Archegos, 2021Medium

One family office, five prime brokers, and a position none of them could see in full — until the race to sell began.

3 min read · 490 words · Updated 4 Sep 2026

What happened

• 2020–early 2021 — Archegos, the family office of Bill Hwang, builds enormous concentrated positions in a handful of media and Chinese technology stocks.
• The structure — most exposure is held through total return swaps with several prime brokers rather than as shares. The bank owns the stock; the client owns the economics.
• Each broker sees only its own book. None knows the aggregate size, and swap positions do not trigger the disclosure thresholds that direct shareholdings would.
• Late March 2021 — a large secondary offering in one holding knocks the share price. Margin calls follow, and Archegos cannot meet them.
• 26 March 2021 onward — the brokers race to liquidate. The fastest escape; the slowest do not. Credit Suisse reports roughly $5.5bn of losses, Nomura around $2.9bn.

The mechanism

• Swaps rebuild ownership as financing. The client gets the price exposure and dividends, pays a financing rate, and posts margin — with no share register entry anywhere.
• Leverage was multiplied by fragmentation: modest leverage at each of five brokers becomes extreme leverage in aggregate, invisible to all of them.
• Concentration removed the exit. The positions were large relative to each stock's daily volume, so liquidation itself crushed the price — the losses were partly manufactured by the unwind.
• Margin was counterparty-blind: each bank's risk model assessed its own exposure against a client it believed to be one of several similar relationships.

What it teaches

• Ask who else finances this client. Counterparty risk is a function of the whole balance sheet, not the part you can see.
• Position size relative to liquidity is a risk factor of its own — a portfolio you cannot exit in a week is not really marked at the screen price.
• Disclosure regimes lag instruments. Regulators subsequently moved to close the swap-disclosure gap; that gap existed because the rules were written for shares.
• In a liquidation race, speed beats analysis. The banks that moved first lost least, which is itself a reason such races are destructive.

The mechanisms behind this

Every case on this site is an instrument or a mechanism doing exactly what it was built to do, in a situation nobody had pictured. These are the pages that explain the machinery:

• Total return swaps — ownership rebuilt as financing, with no entry in any share register.
• Leverage — why modest gearing at five brokers is extreme gearing in aggregate.
• Margin & collateral — the call that could not be met, and what happens in the hour after.
• The other side of the trade — the counterparty question every one of the five banks failed to ask.