Case: Momentum Crash 2009 / Meme Spikes

What momentum is, and what it has paid

**Cross-sectional momentum** ranks stocks by their return over the past several months — typically the last twelve, skipping the most recent month — buys the top group and sells the bottom group. Jegadeesh and Titman’s 1993 study found that buying past winners and selling past losers over three- to twelve-month horizons earned roughly one percent a month in U.S. stocks, and the effect has since been found in many countries and asset classes. It is related to, but not the same as, the time-series momentum of T20, which asks whether each asset is above its own past rather than how it ranks against others. The explanations are behavioural and structural: investors under-react to news and then herd, analysts revise slowly, and institutions trade gradually. Whatever the cause, momentum has been one of the most persistent findings in the literature — and one with an unusual return distribution. It earns modest, steady gains for long stretches and then gives a large share of them back very quickly. • Cross-sectional: rank stocks against each other; long the winners, short the losers. • Time-series (T20): each asset against its own history. • Typical formation: past 12 months, skipping the latest month. • Return profile: steady gains, interrupted by sharp, rare crashes.

How momentum crashes

Daniel and Moskowitz found that momentum’s worst losses are not random. They cluster in what they call panic states: after a market decline, when volatility is high, the market then rebounds sharply. In those moments the short side of a momentum portfolio — last year’s losers — is full of high-beta, beaten-down stocks with the most room to recover, and they rise far faster than the winners the strategy owns. The long side rises a little; the short side explodes; the strategy loses heavily in weeks. Their two signature episodes are the summer of 1932 and the spring of 2009. In 2009 the losers were largely financials, which had fallen furthest in the crisis and rebounded hardest once the worst fears receded. The replay uses the Financial Select Sector ETF against SPY from the March 2009 low to show the size of that rebound on real prices. The useful part of the finding is that crash risk is partly forecastable: it is highest when the market has fallen and volatility is high. Barroso and Santa-Clara showed that scaling momentum exposure down when its own recent volatility is high reduced the crashes substantially. The lesson for any momentum-style method is the same as for trend: the danger is known in advance, and the response is size. Anatomy of a momentum crash — Before: A long market decline: the losers are deeply beaten down, with high beta · Trigger: The market rebounds sharply from a panic low · Long side (winners): Rises modestly · Short side (losers): Rises fastest — the book loses heavily in weeks ←

Meme spikes: the squeeze as a momentum event

A short squeeze is a momentum crash in miniature, inside one stock. In January 2021 GameStop had short interest above its entire float — 122.97% by the SEC staff’s count — and a crowd of retail buyers organised on social media. The popular story added a second engine: option dealers who had sold calls, buying shares to hedge them. As the price rose short sellers did buy to cover, and the rise fed itself: the replay shows the highest close in five sessions at about eight times the starting price. Then the fuel ran out. Once shorts had covered and the crowd’s buying slowed, nothing was left to push, and the price collapsed — the replay shows it down by more than four-fifths within ten sessions of the peak. For a momentum signal the episode is poison either way: a stock ranked as a loser becomes the best performer in days, and a momentum buyer who arrived after the spike bought the top. The SEC staff report tested the story against trade data. Buying by accounts with large short positions was a small fraction of overall buy volume, and the staff found no evidence of a gamma squeeze: the surge in customer options volume was mostly put buying, and market makers were buying calls rather than writing them. Its conclusion was that positive sentiment, not covering, sustained the rise — and that the brokerage restrictions came from clearing-house deposit requirements, not a decision about the stock. Positioning, a crowd and plumbing — not fundamentals — wrote the chart. • Fuel: short interest above the float and a coordinated crowd. • Engine: the crowd’s buying with covering on top — the SEC staff found no evidence of a gamma squeeze. • Exhaustion: when the fuel is spent, the price collapses. • For momentum: ranks flip in days, and late buyers buy the top.

Living with a strategy that crashes

Strategies with crash risk are not necessarily bad; they are mispriced by people who judge them on their average. A momentum or trend method should be sized for its worst episode, not its typical month, and the worst episode is predictable in shape: a rebound after a panic for momentum, a crowded short meeting forced buying for a single stock. Four practical rules follow. Scale exposure to volatility: when a strategy’s own volatility is high, risk less, which is exactly what reduced momentum’s crashes in the research. Diversify across names and markets, so that one squeeze is a small loss rather than the account. Avoid being short the most crowded, most beaten-down names after a crash — the short side is where the crash lives. And treat a squeeze as a positioning event: when a stock rises because shorts are covering and dealers are hedging, the fundamentals have not changed, and the move ends when the forced buying does. Sizing for the crash, not the average — Volatility scaling: Risk less when the strategy’s own volatility is high · Diversification: Many names and markets — one squeeze stays small · Crowded shorts after a crash: The short side is where momentum crashes live · Squeezes: Positioning events: they end when the forced buying ends ← A strategy that has not crashed in your sample has not been tested by the condition that defines it.

Where momentum comes from

Momentum has been documented across countries, asset classes and more than a century of data, and researchers still argue about why it exists. The behavioural explanations start from slow reactions to news. Hong and Stein argued that information spreads gradually through a market of investors who each see only part of it, so prices adjust in steps rather than all at once. Barberis, Shleifer and Vishny described investors who underreact to news at first — conservatism — and later overreact to long runs of it. Daniel, Hirshleifer and Subrahmanyam traced momentum to overconfidence and self-attribution (P2, P11). The disposition effect (P5) supplies a mechanism too. Investors who sell winners too early and hold losers too long slow the price adjustment in both directions: good news is sold into, so prices rise less than they should at first and keep drifting up; bad news is held through, so prices fall less at first and keep drifting down. Grinblatt and Han showed that stocks with more unrealised gains among their holders tend to have more momentum, and Frazzini found mutual fund managers show the same pattern. Risk-based explanations hold that momentum stocks carry exposures that deserve a premium, including exactly the crash risk this lesson describes. The two views are not exclusive, and the practical point survives either way: momentum is a premium paid for bearing a risk that arrives suddenly, and like every published pattern (T1) it has been more modest since it became widely known. • Gradual diffusion of news (Hong & Stein): prices adjust in steps. • Underreaction, then overreaction (Barberis, Shleifer & Vishny; Daniel, Hirshleifer & Subrahmanyam). • The disposition effect slows adjustment both ways (Grinblatt & Han; Frazzini). • Or a premium for crash risk — and smaller since publication either way. Momentum, explained four ways — Slow diffusion of information: Prices catch up in steps · Underreaction then overreaction: Drift, then excess · The disposition effect: Winners sold early, losers held — both drift ← · A risk premium: Paid for bearing sudden crashes

The two great crashes, and the fix the research found

Daniel and Moskowitz found that momentum’s two worst months on record were July and August 1932, and that it crashed again from March 2009. Both came after deep market falls, and both had the same anatomy. After a long decline, the stocks that had fallen most were the most leveraged and most fragile, and their sensitivity to the market — their beta — had risen above three, while the past winners’ had fallen below one half. A momentum book short the losers and long the winners was therefore heavily short the market itself, without anyone having chosen that. When the market rebounded sharply, the losers it was short rose fastest of all. Two research fixes followed. Barroso and Santa-Clara showed that momentum’s own volatility is predictable, and that scaling the position down when its recent volatility is high — and up when it is low — kept most of the return while sharply reducing the crashes, roughly doubling the strategy’s Sharpe ratio in their sample. Daniel and Moskowitz built a dynamic version that also cut exposure in the “panic states” after market falls, when crash risk is highest. For a trader the lesson generalises beyond momentum. Every strategy has states in which its hidden exposures grow — beta, leverage, crowding — and the crash comes from the exposure no one chose. Measure the strategy’s exposure to the market and its own volatility regularly, and size down when either spikes; that is the cheapest insurance any strategy can buy. • Momentum’s worst months: July and August 1932; it crashed again from March 2009. • After a fall, past losers’ betas rose above 3 and winners’ fell below 0.5 — a hidden short on the market. • Volatility scaling (Barroso & Santa-Clara) kept most of the return and cut the crashes. • Measure hidden exposures and volatility; size down when they spike. Anatomy of a momentum crash — After a deep fall: Losers’ betas above 3, winners’ below 0.5 · What the book has become: A hidden short position on the market ← · The sharp rebound: The shorted losers rise fastest · The fix: Scale exposure down when volatility is high

Sources

• Momentum returnsJegadeesh & Titman, “Returns to Buying Winners and Selling Losers” (Journal of Finance, 1993)
• Momentum crashesDaniel & Moskowitz, “Momentum Crashes” (Journal of Financial Economics, 2016)
• Volatility-managed momentumBarroso & Santa-Clara, “Momentum Has Its Moments” (Journal of Financial Economics, 2015)
• The January 2021 squeezeSEC Staff Report on Equity and Options Market Structure Conditions in Early 2021 (October 2021)