Momentum Resembles a Wild Horse
"The Premier Anomaly"
Many investors know that momentum delivered exceptional returns in 2024. By one measure -Goldman Sachs’ U.S. HiLo Momentum Basket, high-momentum stocks outperformed low-momentum stocks by a staggering 42% last year.
What’s less well known is that momentum was also the dominant factor throughout the three decades leading up to 2000, with performance that was both consistent and persistent. In fact, in Dissecting Anomalies (2008), Fama and French identified momentum as THE strongest market anomaly.
Easy to Observe, Hard to Explain
If momentum is such a star factor, why isn’t it part of the Fama-French framework? Because momentum lacks a clear, risk-based explanation. The phenomenon of past winners continuing to outperform past losers directly challenges the efficient market hypothesis, as well as the core principle of classic asset pricing that expected returns are compensation for risk.
Defying theory, momentum proves equally elusive in the real world. It isn’t bound to a specific theme or style - it rides the waves of other factors. During the Fed-driven decline in 2022, Value and Low Volatility held up best, and momentum became heavily tilted toward them. In 2023, as the “Magnificent Seven” soared and Quality rebounded, momentum rotated. Most recently, with the rise of AI, momentum has loaded up Growth and AI-related stocks.
Even More Difficult to Implement
Momentum exhibits fat tails and is prone to sharp, sudden drawdowns, often referred to as “momentum crashes.” The chart below shows the distribution of daily factor returns over the past 20 years. Compared to the Quality factor, momentum displays significantly higher volatility, greater kurtosis, and more negative skewness.
Historically, when momentum 'crashes', the consequences can be severe. The 2022 study “Isolating Momentum Crashes” found that between June and August 1932, a momentum portfolio lost approximately 91%, followed by a second drawdown from April to July 1933. A more recent crash occurred in 2009, when momentum declined more than 73% in just three months. Even smaller episodes - such as those in 1938–1939, 1974–1975, and 2001–2002—included at least one month with losses exceeding 19%.
Ways to Saddle Break Momentum
More is More
A common approach to measuring momentum is the 12–1m trailing return. Trend-following managers supplement with technical indicators to refine timing and signal strength. But as this year’s CTA performance shows, this combination is still not sufficient to mitigate downside risk. While stocks are up 7.4% year-to-date, bonds 3.6%, and commodities 3.9%, the average trend follower is down 4.8%.
Momentum is dynamic, volatile and inherently risky - it can't be implemented naively. In our stock selection process, we use 12 distinct momentum signals to cross-validate one another. We’ve observed a strong amplified additive effect: stocks flagged by multiple signals tend to perform better on both the buy and sell sides, and with greater consistency over time.
Reining In Hidden Traits
Beneath its conceptual simplicity, momentum holds a nuanced complexity. Understanding its unique characteristics is essential before attempting implementation.
- A momentum crash is often triggered by a ‘junk rally’'.
During “normal” markets, negative momentum stocks tend to significantly underperform. However, occasional junk rallies can trigger sharp reversals, leading to momentum crashes. Effectively managing the behavior of negative momentum stocks is therefore crucial for successful momentum implementation.
- Losers are easier to identify than winners.
Typical momentum signals produce asymmetric return profiles: positive and neutral momentum stocks tend to perform similarly, while negative momentum stocks significantly underperform. Discovering ways to better identify true winners can meaningfully enhance the alpha potential of a momentum strategy.
- Momentum rankings do not produce monotonic returns.
The market doesn’t always favor the highest or lowest momentum rankings. In some cases, "the middle" of the distribution performs best, further complicating simplistic implementations.
Pay Attention to Trend Breaks
Momentum premiums are not permanent, and some research indicates that they dissipate within two years. As a result, paying attention to trend breaks is important in momentum implementation. In their 2023 and 2024 papers, Goulding, Harvey, and Mazzoleni found that divergence between short-term and long-term momentum helps identify momentum turning points, which are predictive of subsequent returns.
Summary
Momentum is a highly profitable stock selection factor, but successful implementation requires:
- A deep understanding of its behavior and the underlying characteristics.
- The skill and experience to manage downside risks and sharp drawdowns.
- The willingness to sacrifice some return potential in exchange for greater performance consistency.