Momentum is not one strategy. It is two strategies wearing the same coat, and the difference between them is the difference between a 15% annualised gai...
Momentum is not one strategy. It is two strategies wearing the same coat, and the difference between them is the difference between a 15% annualised gain and a 14.8% annualised loss. Reihaneh Haghighi Zadeh's work pulls that coat open. What she finds reframes the whole conversation about why momentum crashes: the crashes are not an unlucky external shock that occasionally hits an otherwise sound factor. They are a structural feature of a specific subset of stocks that happen to dominate momentum's short leg.
The subset is lottery stocks, and understanding why they behave the way they do is the whole point.
Two Sorts That Should Not Belong Together
Start with the two dimensions Zadeh sorts on, because at first glance they look like they measure the same thing and they do not.
The first is ordinary price momentum: rank US stocks on their trailing twelve-month return, skipping the most recent month to avoid short-term reversal, then buy the top and sell the bottom. In her sample running from 1962 through the end of 2023, the high-momentum winners returned 16.5% annualised and the low-momentum losers returned 0.1%. That gap is the momentum premium in its familiar form, and it is large.
The second sort is the lottery characteristic. Here she uses each stock's single highest daily return over the recent period as the proxy. A stock that spiked 18% on one day last month looks like a lottery ticket: a rare, large payoff surrounded by ordinary or negative days. The established finding on these stocks is that they underperform over the long run. Investors overpay for the small chance of the jackpot, bid the price above fair value, and then absorb the loss when the price drifts back to earth. This is a well-documented behavioural drag, not a controversial claim.
The interesting question is what happens when you cross the two. Does momentum work differently inside lottery stocks than inside boring ones?
Where the Return Actually Comes From
It does, and the asymmetry is the finding.
Split momentum strategies by whether the underlying stocks carry lottery-like features and the lottery version shows a visibly stronger cumulative momentum effect, in both value-weighted and equal-weighted constructions. The natural assumption is that this extra juice comes from the winners: lottery stocks that are also flying high must be flying higher than everything else. That assumption is wrong.
The lottery winners returned 15.0% annualised. The plain momentum winners returned 16.5%. The lottery ticket adds nothing to the long leg; if anything it subtracts a little.
The entire difference lives in the losers. Ordinary low-momentum losers returned 0.1% a year, roughly flat. Lottery-like losers returned negative 14.8% a year. That is the number that carries the whole result. The extra momentum effect in lottery stocks is not a story about winners winning more. It is a story about losers losing catastrophically.
This is the mechanism the headline gets backwards. People associate lottery stocks with upside, with the meme rally and the parabolic chart. The data says the danger sits on the other side. A lottery stock with positive momentum behaves acceptably, riding its trend like any other winner. A lottery stock whose momentum has turned does not merely decline. It falls a long way and stays down.
Why the Downside Is Structural, Not Bad Luck
The reason connects two behavioural facts that reinforce each other.
A lottery stock is, by construction, a stock that investors have overpaid for on the hope of a rare rally. That overpayment is a spring loaded with downside. As long as sentiment holds and the price trends up, the spring stays compressed and nobody notices. The stock reads as a healthy momentum name. But momentum turns are exactly the moments when sentiment breaks. When a lottery stock rolls over, two forces fire at once: the ordinary momentum reversal that hits any falling stock, plus the unwinding of the lottery premium that was propping the price above fair value. The stock has further to fall because it started higher than it should have, and it has less support on the way down because the buyers who held it were holding it for a jackpot that is no longer coming.
That is why the crash is deep and durable rather than a quick air pocket. The overvaluation and the reversal are not independent risks that happen to coincide. They are the same overpayment expressed first as inflated price and then as an outsized drawdown.
And this is the bridge to a market-level problem. Momentum crashes, the periodic episodes where the factor gives back years of gains in weeks, have long been the strategy's defining flaw. Zadeh's evidence points to a specific culprit. The short leg of a momentum book is disproportionately populated by these lottery-like losers, and they are precisely the stocks that gap down hardest and rebound most violently when a sharp market rebound flips the losers into the biggest gainers. If you want to know where momentum's tail risk concentrates, it concentrates here.
The Case Against Reading It This Way
The result is a single-study finding on a single characteristic proxy, and that deserves scrutiny before it becomes conviction.
The maximum-daily-return measure is one way to define a lottery stock, not the only way, and it is mechanically correlated with volatility. High-volatility stocks have larger daily moves in both directions, so a sort on maximum daily return partly reloads a sort on volatility. Low-momentum high-volatility stocks have been known to perform poorly for reasons that have nothing to do with lottery preferences. It is fair to ask how much of the negative 14.8% is a lottery effect and how much is a volatility effect wearing a lottery label. Zadeh's framing attributes it to the lottery mechanism; a sceptic would want to see the result survive controls that strip volatility out.
There is also the survivorship-of-findings problem. A long sample from 1962 gives statistical power, but factor interactions that look clean over six decades often fade once trading costs, borrow costs on the shorts, and capacity constraints enter. The lottery losers are frequently small, illiquid, and expensive to short. The paper 15% spread between lottery winners and losers is not the same as a spread an investor could actually harvest.
Neither objection breaks the core observation. Even if part of the effect is volatility in disguise, the practical instruction is unchanged: the stocks that blow up your momentum book sit in the short leg and carry lottery characteristics. What the objections do is caution against treating the exact numbers as tradeable coefficients rather than as a map of where the risk lives.
What Would Confirm It, and What to Do With It
The read is confirmed if the pattern holds out of sample: a momentum strategy that screens lottery-like losers out of its short leg should show meaningfully smaller crash drawdowns without giving up much of the long-leg premium, since the winners were never the source of the edge. That is a testable, near-term claim, and it is the version of the finding that matters for anyone running the factor rather than reading about it.
The practical implication follows directly from where the return comes from. The winners of a momentum strategy do not need lottery filtering, because lottery features add nothing on the long side. The losers do. The cleanest improvement to a momentum book is not to chase the highest-momentum names harder but to refuse to short the worst-momentum lottery stocks, or to hedge them explicitly, precisely because their crash behaviour is what turns a good year into a lost decade during a momentum unwind.
The thesis breaks if a volatility-neutral version of the sort erases the loser asymmetry, because then the lottery label is decorative and the real driver is something already priced by existing risk models. Until that test fails, the sharper reading stands: momentum's premium and momentum's catastrophe are not opposite ends of the same factor. They are separable, they live in different stocks, and the crash risk everyone fears is concentrated in a corner of the short leg that can, in principle, be fenced off.
