Archegos: How Bill Hwang's Family Office Cost Wall Street $10bn (2021)

At 9:30 on the morning of Friday, 26 March 2021, Goldman Sachs sent its institutional clients a block-trade offering for around $10.5bn of stock. ViacomCBS, Discovery, Baidu, Tencent Music, Vipshop and GSX Techedu all featured on the list, sold in unusually large size and at discounts of 5–20% to the previous close. The sellers were not named. By lunchtime in London, traders were calling it "the Archegos liquidation" because Morgan Stanley had begun its own block sale the evening before, and by Monday's close Credit Suisse had admitted to a loss that would eventually reach roughly $5.5bn. The client behind the unwind was a private family office at 888 Seventh Avenue that had filed no equity disclosures because it was not required to file any.

The firm was Archegos Capital Management, run by Sung Kook "Bill" Hwang, a former Tiger Management analyst who had spent two decades as a Tiger Cub. In a single business week, Hwang's concentrated bets detonated a chain of prime-brokerage losses that the Financial Stability Board later tallied at more than $10bn across six banks, and that regulators across three continents used as the template for redesigning how they supervised shadow leverage. The collapse had no policy trigger, no central-bank signal, no macro shock — only a series of margin calls into a thin tape.

From Tiger Cub to family office

Hwang arrived in the United States from Seoul as a teenager, worked as a Hyundai Securities salesman, and in 1996 joined Julian Robertson's Tiger Management as an analyst. When Tiger wound down in 2000, Robertson seeded a cohort of younger managers — the Tiger Cubs — and Hwang set up Tiger Asia Management with roughly $25m of backing. By 2007 Tiger Asia was running close to $8bn, focused on Japan, Korea, and China long-short equity. The business ended in December 2012, when Hwang pleaded guilty on behalf of Tiger Asia to wire fraud and insider trading in Chinese bank stocks, paid $44m to settle parallel SEC charges, and agreed to return external capital and exit the advisory business for five years (DOJ, 2012).

The settlement left Hwang with roughly $500m of his own money and a ban on managing outside capital. In 2013 he incorporated Archegos Capital Management as a family office. The structure mattered: under the Dodd-Frank Family Office Rule (Rule 202(a)(11)(G)-1), a firm managing only the wealth of a single family and its key employees is exempt from registration as an investment adviser, from Form PF reporting to the SEC, and from most Rule 13F disclosure on long positions. Hwang could rebuild a concentrated equity book without being visible to the market he was operating in.

He rebuilt fast. Archegos grew from $1.5bn of equity in 2018 to somewhere between $35bn and $36bn of net asset value by mid-March 2021, an implied compound return of more than 100% per year (Kelly, 2023). Nearly all of that return came from a handful of concentrated single-name positions held on leverage through total return swaps.

The architecture of hidden leverage

A total return swap is a deceptively simple instrument. The client — here Archegos — agrees to receive from the bank any price appreciation plus dividends on a reference stock, and to pay the bank a financing rate (typically a LIBOR-plus spread while LIBOR still existed) and any price declines. The bank hedges by buying the stock in the cash market, which parks it on the bank's balance sheet rather than the client's. From the outside, the bank looks like a large long holder; Archegos, on paper, holds nothing.

For Archegos this structure offered three things at once. It gave leverage far beyond what Regulation T would permit in a margin account — roughly 5× to 8× gross exposure against posted equity, by the Paul Weiss tally (Credit Suisse Group Special Committee, 2021). It kept the positions off the 13F filings that a registered fund would have to make each quarter, so competitors and targets never saw them build. And it let Hwang spread the same playbook across multiple counterparties without any of them seeing the whole.

The Credit Suisse post-mortem reconstructed the client's book by the Friday before the crash:

Figures drawn from Credit Suisse Group Special Committee (2021), Nomura's May 2021 interim report, Morgan Stanley 1Q 2021 earnings commentary, UBS 2021 annual review, and Financial Stability Board (2023).

By the time the liquidation began, Archegos held economic exposure to about 10% of the float in ViacomCBS, a similar share of Discovery Class A, meaningful chunks of Farfetch, GSX Techedu, Vipshop, Baidu, Tencent Music, and iQiyi, and smaller positions in Rocket Companies and Shopify. A Credit Suisse risk officer quoted in the Paul Weiss report called the Archegos account "the single largest client exposure in the Prime Services division globally," adding that "scenario losses at the portfolio level had been running above limits since late 2020" (Credit Suisse Group Special Committee, 2021).

How supervision failed at each bank

Each prime broker had in principle the same information about its own relationship with Archegos: concentration reports, scenario loss estimates, initial-margin shortfalls, and a counterparty risk rating. None had visibility into what the other banks were doing. The Credit Suisse report concluded that the bank had ignored its own dynamic margining framework and charged Archegos a flat 7.5% initial margin on its swap book — a figure "materially below" what the framework would have required once concentration rose, and a figure the bank left untouched even as exposures doubled through early 2021. "Dynamic margining would have required approximately $4.7bn of additional initial margin by the beginning of March 2021," the report noted, "but the framework was never implemented for Archegos" (Credit Suisse Group Special Committee, 2021).

Nomura's own review found a comparable story — concentrated exposure, margin terms that drifted below the bank's policy, and internal risk escalations that stalled. At Morgan Stanley and Goldman, the risk architecture behaved similarly but the execution diverged: both houses had more recent experience of unwinding forced blocks (the 2014 Caesars, the 2019 Anaplan IPO secondaries) and moved decisively once the margin call missed. That difference between first-mover and last-mover on the 25th and 26th is what separates near-zero loss from $5.5bn.

Kelly (2023) describes a conference call on Thursday, 25 March 2021, in which representatives of Credit Suisse, Nomura, Morgan Stanley, UBS and Goldman attempted to agree on a joint, orderly unwind of Archegos's book. The call broke down within hours. Goldman and Morgan Stanley chose to protect themselves by selling first; Credit Suisse and Nomura waited, partly out of a stated preference for orderly handling and partly because their systems could not price the full book quickly enough.

The ViacomCBS trigger

The position that lit the match was ViacomCBS. Archegos had accumulated roughly 50 million shares of economic exposure in the name through 2020 and into early 2021, helping drive the stock from the mid-$30s in November 2020 to $100.34 on 22 March 2021. On that morning, ViacomCBS announced a $3bn secondary offering priced at $85 per share, below the prior close. The stock fell through the offering price immediately and kept falling. Discovery, another Archegos flagship, moved with it.

Over the five trading days from 22 to 29 March 2021, ViacomCBS lost roughly 55% of its value, Discovery lost close to half, and the Chinese ADRs held by Archegos — Baidu, Vipshop, Tencent Music, iQiyi, GSX — fell between 20% and 50%. The concentrated basket was moving together because one seller was liquidating all of it simultaneously. Every dollar of drop triggered a further margin call on Archegos's swap book across every bank at once.

On Wednesday 24 March Archegos failed to meet a margin call at Credit Suisse and Nomura. On Thursday 25 March Goldman Sachs and Morgan Stanley sold blocks after the close. On Friday 26 March Goldman launched the formal $10.5bn block offering that made the unwind public. Over the following two weeks roughly $20bn of aggregate stock changed hands in liquidations traceable to the Archegos book.

The butcher's bill

Credit Suisse absorbed the heaviest blow. The bank reported a CHF 4.4bn loss in Q1 2021 and a final figure of around $5.5bn once all unwinds had cleared. The loss arrived while the bank was still processing the Greensill supply-chain fund collapse of early March. Together, the two events cost Credit Suisse its CEO, its investment-bank chief, its chief risk officer, its prime-services head, and several layers below — and set off a two-year slide in confidence that concluded with the March 2023 emergency sale to UBS as part of a broader retelling of how systemically important banks can die by reputation rather than capital.

Nomura acknowledged a $2.9bn loss and scaled back its US prime brokerage ambitions. Morgan Stanley's $911m hit was modest against its group earnings but triggered internal changes to how its prime services division sized single-counterparty risk. UBS reported $774m. Mitsubishi UFJ's securities arm took roughly $300m. Goldman Sachs and Wells Fargo took nil to negligible losses — both had moved the moment the margin call missed.

The regulatory response followed two tracks. In April 2022 the US Attorney for the Southern District of New York indicted Hwang and his CFO Patrick Halligan on eleven counts including racketeering conspiracy, securities fraud, and market manipulation. The indictment alleged that Archegos had misled its prime brokers about the size and concentration of its positions in order to extract ever-greater leverage (DOJ, 2022). In July 2024 a Manhattan federal jury convicted Hwang on ten of those counts. On 20 November 2024 Judge Alvin Hellerstein sentenced him to 18 years in federal prison — longer than the 14 years initially handed to Barings trader Nick Leeson's rough historical analogue in the 1995 Barings collapse, though against a much larger dollar scale. A later resentencing reduced the term slightly on pension-related counts.

In parallel, the Financial Stability Board published a cross-jurisdictional review drawing three structural lessons (FSB, 2023). First, prime brokers had relied on bilateral margin terms that failed to respond to position concentration; dynamic initial margin became a supervisory expectation rather than an option. Second, non-bank financial intermediation — particularly family offices using total return swaps — produced leverage that was invisible in aggregate, and the SEC needed to close the reporting gap. Third, banks' own governance had permitted commercial incentives to outrun risk escalation, a finding that echoed the post-mortems on Long-Term Capital Management in 1998.

The SEC moved on the disclosure gap with new Rule 10B-1, proposed in December 2021 and finalised in 2023, which requires prompt public reporting of large security-based swap positions above specified thresholds. Family offices that use total return swaps at Archegos scale now have to surface the same positions that Hwang kept hidden for three years.

What Archegos actually was

Almost everything about the collapse was ordinary. Total return swaps were not new; the family-office exemption was not new; the concentration was extreme but not in any technical sense novel. What was different was the combination. Six independent banks, each applying its own risk framework, each believing the client held a reasonable book on a reasonable amount of collateral, collectively underwrote a position that would have been impossible for any one of them alone. The industry had built a structure in which a single investor could be the largest holder of several mid-cap listed companies without ever appearing in any register or filing — and when that investor stopped being able to meet margin calls, the unwind was always going to be a race, not a negotiation.

A former Goldman risk officer told Kelly (2023) that the lesson of Archegos was not that leverage was dangerous — the industry knew that — but that "you can't manage what you can't see, and we'd built a market where you couldn't see." A year later, one of the six banks was gone. Three years later Bill Hwang was in federal prison. ViacomCBS, renamed Paramount Global, traded below $20 a share by early 2023 and was eventually absorbed into a merger that made its independent life a parenthesis in a longer streaming consolidation.

The tape on Friday, 26 March 2021, printed the largest block trades in Wall Street history to that point. Not a single prime broker had seen them coming.