The Ubiquitous Momentum Effect
Since Jegadeesh and Titman's (1993) groundbreaking study validated that "buying past winners and selling past losers" generates significant alpha, momentum investing has become one of the most persistent market anomalies. This cross-sectional momentum strategy contradicts the weak-form efficient market hypothesis while delivering robust returns across:
- Global equity markets (Rouwenhorst 1998)
- Mutual funds (Carhart 1997)
- Commodity futures (Miffre & Rallis 2007)
- Corporate bonds (Jostova et al. 2013)
- Cryptocurrencies (Liu et al. 2022)
As Asness et al. (2013) demonstrated, momentum proves "ubiquitous" when applied to US/UK/European stocks, government bonds, currencies, and commodities. This global persistence suggests either systematic risk compensation or deep-rooted behavioral biases.
👉 Discover how momentum strategies outperform in volatile markets
Three Momentum Strategy Variants
1. Cross-Sectional Momentum
The classic approach involves:
- Ranking assets by past 3-12 month returns (excluding the most recent month to avoid reversal effects)
- Going long top decile winners, shorting bottom decile losers
- Holding for 1-12 months
Key Insight: Lewellen (2002) identified three profit sources:
- Positive stock autocorrelation
- Negative cross-asset serial correlation
- Persistent differences in expected returns
2. Time-Series Momentum
Moskowitz et al. (2012) found that volatility-normalized past 12-month returns positively predict next-month returns for:
- 52 of 58 tested assets
- Sharpe ratios exceeding traditional cross-sectional momentum
This works because it captures:
- Positive inter-asset serial correlation
- Reduced factor exposure noise
3. Residual Momentum
Blitz et al. (2011) improved risk-adjusted returns by:
- Adjusting raw returns for Fama-French three factors
- Ranking stocks by residual returns
- Achieving 0.90 Sharpe ratio (vs. 0.45 for raw momentum)
Behavioral vs. Risk-Based Explanations
| Theory | Key Mechanism | Empirical Evidence |
|---|---|---|
| Behavioral | Investor underreaction/overreaction driven by: | - 52-week high effect (George & Hwang 2004) |
| - Overconfidence (Daniel et al. 1998) | - Industry momentum (Moskowitz & Grinblatt 1999) | |
| - Conservatism bias (Barberis et al. 1998) | - Style momentum (Chou et al. 2019) | |
| Risk-Based | Compensation for: | - Conditional factor exposures (Kelly et al. 2021) |
| - Growth rate risk (Johnson 2002) | - Factor momentum (Arnott et al. 2021) | |
| - Factor autocorrelation | - Volatility clustering |
Cutting-Edge Developments
Factor Momentum
Recent studies reveal:
- Factor portfolios exhibit stronger momentum than individual stocks (Gupta & Kelly 2019)
- Industry-neutral factors fully explain industry momentum (Arnott et al. 2021)
- Principal components of factor returns capture all industry momentum
Ehsani & Linnainmaa (2022) mathematically prove that:
- Cross-sectional stock momentum decomposes into factor momentum terms
- Pure factor-driven momentum coexists with short-term reversal
👉 Explore advanced momentum portfolio construction techniques
Practical Implementation Guide
Dynamic Momentum Enhancement
Daniel & Moskowitz (2016) overcame momentum crashes by:
- Scaling positions by conditional Sharpe ratio
Using GJR-GARCH to adjust for:
- High market volatility
- Post-crash rebounds
- Achieving 1.20 Sharpe ratio (vs. 0.68 for static momentum)
Optimal Parameterization
| Parameter | Recommendation | Rationale |
|---|---|---|
| Formation Period | 6-12 months | Avoids short-term reversal |
| Holding Period | 1-6 months | Captures intermediate-term persistence |
| Delay Period | 1 month | Skips most recent performance |
| Weighting | Volatility-scaled | Reduces crash risk |
FAQ Section
Q: Does momentum still work post-2008?
A: Yes - Ehsani & Linnainmaa (2022) show factor momentum remained robust through 2020, with risk-adjusted returns actually increasing after financial crises.
Q: How to handle momentum crashes?
A: Combine volatility scaling with defensive positioning when:
- VIX > 30
- 6-month market return < -10%
Q: What's the best asset class for momentum?
A: Commodities and currencies show highest risk-adjusted returns historically, but equity factor momentum offers the most theoretical clarity.
Q: How many factors should I include?
A: Arnott et al. (2021) found 3-5 principal components capture ~90% of explanatory power for US equities.
Conclusion
Three decades of research confirm momentum as:
- Universal - Works across assets, geographies, and time periods
- Enhanceable - Improved via residuals, volatility scaling, and factor integration
- Theoretically rich - Bridges behavioral finance and asset pricing theories
The next frontier involves:
- Disentangling firm-specific vs. factor-driven momentum
- Developing unified "meta-momentum" frameworks