Wow. The obtuseness of finance academics never fails to astonish me. First, when discussing the tendency of stocks to revert to the mean over a one-month period, they ignore the law of regression to the mean, which is a statistical law, not a tendency in stock returns. (It states, As long as there is a meaningful average of values, an extreme value will be more likely to become less extreme over time than to continue to be extreme.) Second, they do not discuss the variability of the five major asset classes. If they exhibit low variability, of course their results will be the same from month to month.
Let me illustrate this with an intuitive example. Let’s say you have five funds. One grows at a steady 1% per month. Another grows at a rate that varies between 0.25% and 0.5%. Another grows at a rate that varies between -2% and +3%. A fourth grows at a rate between 1.25% and 1.85%. And a fifth grows at a rate between -0.25% and 1%.
Obviously, the first fund is always going to beat the second and fifth fund, and the fourth fund is always going to beat all the other funds except, sometimes, the third. By the authors’ own measure, this set of funds is going to exhibit very strong short-term momentum!
Now this is an extreme example, and asset classes don’t behave like the funds above. But there are some similarities. Equity indices tend to outperform the other classes; currencies tend to underperform; equities and commodities are much more variable than government bonds and treasury bills; treasury bills are a subset of government bonds; and so on. A study of five asset classes that each show a great deal of persistence in their overall behavior cannot be compared to a study of thousands of stocks that all behave wildly, like stocks normally do.
To sum up in statistical terms, the law of regression to the mean is mitigated by time-series stationarity. The word “stationarity” does not appear once in the paper.