How did Archegos collapse in 2021 expose hidden leverage?

The short answer

Bill Hwang ran a family office that, by 2021, had built enormous concentrated positions in a handful of US and Chinese stocks. The trick was that he did not own the shares directly. He held total return swaps (TRS) with eight major banks—Credit Suisse, Nomura, Goldman Sachs, Morgan Stanley, UBS, Deutsche Bank, MUFG and Wells Fargo—each of which hedged its exposure by buying the underlying shares.

The economic effect was identical to direct ownership. The reporting effect was not. Because the banks held the shares (and Archegos held only the swap), Archegos never appeared on the 13F filings or 5% beneficial-ownership thresholds that would normally disclose a major holder.

When ViacomCBS dropped 27% on March 24-25, 2021 after a stock issuance, Archegos faced margin calls it could not meet. The banks discovered, in real time, that they each held large positions in the same names against the same client. The race to liquidate produced approximately $10 billion in combined losses.

→ New to derivatives mechanics? Investment vehicles framework

What the data shows

The Archegos collapse remains one of the largest single-client losses in modern prime brokerage history.

The empirical picture (bank quarterly reports, Credit Suisse Special Committee Report July 2021):

• Default trigger: March 26, 2021. Margin call on ViacomCBS-related TRS positions following a 27% single-day stock decline
• Combined bank losses: approximately $10 billion across all prime brokers involved
• Credit Suisse: $5.5 billion (largest single-bank loss); forced to raise $1.9 billion in capital and cancel executive bonuses; CRO and head of investment banking departed
• Nomura: $2.85 billion (initially reported $2 billion in April 2021, revised upward); led to scaling back of US/European prime brokerage
• Morgan Stanley: approximately $911 million to $1 billion; UBS $774 million; Goldman Sachs limited losses through speed
• Archegos peak NAV: estimated at $20-36 billion at various points 2020-2021; effective gross exposure across all banks estimated at $50-100 billion
• Credit Suisse fee income from Archegos in 2019: $17.5 million (the bank carried multi-billion-dollar exposure for what amounts to a low-fee relationship)

The exception is Goldman Sachs and Morgan Stanley: both moved within hours of the default to liquidate positions before broader announcements were made; Goldman reportedly sold over $10 billion of Archegos-linked shares on March 26 alone, escaping with minimal losses (David Solomon reported Q1 2021 prime brokerage record balances).

→ Dataset: Credit Spread vs VIX

Why it happens — the macro mechanism

The Archegos collapse was not a story of excess leverage in the abstract—it was a story of leverage that escaped reporting because it lived in derivatives, not in shares. Cleared products do not allow that concealment: margin is called against the aggregate position, the mechanism at the centre of central clearing and counterparty risk.

The total return swap mechanism. A TRS is a contract where one party (the bank) pays the other (the client) the total return on a stock—price appreciation plus dividends—in exchange for a fixed funding rate. The client posts margin (typically 15-25% of notional) but does not own the stock; the bank hedges by buying the actual shares. Economically, Archegos owned the upside and downside of the position. Legally and regulatorily, the bank did.

The reporting blind spot. SEC rules require institutional investment managers to disclose positions on Form 13F, and any holder reaching 5% of a publicly listed company must file Schedule 13D. Both rules apply to direct ownership and certain types of derivative exposure—but not to TRS positions where the bank holds the shares as a hedge. This is the angle most coverage misses: Archegos was not a Bill Hwang character flaw, it was a structural reporting gap that allowed any sufficiently creative client to build position sizes that no single bank could see across the system.

Each of the eight prime brokers saw its own exposure but had no view of the aggregate book.

Concentration and synchronized hedge. Because every bank hedged its TRS by buying the same shares (ViacomCBS, Discovery, Baidu, Tencent Music), the banks collectively held a very large block of those names. When Archegos defaulted and banks needed to liquidate hedges, they competed against each other to sell the same stocks—creating the 27% one-day moves and the loss cascade.

Synthesis by regime. In calm regimes (2019-early 2021), the TRS structure was a profitable, non-disclosed financing channel for funds wanting concentrated exposure without 13F friction; banks earned attractive financing spreads and the structure operated cleanly. In a stress regime (March 2021), the exact features that made TRS attractive—non-reporting, off-13F, leveraged—turned the structure into a synchronized fire-sale generator. The pivot was a single-name decline (ViacomCBS dilutive issuance) rather than a macro shock; this distinguishes Archegos from LTCM 1998 (macro: Russia/Asia) and from 2008 (macro: subprime).

Archegos was not a leverage failure but a visibility failure: every bank saw its piece, none saw the whole, and the regulator’s reporting form had been written for a different financial era.

→ Framework: Systemic risk indicators & stress signals

What it means for different economic actors

Banks (the immediate losers). Credit Suisse and Nomura suffered concentrated losses that triggered strategic retrenchment; Credit Suisse exited prime brokerage entirely. Goldman Sachs and Morgan Stanley demonstrated that operational speed and pre-defined liquidation playbooks made the difference between escaping unscathed and absorbing billions.

Hedge fund clients (the affected adjacent population). Stocks held by Archegos collapsed 30-50% in days, transmitting losses to long-only and other hedge fund holders of those same names. Hedge funds with overlapping positions had to disclose Q1 2021 drawdowns sometimes attributable to the Archegos liquidation.

Regulators. The SEC adopted new rules in 2023 requiring 10-day disclosure of large derivative positions (the “security-based swaps” rule under Dodd-Frank Section 763) and tightened beneficial-ownership reporting timelines. Prime brokerage arrangements have been repriced by all surviving providers since.

A common error is to focus on Bill Hwang’s personal trading style. The structural lesson is that the same TRS mechanism is still legal, still profitable, and still creates the same reporting gaps for any sufficiently large family office or hedge fund willing to use it.

Practical observation

Go deeper

Related questions

• What are prime brokerage arrangements?
• How do hedge funds affect market stability?
• What is rehypothecation?
• What is shadow banking?

Frequently asked questions

Why did regulators not catch Archegos before March 2021?

Archegos was a family office, exempt from registration as an investment adviser under the Dodd-Frank “family office rule” finalized in 2011. Combined with TRS structures that placed beneficial ownership at the bank rather than the fund, the firm fell into a regulatory blind spot: not registered with the SEC, not visible in 13F filings, and visible only bilaterally to each prime broker as their own exposure. No single regulator or bank had a system-wide view.

Could the same structure cause another Archegos-style event today?

The mechanism remains legally available. SEC rules adopted in 2023 require 10-day disclosure of large positions in security-based swaps (under Dodd-Frank Section 763) and shortened beneficial-ownership reporting timelines, partially closing the gap. Banks have also tightened single-name concentration limits internally and improved cross-counterparty information sharing through industry working groups. But the structural visibility problem—that no single regulator sees aggregate TRS positions across all major banks—has not been fully solved. The angle worth highlighting: the lesson of Archegos is reporting infrastructure, not personality.

How did Goldman Sachs and Morgan Stanley escape the worst losses?

Both banks identified the deteriorating risk profile faster and acted more decisively. According to public reporting, Goldman Sachs began liquidating Archegos-related hedges on March 26, 2021—the day of default—selling more than $10 billion of stocks in a single session through block trades and algorithmic execution. Morgan Stanley executed similar speed. Credit Suisse and Nomura stayed in the position longer, in part because of internal coordination delays, and incurred their losses as the underlying stocks continued to decline through early April. Speed of execution distinguished outcomes more than initial exposure size.

Last updated — 28 July 2026