Daniel, Kent D., and Tobias J. Moskowitz. “Momentum Crashes.” (2013)

Abstract

Across numerous asset classes, momentum strategies have historically generated high Sharpe ratios and strong positive alphas relative to standard asset pricing models.

However, the returns to momentum strategies are negatively skewed: they experience infrequent but strong and persistent strings of negative returns.

These momentum crashes are partly forecastable. They occur in what we term “panic” states – following market declines and when market volatility is high, and are contemporaneous with market “rebounds.”

We show that the low exante expected returns in panic states result from a conditionally high premium attached to the option-like payoffs of past losers.

An implementable dynamic momentum strategy based on forecasts of each momentum strategy’s mean and variance generates an unconditional Sharpe ratio approximately double that of the static momentum strategy.

Further, we show that momentum returns in panic states are correlated with, but not explained by, volatility risk. These results are robust across eight different markets and asset classes and multiple time periods.