A Lesson In Minimum Volatility Gleaned At The Horse Races

by: Dan Morillo

At the Belmont Stakes horse race coming up June 9, I’ll Have Another will attempt to become the 12th horse to win the Triple Crown. Now, I’m not a huge fan of horse racing, but the economist in me is always intrigued by what sporting events can teach us about investor behavior. In particular, I’m interested not in which horse wins the race but in how those who follow the races place their bets. Let me explain why - and how it applies to minimum volatility portfolios.

In an earlier blog post, I argued that equity minimum volatility portfolios are potentially a good replacement for a traditional equity investment in a cap-weighted market portfolio. Key to this claim is the fact that a wide range of empirical studies show that minimum volatility portfolios have performed equally well, if not better, than the market but have done so with lower risk.

Still, no matter the strength of the historical observation, I am continually asked whether minimum volatility portfolios can continue to perform as they have in the past. One way of answering the question is to focus on what are the sources of the performance of minimum volatility portfolios and make an assessment of their likely persistence.

Today I want to discuss “behavioral” sources for the low-risk anomaly. “Behavioral” is a catch-all term that encompasses a wide range of theories of how people’s behavior appears to deviate from the theoretically ideal behavior when dealing with uncertainty.

My favorite example is what has been known since the 1950s as the “favorite/long-shot bias”[1]. This is the finding that people who bet on horse races bet on the long-shot horses more than would be justified by the actual odds of winning (and conversely bet less than they should on the favorite horses). The story generally goes like this: Horse-racing bettors have difficulty evaluating small differences in outcome probabilities, for example the difference between a 1-in-50 odds horse and a 1-in-100 odds horse. That’s a 2% probability of winning versus a 1% probability of winning. This results in a large proportion of bettors placing too high of a bet on the longer odds because the potential winnings appear larger in that case[2].

If I replace horses with stocks, you can see where this leads: Stocks with high beta (or high risk) may be “overbought” (leading to future underperformance) and vice-versa for low-beta (or low risk) stocks. While the various theories around this behavior can get complicated (as theory always does), there is a common theme: Investors have difficulty evaluating small differences in probabilities, particularly when large outcomes are possible. This results in an apparent preference for risky assets of various kinds, thus leading to their underperformance[3].

Why is this explanation interesting? It tells us that the “hurdle” for thinking that the low-risk effect may persist in the future could be similar to other factors that equity investors have already come to rely on, like momentum investing, which are tied to behavioral explanations.

There is an additional reason that the low-risk effect may persist. This is the fact that the widespread use of active-management benchmarks may induce money managers to prefer high-beta stocks even in the absence of any behavioral biases. More on this on my next post.