Almost every academic paper on cross-sectional stock returns from the last thirty years mentions momentum. The finding is remarkably robust: stocks that have outperformed over the past 6–12 months tend, on average, to continue outperforming over the next 3–6 months. It works across countries, across asset classes, and across time periods where almost nothing else has. Understanding why matters more than the pattern itself.
The empirical record
Jegadeesh and Titman's 1993 paper documented that portfolios of past winners (top-decile 6-month performers) outperformed portfolios of past losers by roughly 1% per month over the following 3–12 months. Subsequent research has extended the finding to nearly every major equity market, to bonds, to commodities, and to currencies. The premium has compressed since the mid-2000s as awareness has grown and capital has crowded in, but it has not disappeared.
Momentum is one of the five factors in most factor models — Carhart's four-factor model added it to the Fama-French three explicitly because the anomaly could not be explained by size, value, or market beta.
Why it works, if it works
There is no single accepted explanation, but three are commonly cited.
Behavioural underreaction. Investors take time to update beliefs after new information. A company reporting a positive earnings surprise sees its stock rise, but not far enough to reflect the full change in fundamentals. Over the next several months, as analysts revise estimates upward and more investors incorporate the news, the stock keeps drifting up. The pattern is the same in reverse for negative surprises.
Behavioural anchoring and herding. Once a stock has established a trend, later observers rely on the trend itself as information. This produces a self-reinforcing dynamic — the trend attracts flows, which extend the trend, which attracts more flows.
Compensation for risk. A less popular explanation: momentum stocks have higher exposure to some risk factor (macro sensitivity, drawdown risk during regime shifts) that requires ongoing compensation. The evidence for this explanation is mixed.
The truth is likely a combination. Momentum probably reflects a real behavioural pattern that also carries real risk, and the two effects are hard to disentangle empirically.
Why it feels wrong
Momentum is the most counterintuitive of the well-documented factors. The mental model most beginners bring to investing is buy-low-sell-high, which is almost the opposite of momentum's mechanic (buy high, hope it stays there, and don't hold too long). This is why the strategy is one of the most consistently under-implemented by retail investors despite decades of published evidence: it feels wrong to buy something that has already risen substantially.
The best way to reconcile this is to notice that momentum and value are not opposites. Value says "buy things that are cheap relative to fundamentals." Momentum says "buy things that have been rising recently." A stock can be both — cheap on fundamentals and rising — and the combination of the two factors historically has produced better outcomes than either alone.
The dark side of momentum
Momentum's Achilles heel is that when it fails, it fails catastrophically. The two most-documented episodes are 2009 and early 2016, both cases where past-loser stocks reversed sharply against past-winner stocks over short windows. In 2009, the momentum factor lost roughly 50% peak-to-trough — a "momentum crash" that took years to recover.
The reason: when a market has spent a long time falling, the "past losers" become extraordinarily depressed, and when the market turns, they rebound faster than the past winners. Momentum strategies, having built up large short exposure to the past losers, take the full force of that rebound in reverse.
Risk-managed versions of momentum — using volatility filters, sector-neutral construction, or trend confirmation — reduce these crashes at the cost of some of the underlying premium. The trade-off is real and not free.
How the pattern shows up in practice
Sector rotation. Sectors that outperform over one quarter show elevated probability of continuing to outperform in the next quarter. This is momentum expressed at the sector level, and it is the most-commented-on version of the effect in short-term market commentary.
Individual-stock trend persistence. Named "trend following" in the retail literature and "momentum" in the academic one, the same pattern shows up in single-stock returns.
Cross-asset. Trend-following funds (CTAs) trade momentum across futures markets in commodities, currencies, and bonds. Their long-term track record is real but volatile, and they have delivered particularly strong returns during crises when almost everything else has fallen.
The rule to internalise
Momentum is one of the few anomalies with decades of empirical support. It is also one of the hardest to implement, because it requires holding recent winners and selling recent losers — the exact opposite of what most emotional impulses suggest. The premium exists partly because it is uncomfortable, and the discomfort is a feature of the strategy, not a flaw in it.
Educational content only. Not investment advice.