Sunday, 1 June 2014
Profiting from momentum strategies - Part 2
The previous post concerned momentum - a strategy of buying past winners and selling past losers. It discussed how this strategy does well on average, but on rare occasions (recent market downturns and high market volatility) does very poorly.
Another reason why momentum may perform poorly is because other investors are chasing the same strategy. One of the "behavioral" explanations for momentum is underreaction, and goes as follows. Suppose a company experiences good news, which increases its true value by 10%. However, the stock price may only rise by 6%, because (a) only some investors notice the news, due to limited attention - they follow hundreds of stocks, and cannot notice what happens to every single stock on a particular day, and/or (b) investors do notice the news, but have stubborn prior beliefs - an investor may have a long-standing belief that the stock is of low-quality, and may cling to this belief even after receiving the news.
Regardless of the explanation, momentum works. You should buy a company that has risen by 6% over the past six months, because its true value has increased by 10%, and so it may gain the final 4% over the next six months.
However, what if all investors think like that? Then, they may pile into a stock that has risen by 6%. This extra buying pressure causes it to rise another 5% - so that the total increase is now 11% and so it has overshot. How does a would-be momentum investor ensure that he hasn't bought a stock that has overshot?
Christopher Polk and Dong Lou of the London School of Economics study this question in an interesting paper entitled "Comomentum". The idea is as follows. If investors are trying to exploit momentum (causing stocks to overshoot), they would have piled into many past winners, and equivalent sold many past losers. Then, the stock returns of past winners will covary with each other (i.e. move up together), and similarly the stock returns of past losers will covary with each other. They introduce a new measure, comomentum, which is the abnormal correlations among past winners and past losers - the stocks the momentum trader will trade on. When comomentum is high, this suggests that lots of investors are piling into momentum trades, and so the trades are less profitable.
Indeed, they find that their comomentum measure significantly predicts the future profitability of momentum. The effects are economically large. When comomentum is in the top 20% of its range, the momentum strategy earns 10.4% lower returns in its first year than when it is in the bottom 20% of its range. Simply put, when comomentum is high, other investors are pursuing the momentum strategy. This strategy is now crowded, so you should get out.