, with a restitution order exceeding $9 billion. It stands among the heaviest sentences ever imposed for conduct centered on market manipulation.
The Risk That Lived Between the Banks
Regulators treated Archegos as a syllabus. The SEC moved to expand swap-position reporting; supervisors on both sides of the Atlantic reviewed prime brokerage practices; and the Basel Committee’s final guidelines on counterparty credit risk management, published in December 2024, are shot through with Archegos’s fingerprints — demands for real due diligence on opaque clients, complementary exposure metrics rather than a single number, and margin that responds to concentration and liquidity, not just volatility.
But the case supports one insight sharper than “improve risk management.” Every bank facing Archegos could run an individually defensible model and still be collectively blind, because the decisive variable — how many other lenders were financing the identical trade — appeared in no bank’s data. Each margin calculation implicitly assumed an orderly liquidation into a normal market; the existence of six other prime brokers guaranteed the liquidation would be anything but. The fatal exposure was not Archegos’s leverage at any one bank. It was the correlation among the lenders themselves, a risk that lived in the space between institutions, where no single institution’s model looks. That is why the fix cannot be purely internal: only disclosure regimes and information that cross firm boundaries can surface a position that is safe seven times individually and lethal once in aggregate.
Source note: Figures and findings are drawn from SEC and DOJ charging documents and press releases, the Credit Suisse Special Committee (Paul, Weiss) report as filed with the SEC, contemporaneous CNBC reporting, and Basel Committee publications; bank loss figures evolved over time and are stated approximately.