This past spring I asked why everyone isn’t beating the market when countless strategies have been shown to deliver outsized returns—at least in theory. I put it down to self-destructive behaviour, but now two US researchers have a different explanation: they’ve demonstrated that market-beating strategies tend to lose their mojo after they’ve been published.

In their paper, Does Academic Research Destroy Stock Return Predictability?, David McLean and Jeffrey Pontiff examine 82 characteristics that have been offered up in peer-reviewed journals to explain variations in stock returns. These include things like market cap, value measures, liquidity, dividend policy, momentum, credit downgrades, and many others. If an investor could use of any of these characteristics to pick market-beating stocks they would be called market anomalies.

Anomalies can come and go for various reasons, the researchers tell us. They may be the result of statistical biases: in other words, if you were to look at a different data sample the anomaly would probably disappear. For example, if stocks with a given characteristic delivered higher returns in the US but not in other countries, or only during a specific period, chances are there’s no real anomaly.

Anomalies may also be the result of mispricing. In this case, they are real, but as McLean and Pontiff write, “if publication draws the attention of sophisticated investors, then we might expect anomalies to disappear post-publication because of arbitrage.”

It works until doesn’t

Whatever the reason for the fleeting anomaly, investors who try to capitalize on it will end up disappointed. And that’s exactly what the researchers discovered. For each of the characteristics they examined, McLean and Pontiff conducted two statistical tests to see how well they predicted future stock returns. They found the average anomaly became significantly less meaningful after it was reported in the academic literature: “We estimate the average anomaly’s post-publication return decays by about 35%. Thus, an in-sample alpha of 5% is expected to decay to 3.25% post-publication.”

It should be noted that a 5% alpha is huge: most strategies designed to beat the market promise much more modest results. The value and small-cap premiums identified by Fama and French, for example, are much smaller: since 1927, large-cap value stocks outperformed the S&P 500 by 1.5% annually, while small caps managed an extra 2%. Reduce these by 35% and you’re not left with much outperformance at all.

Moreover, McLean and Pontiff’s paper was unable to account for the higher fees usually charged by funds that try to capture these anomalies, nor could they measure transaction costs and taxes. But McLean did comment on this important point later: “Investors should also keep in mind that these papers are written by researchers whose first priority is to better understand how financial markets work, and not to identify money making mechanisms. As a result, most studies do not estimate the costs of implementing the strategy, which can be substantial.”

I’ve made this argument before in regards to fundamental indexing, though it applies equally to any strategy that requires frequent portfolio turnover. It explains why ETFs that use strategies other than cap-weighting typically have higher tracking errors.

These findings are worth keeping in mind as the ETF world explodes with new “enhanced indexing” products designed to deliver higher returns than the broad market. Many of these are based on peer-reviewed research that may have identified legitimate market anomalies. But it seems likely many will not persist, or their alpha will evaporate after fees and costs are subtracted.