Thursday, July 14, 2016

Avoid the losers - a min/max strategy based on trends

GaveKal, “Most investors go about their job trying to identify ‘winners.’ But more often than not, investing is about avoiding losers. Like successful gamblers at the racing track, an investor’s starting point should be to eliminate the assets that do not stand a chance, and then spread the rest of one’s capital amongst the remainder.” 

On the surface this is impeccable logic; however, the question is whether managers have better skill at picking losers overs winners. We can marshal evidence that shows that this is not as easy as many would think. Many hedge funds have sold over-valued names and bought cheap only to have the pairs do the exact opposite of what was expected. Finding what to avoid may be as hard as picking the winners. Minimizing the maximum loss is never a bad strategy; however, it is the operational details that often make this difficult. 

This is one of the reasons for momentum and trend-following strategies as a simple way to employ the min/max strategy of avoiding losers. For a long-only trend-followers, the losers are avoided by not holding anything that is not going up. For the long-short managers, short losers can add to performance. For momentum strategies, the investor will hold the top names as measured by past total returns and short the names that had the poorest performance and avoiding the rest. In both cases, the approach is simple - follow what the consensus of capital is telling the market. Poor performers will continue to be poor. Obviously this cannot continue forever, but it can be a starting point for construction. 

A simple addition is to avoid those names trending lower and buy those trending higher is to sort based on some valuation criteria. Buy cheap up trends and sell expensive downtrends. At the very least, using a trend sort provides a good Bayesian foundation for any decision. If my priors are telling me to avoid, then only buy if there is new information that will offset current price evidence. 

No comments: