Price Momentum Crashes — Should We Worry?
November’s big negative return for price momentum was a reminder for investors to also hold exposures that are uncorrelated with momentum signals.
At its worst point in November, a simple 10x1 price momentum factor portfolio1 was down around 9%. It recovered into the end of November, ending up nearly flat for the month. And while a flat return may typically not be worth writing about, the two days that delivered the 9% drop deserve further scrutiny.
Going back to 1925, a Winners Minus Losers (WML) portfolio,2 whose results we use here to represent “price momentum,” had two disastrous return periods — one in 1932 and one in 2009. On both occasions, price momentum crashed during the huge rebound that followed a major negative stock market return. In both cases, the loss sustained by the price momentum portfolio during the crash (and rebound) did not recover for many, many years. See Figure 1.
Figure 1 : Theoretical Cumulative Growth of $1 Invested in WML (Momentum) Portfolio Since 1927
Monthly
Other historical occurrences of a monthly decline of 9% or worse for a price momentum portfolio, excluding the two episodes described above, show a maximum drawdown between 11% and 46% and an average of around two years to recover the loss.
The distribution of returns for price momentum is negatively skewed. This means, the worst negative returns are typically a lot bigger than the best positive returns. So the large negative drawdown over the two days in November is within expectations for the momentum signal.
Given the observed tendency for momentum to exhibit this negatively skewed characteristic, we believe it’s important to combine momentum with other, uncorrelated signals. Signals like value tend to have a positive skew and generally have low or negative correlation with momentum’s returns.
The Bottom Line
Because momentum quickly recovered to flat in November, we avoided a true price momentum crash, but the event is a cautionary tale for investors. Given the difficulty of predicting the timing of price momentum declines, it’s wise to invest using a diversified set of signals that contain complementary payoff profiles.
When AQE researches new signals to add to our models, we always look for uncorrelated signals. In this instance, based on near-term and long-term characteristics, we seek measures of sentiment other than price momentum — with different payoff profiles that optimize total portfolio performance.