Momentum investing uses defined past-return or trend signals to rank assets, form portfolios, and rebalance while accepting reversal and trading-cost risks.
Momentum investing is a rules-based strategy that favors assets with stronger recent returns or positive trends and may underweight, avoid, or short assets with weaker recent returns. A complete momentum strategy must define the universe, lookback period, ranking or trend rule, holding period, weighting, rebalance schedule, and risk controls. Recent performance can reverse abruptly and does not predict future returns with certainty.
| Form | Comparison | Example signal | Typical decision |
|---|---|---|---|
| Cross-sectional momentum | Each asset against peers | Rank trailing total returns | Overweight recent relative winners and underweight losers |
| Time-series momentum | An asset against its own history | Return above zero or price above a trend measure | Take positive, neutral, or negative exposure based on trend |
| Technical momentum indicator | Recent gains, losses, or moving averages | RSI or MACD | Generate a trading signal under a defined rule |
| Earnings momentum | Fundamentals or expectations through time | EPS growth, surprise, or estimate revision | Rank changes in business performance or expectations |
The methods may point in different directions. A stock can rank highly on 12-month relative return while a short-term oscillator signals that recent gains are stretched.
One research convention measures cumulative return from month (t-12) through month (t-2), omitting the most recent month. Using total-return relatives, the signal can be expressed as:
where (r_{i,t-k}) is security (i)’s return in a prior month. The exact labels for this convention vary, so the formation window should be written as calendar endpoints rather than assumed from shorthand alone.
The Fama-French Data Library describes its U.S. momentum portfolios as formed monthly using prior 2-12 returns. Its published Mom factor averages high-prior-return portfolios and subtracts low-prior-return portfolios while controlling the construction for size.
Skipping the most recent month is a research design choice intended to keep the intermediate-horizon signal separate from very short-term return behavior. It is not a universal requirement for every momentum strategy.
Assume five eligible stocks have these hypothetical trailing total returns over the defined formation window:
| Stock | Formation-period return | Rank |
|---|---|---|
| D | 24% | 1 |
| A | 18% | 2 |
| B | 7% | 3 |
| C | -2% | 4 |
| E | -10% | 5 |
A simple long-only rule might select D and A. A research long-short portfolio might hold D and A long while selling C and E short. Those are materially different implementations: the long-short version requires borrowing, collateral, financing, and short-risk controls.
Suppose equal-weighted D and A return -8% and -4% in the next month, while C and E rebound 6% and 10%. The hypothetical long side returns:
The loser basket returns 8%, so the gross winner-minus-loser spread is (-14%) before borrowing and trading costs. This illustrates reversal risk; it is not an estimate of a normal or maximum momentum loss.
The Relative Strength Index and Moving Average Convergence Divergence are technical indicators, not synonyms for the academic momentum factor.
| Measure | Input | Output | Key distinction |
|---|---|---|---|
| Cross-sectional return rank | Comparable past returns across securities | Relative rank or portfolio assignment | Requires a peer universe |
| Relative Strength Index | Magnitudes of recent gains and losses | Bounded oscillator | Describes one asset’s recent price behavior |
| MACD | Difference between selected exponential moving averages | Trend-following indicator | Depends on chosen moving-average lengths |
| Time-series return signal | Asset’s own trailing return or trend | Directional signal | Does not require ranking against peers |
Relative Strength can also mean either comparative performance or a technical calculation, depending on context. The formula and benchmark should always be stated.
Overlapping portfolios are common in research designs with multi-month holding periods. An analyst should identify whether each month’s reported return combines several formation cohorts and should reproduce that logic consistently.
Proposed explanations include delayed reaction to information, investor behavior, institutional trading, and compensation for risks that become severe during reversals. The existence of several explanations matters: a historical return pattern alone does not prove one mechanism, and the dominant mechanism can differ across markets and periods.
The original evidence and later research are empirical. They do not establish a law that winners must continue winning.
| Strategy | Main selection basis | Main difference from price momentum |
|---|---|---|
| Value Investing | Price relative to estimated fundamentals | Can buy recent underperformers if they appear undervalued |
| Growth investing | Expected business growth | Focuses on company fundamentals and valuation, not past return alone |
| Contrarian investing | Reversal or mispricing thesis | Often takes the opposite side of recent market behavior |
| Trend following | Direction of an asset or market | Commonly time-series rather than peer-relative |
| Earnings momentum | Reported or expected earnings changes | Uses fundamental information rather than price history alone |
A security can satisfy more than one style definition. A momentum portfolio may unintentionally acquire growth, size, sector, beta, or volatility exposure.
Confirm total-return calculations, corporate-action adjustments, time zones, stale prices, missing data, and the exact point at which the signal becomes tradable.
Separate the period used to design the strategy from out-of-sample and live periods. Repeatedly changing the lookback, skip, quantile, and holding period after viewing results creates data-mining risk.
Momentum portfolios can trade frequently and may seek to buy or sell the same names as other systematic strategies. Estimate transaction costs under realistic volume, spread, and market-impact assumptions.
Use factor models and holdings analysis to identify market, size, value, sector, country, currency, and volatility exposures. A positive historical alpha under one model can disappear under another.
Inspect concentrated loss periods, rebound markets, liquidity stress, and the behavior of the short side. Average return and annualized volatility can hide abrupt nonlinear losses.
This article provides general financial education. It does not recommend momentum trading, short selling, a signal, security, fund, or portfolio allocation. Past performance and backtests do not guarantee future results.