March 2025: Momentum Crash Driven by Forced Liquidation, Not Just Factor Rotation
Equity markets have just experienced one of the sharpest momentum drawdowns in decades. Over a three-week period, the long side of the momentum trade—dominated by large-cap growth and AI-linked equities—collapsed, giving back two years of gains. This was not a slow rotation but a violent unwind, marked by indiscriminate selling and massive market cap destruction.
The S&P 500 erased approximately $5.8 trillion in value from peak to trough. Nearly 40% of that decline came from just 50 stocks—the top performers over the prior 12 months—highlighting the extent of concentration risk embedded in the market.
While some strategists attribute the move to “crowding” or factor rotation, those labels are downstream of the actual drivers: forced deleveraging, margin calls, and systemic risk reduction. Large, leveraged exposures—many structured through swaps and options—were unwound rapidly, not because of a change in fundamental view, but because risk systems, collateral triggers, and liquidity demands left no alternative.
This is not a CTA-driven event in the pure sense. While CTAs do model momentum and may have contributed mechanically to some of the pressure, the scale and speed of the move point to more acute factors:
Margin-driven selling by hedge funds and levered players
Dealer hedging in short-gamma regimes, amplifying downside
ETF and derivative structures forcing price-insensitive flows
Index rebalancing effects hitting cap-weighted exposures in mega-caps
The shift in narrative—from high-for-longer rates and AI euphoria to growth concerns and political risk—may have been the trigger, but it was the embedded leverage and exposure that dictated the severity of the move.
This structure is not new. In 1929, the mechanism was margin debt, with investors borrowing up to 90% of capital to buy stocks. Brokers, in turn, borrowed from banks and corporations through the call loan market, often recycling capital back into the market. The system was held together by collateral that evaporated as prices fell, triggering a reflexive spiral of liquidation.
Today’s version substitutes derivatives for margin accounts, and prime brokers for retail brokerages. Leverage is embedded in swaps, options, and leveraged ETFs. But the result is the same: once collateral is impaired or volatility increases, exposure must be cut—mechanically and often simultaneously. Dealer gamma positioning, risk-parity compression, and VaR breaches fill the role once played by margin clerks and callable loans.
This structure extends well beyond institutional trading. As of year-end 2024, the table below shows the largest U.S.-listed leveraged bullish ETFs by assets under management (AUM). AUM reflects actual investor capital, while notional exposure represents the gross market exposure created through internal leverage.
While investor losses are limited to principal, leveraged ETFs amplify both returns and volatility by the stated leverage factor. This increases the probability of severe drawdowns and capital loss—even though absolute losses are capped. In a momentum-led drawdown, these instruments can act as accelerants, creating mechanical selling flows during rebalancing and feeding volatility back into the broader system.
Flows rotated into defensives—utilities, insurance, and other low-volatility sectors—resulting in rapid multiple expansion for those names. However, this was not a broad move into value or cyclicals. The short leg of the momentum unwind showed minimal strength, confirming that this was a de-risking event, not a low-quality rally.
There is no evidence yet of structural regime change. The drawdown is consistent with historical momentum crashes in magnitude (~12–13%), and while further downside is possible if macro conditions worsen, the liquidation phase appears to have run its course—at least for now.
What matters going forward is not sentiment or style preference, but who still needs to reduce exposure, and how much risk remains embedded in derivatives and leverage across the system.