Momentum crashes should be expected - Herds and information

One interesting feature of a momentum strategy across all asset classes is that it is subject to crashes or reversals. This should not be surprising given that momentum can be viewed as the manifestation of herding behavior by investors. The strong herding can be defined as crowded trades where flows are pushing prices to extremes. Herding when delinked from fundamentals is the basis for price bubbles; the more successful a momentum trend, the more likely there will be a reversal crash. However, the measurement of this reversal risk is difficult. Ideas for measuring oversold and overbought have been tested for decades with mixed results. Take away the crowd or provide a change information that adjusts herd expectations quickly, and the asset flowing into momentum moves in reverse.  

Some have argued that the potential for crashes is the reason for momentum compensation. There is a premium for taking on crash risk. It can also represent skew in the return distribution, or from the perspective of the trend-follower, a reversal of profits from the market high. 

Time series momentum investors should be ready for crashes. The simplest contingency for downside protection with time series momentum is a stop-loss. While the use of stop-losses may be suboptimal, it is the most direct method for stopping losses from a momentum crash. The stop-loss will limit downside regardless of momentum reversal size.  Similarly, cross-sectional momentum needs to be ready for shorts that rise and long that fall contrary to expectations; however, cross-sectional stop-losses deconstruct the momentum trade. 

Investors want to extract any momentum premium but also have a contingency for when momentum decides to reverse. Hence, there is significant value with measuring traded crowded, herd behavior, and information flow changes.