Measuring Stock-Picking Skill

Deciding whether someone has skill in picking stocks that will give higher than average returns is a tricky business. You’d think that having a long-term track record of beating the market would be proof. However, some have found ways to argue that such records aren’t proof at all. I have my doubts about the arguments.

When investment managers have the ability to pick superior stocks, we call this alpha. If they beat the market averages by 2% per year, we say that they have an alpha of 2%. When we just invest in market index funds, we call the source of these returns beta. These returns come from putting your money at risk, but they don’t come from investment skill.

Complicating the situation is the existence of types of stocks that give superior returns. It’s well known that stock in small businesses and low-priced businesses have given superior returns over the long run. Such categories of stocks are called factors. These two examples are called the size factor and value factor. There are other lesser-known factors (e.g., momentum). Researchers are finding new possible factors all the time.

In the first chapter of their book The Incredible Shrinking Alpha, Larry Swedroe and Andrew Berkin argue that if we invest in factor stocks rather than just a regular index, the outperformance we get isn’t really alpha; it’s just another kind of beta. By this they mean that we’re not showing stock-picking skill; we’re just invested in a category of stocks know to perform better than others.

Swedroe and Berkin go on to use factors to show that investors with strong long-term investment records did it with various types of factor-based beta rather than using alpha. I have concerns about this type of argument.

My main concern is best illustrated by taking factors to an extreme. Suppose we invent so many fine-grained factors that each factor actually represents just a single stock instead of a broad class of stocks. Then by definition, alpha is impossible. Whatever stocks you pick, there are corresponding factors saying your returns are the due to beta rather than alpha.

Now, I’m not saying that factor research has gone this far, but there is no guarantee that any given factor will persist into the future. Suppose that in the next 50 years a given factor disappears because we were guilty of data mining or for some other reason. Past investors should be given credit for alpha when they recognized stocks covered by this phantom factor as undervalued.

Factor researchers work hard to avoid data-mining. They look for sensible reasons why a factor should exist in addition to just observing it’s outperformance in returns data. However, even 100 years of stock returns is only a modest amount of data. We can’t eliminate the possibility of data mining. There are a few factors that seem fairly solid, but a great many others are not.

When we examine investment records during periods of time long before the existence of a given factor was widely-known, declaring an investor’s performance to be “merely beta” seems like 20/20 hindsight. On the other hand, if a modern era investment manager used well-known factors to increase returns, we’re justified in saying any outperformance is the result of beta, not alpha.

Of course, the main point of the book that it’s not worth it to pursue alpha still stands. Alpha is scarce and trying to get it can be very expensive. I’m not a fan of venturing too far into the world of factors either. Pursuing factors increases investment costs. If the factors don’t outperform by as much as we hope, the net effect may be lower returns.