Momentum trading strategy: rules, the edge and the cost
Momentum trading buys what is going up and sells what is going down — and it is one of the most robustly documented edges in finance, surviving across decades and asset classes. But it pays for that edge with painful whipsaws and sudden crashes. This guide explains the two flavours, the rules, and the honest risks.
Why momentum is a documented edge
Decades of academic research — most famously Jegadeesh and Titman's work on relative-strength portfolios — found that assets that outperformed over the past 3–12 months tended to keep outperforming over the next few months. The effect appears across stocks, currencies, commodities and crypto. It is not a guarantee; it is a statistical tilt that pays off on average while losing on plenty of individual trades. That is the opposite profile of mean reversion.
Two flavours of momentum
- Time-series (absolute) momentum — trade each asset against its own past: go long if it rose over the lookback window, flat or short if it fell. This is close to trend following.
- Cross-sectional (relative) momentum — rank a basket of assets and hold the strongest, short or avoid the weakest. This is the classic stock-market momentum factor.
Entry and exit rules
A simple bot ranks assets by their N-period return (say 90 days), buys the top performers, and rebalances monthly. Exits matter as much as entries: momentum strategies suffer "momentum crashes" during sharp reversals, so a volatility-scaled position size and a trailing stop soften the worst drawdowns. The whipsaw cost is real — you re-enter near tops and exit near bottoms during choppy regimes.
The code
python · momentum.pydef ret(closes, n=90): return closes[-1]/closes[-n] - 1
ranked = sorted(universe, key=lambda s: ret(prices[s]), reverse=True)
longs = ranked[:3] # hold the strongest, rebalance monthly
Backtest momentum across a full cycle — including a crash and recovery — in the backtester, because its risk lives in the rare reversal, not the calm trend.
Frequently asked questions
Does momentum trading actually work?
Momentum is one of the most robustly documented anomalies in finance, persisting across markets and decades. It is a statistical edge, not a sure thing — it loses on many individual trades and suffers sharp 'momentum crashes' during reversals.
What lookback period is best for momentum?
Research commonly uses 3 to 12 months for the lookback. Shorter windows react faster but whipsaw more. The robust choice performs across a range of lookbacks rather than relying on one tuned value.
How is momentum different from trend following?
They overlap heavily. Trend following usually trades each asset against its own history (time-series momentum), while the classic momentum factor ranks a basket and holds the strongest (cross-sectional). Both buy strength and sell weakness.
What is a momentum crash?
It is a sharp loss that momentum strategies suffer when a beaten-down market violently rebounds — the previously weak assets the strategy avoided or shorted surge. Volatility scaling and stops reduce, but do not eliminate, this risk.