Following Trends Across and Within Industries

Last week, I posted about two ways to improve a momentum strategy by using value and quality to screen out dangerous momentum stocks. It so happens that lots of the momentum high-fliers from 2013 were concentrated in several key industries, like social media and biotechnology. This concentration begs the question, how much of the momentum effect comes from riding the hottest industries?

One of the more interesting elements of factor/quantitative investing is the industry effect.  For any style/factor—like value or momentum—how much alpha is the result of just making industry bets? For example, any unconstrained value strategy over the past 50 years has been chronically underweight tech stocks (until very recently), and this underweight has been beneficial because tech has been one of the worst performing sectors. The ability to make industry bets—like the one against tech— can be a significant advantage even though it increases some traditional measures of risk like tracking error.

So what about momentum? Does the excess return from momentum investing come mostly from rotating between industries, or can you also benefit from buying stocks within each industry that have the best momentum relative to other companies in the same line of business.  

For this analysis, I use two different definitions: industry group, and industry. There are 24 industry groups and 68 different industries (you can even get more specific to sub-industry, but in many cases there aren’t enough companies in each sub-industry for this type of analysis).

Here are the results (return and volatility), by decile since 1964, for 3-calculations of momentum: raw 6-Month return (which would allow you to rotate between industries freely), and 6-Month return relative to both industry group and industry. These second two calculations would result in a portfolio that was more spread out across industries and looked more similar to the overall market[i].  

A few interesting points stand out. First, momentum works both within and across industries.  This means that you can do well by using momentum to determine your industry allocations, but you can also do well by using momentum to select stocks in the industries that you like. Second, the unconstrained, raw version of momentum provides the best overall returns, but requires a bumpier ride. The annualized volatility is much higher than the industry relative calculation.

The chart below shows the rolling 3-year excess return for high momentum strategies. As you can see, momentum investing is quite cyclical, and returns can be painful for years at a time. The period following 2009, for example, was one of the worst factor inversions we’ve ever seen.

Bottom line, the flexibility to rotate more between industry groups and industries has added a significant amount of alpha over the long term. Of course, you pay for this alpha by having to endure higher volatility. Either way, the trend is your friend more often than not.

[i] Specifically, these are percentiles calculated within different groups.  The percentiles which are relative to industry or industry group would result in portfolio with lower overall momentum, because you’d own some stocks that may not have great momentum compared to the market, but do have good momentum compared to other stocks in the same industry (think Telco stocks in 2013, or some other laggard).