*Momentum is the closest thing finance has to an embarrassment. It shouldn’t work, buying what has already gone up violates every instinct about buying low, and yet it is one of the most stubbornly documented patterns in markets, replicated across decades, countries, and asset classes. Here is what momentum investing actually is, the leading explanations for why it persists, the failure mode that can undo it in a single violent turn, and how a disciplined system tries to keep the edge while side-stepping the catastrophe.*
# What momentum investing is
Momentum investing is a strategy of buying the assets that have most strongly outperformed recently and holding them while that relative strength persists, on the empirically documented tendency for recent winners to keep outperforming recent losers over intermediate horizons. That is the whole idea, and it is deliberately mechanical: rank a universe by past return, tilt toward the top of the ranking, and refresh as the leadership rotates. No view about fair value, no story about the company , just the observation that price trends, once established, have a measurable habit of continuing for a while.
The horizon matters, because momentum is a middle-distance phenomenon. Over very short windows, days to a few weeks , prices tend to *reverse*, not continue. Over very long windows, several years, they mean-revert, and yesterday’s darlings become tomorrow’s value traps. It is in the intermediate band, conventionally something like the trailing three-to-twelve months, that the continuation shows up most reliably. Momentum is not “trends last forever.” It is “trends last a little longer than the market has yet priced in.”
# The evidence: a remarkably durable anomaly
What makes momentum more than a folk belief is how hard it has been to make it disappear. The academic literature dates the modern formulation to work by Jegadeesh and Titman in the early 1990s, who showed that ranking US stocks on past returns and holding the winners produced returns that ordinary risk factors couldn’t explain away. The interesting part is everything that came after: the effect was found again in international equity markets, in currencies, in commodities, in bonds, and even running back through more than a century of historical data that the original researchers never saw. An anomaly that survives in samples chosen *after* it was published, in asset classes it wasn’t discovered in, is rare. Momentum is one of the few.
None of that makes it a guarantee, and the honest framing is important: momentum is a *tendency*, measurable across large populations and long windows, not a property of any single position. Plenty of individual winners roll over the day after you buy them. The edge is statistical, a small, repeatable tilt that shows up in the shape of a whole distribution, the same way every honest systematic edge does, as how a rule-based signal works spells out. Treating it as a promise about the next trade is how people get hurt by a strategy that is, in aggregate, sound.
# Why does momentum work?
The uncomfortable truth is that the profession does not fully agree, and a strategy whose own advocates can’t settle the mechanism deserves a careful reader. There are two broad camps, and the most credible position borrows from both.
The behavioural story is the more intuitive one. Investors underreact to news: when a company posts genuinely good results, the price drifts toward fair value over weeks rather than repricing instantly, so the recent winner keeps climbing as the information slowly diffuses. Then a second, opposite bias takes over, herding and trend-chasing, where rising prices attract buyers simply *because* they are rising, pushing the move past fair value before it eventually corrects. Underreaction starts the trend; overreaction extends it. Both are well-documented features of how humans process information, and together they describe momentum’s characteristic arc rather well.
The risk-based story is less satisfying but harder to dismiss: if winning stocks are simply riskier in some way the standard models miss, more exposed to bad states of the world, more fragile to a sudden change in conditions, then their extra return is just fair compensation for bearing that risk, not a free lunch at all. There is something to this, and the next section is essentially its evidence. The pragmatic conclusion is that the debate is genuinely unsettled, and the safest assumption is that the premium is part mispricing and part risk, which means it can shrink, and it can bite.
# The honest catch: momentum crashes
Every strategy has an Achilles heel, and momentum’s is unusually sharp. Because a momentum book is, by construction, long whatever has been working and short or absent from whatever has been failing, it carries a hidden bet on the current regime*continuing*. When the regime snaps, a violent market bottom where the most-beaten-down names rocket and the recent leaders are dumped, that bet inverts all at once. The result is the “momentum crash”: short, brutal drawdowns that cluster precisely at major turning points, when the strategy is most confidently positioned the wrong way. The sharpest historical examples come at the violent recoveries off market bottoms, when leadership flips overnight and yesterday’s winners become the day’s worst losers.
There is a quieter cost too. Momentum needs trends to feed on, and in choppy, directionless, mean-reverting tapes there are none. The strategy buys the recent leader just in time for leadership to rotate, sells into the rotation, and buys the next leader just before *that* rotates, a slow bleed of small losses and transaction costs that the textbook backtests, run on clean data, tend to understate. So momentum has two distinct enemies: the rare regime-turn crash that hurts a great deal at once, and the common sideways grind that hurts a little, continuously. A strategy that ignores either one is selling you only half the picture.
# So, does momentum investing work?
Yes, with two large asterisks. The momentum premium is real, durable, and about as well-replicated as anything in empirical finance, which is why it survives as a strategy at all. But it is a statistical tendency, not a promise on any single position, and it carries a specific, violent failure mode at regime turns plus a steady drag in trendless markets. So momentum works in the way a sound edge works: applied with discipline, across many names and many days, by someone who has planned for the times it doesn’t.
Which is the entire reason the gate exists. Capturing the premium is the easy half; surviving the crash is the half that decides whether you are still around to compound. A momentum strategy without an answer for its own Achilles heel is a bet that the regime never turns, and the regime always turns. The useful version is the one that treats momentum’s weakness as a design problem to solve in the open, not a footnote to omit from the brochure.