Momentum is the tendency for securities that have performed well over the past 3-12 months to continue outperforming, and for past losers to continue underperforming.
Construction
The standard academic measure uses 12-1 month returns: the cumulative return over months t-12 to t-2, skipping the most recent month (to avoid short-term reversal). The factor UMD (Up Minus Down) or WML (Winners Minus Losers) goes long the top decile/tercile of past winners and short the bottom.
Discovery
Jegadeesh and Titman (1993) documented momentum using NYSE and AMEX stocks from 1965-1989. All 32 strategies they tested (varying formation and holding periods from 3-12 months) produced positive returns. The best strategy (12-month formation, 3-month holding) yields 1.49%/month. The canonical 6/6 strategy returns about 1%/month (12.01% annualized).
Carhart (1997) incorporated momentum as the fourth factor in what became the Carhart four-factor model, showing that mutual fund performance persistence is largely explained by momentum loading rather than manager skill.
Breadth of evidence
Asness, Frazzini, Israel, and Moskowitz (2014) show momentum has been present in:
- 212 years of U.S. equity data (1801-2012)
- 40+ countries
- Over a dozen asset classes
Among standard factors, momentum has the highest average return (UMD: 8.3%/year) and Sharpe ratio (0.50), exceeding both HML (4.7%/year, 0.39) and SMB (2.9%/year, 0.26).
Myth-busting
Asness et al. (2014) refute ten common myths:
- Momentum works on both long and short sides (roughly equal contribution)
- It is robust among large caps (6.8%/year)
- Real-world institutional trading costs are roughly one-tenth of academic estimates
- No degradation in out-of-sample periods (1991-2013)
Long-term reversal
Jegadeesh and Titman (1993) found that momentum profits partially reverse: the 6-month strategy’s cumulative return peaks at 9.5% at month 12 but loses more than half over the next 24 months. This coexistence of medium-term momentum and longer-term reversal challenges simple overreaction or underreaction stories.
Interaction with value
Momentum and value are negatively correlated (approximately -0.6 within equity markets). This makes them natural complements:
- A 60/40 HML/UMD combination reduces maximum drawdown from -77% (momentum alone) or -43% (value alone) to -30%
- Asness et al. (2013) find the 50/50 combination across all asset classes produces a Sharpe ratio of 1.42
Crash risk
Momentum’s main vulnerability is crash risk: sharp reversals following bear markets (e.g., 2009). These crashes occur when the strategy is short beaten-down stocks that suddenly rally. The combination with value substantially mitigates this risk since value tends to perform well precisely when momentum crashes.
Global factor structure
Asness et al. (2013) find strong common factor structure in momentum returns across eight asset classes. Momentum strategies are positively correlated across markets (0.65 within equities, 0.37 cross-asset). Global funding liquidity risk partially explains these patterns: momentum is positively exposed to liquidity shocks.
Theoretical explanations
Both behavioral and risk-based theories exist:
- Behavioral: Underreaction to firm-specific information (Jegadeesh and Titman), disposition effect, investor inattention
- Risk-based: Cash-flow risk, discount-rate risk, funding liquidity risk