Clint Eastwood’s “Dirty Harry” character famously sneered that “a man’s got to know his limitations”. The same can be said of factor returns, which can be slippery beasts, especially in the short run.

Factor premia typically look compelling in long-term profiles, but results are subject to volatility and more than a trivial amount of noise in any given year — and sometimes over longer stretches. That alone isn’t a reason to ignore factor-based strategies, such as momentum- and small-cap-focused portfolios. But the capacity for sharp price swings (beyond what’s experienced in the broad market) is a reminder that understanding how these premia behave over short-to-medium periods is a key part of managing expectations (and thereby behavioral risk).

One item worthy of deeper study is mean reversion; that is, the tendency of returns to revert to a long-run average after extreme moves in the recent past. As a recent research note from Noah Rumpf, director of quantitative equity research at MFS Investment Management, demonstrates quantitatively, directional change is easier (but still not easy) to forecast when trailing performance is unusually high or low.

“While this does not necessarily mean that a factor that has underperformed will offer unusually strong performance going forward, it does mean that the factor has tended to revert to its long-term average performance,” he writes in “Value, Momentum and Mean Reversion in Factor Returns,” published in January.

To provide context for evaluating how mean reversion stacks up across several factors, Rumpf first lines up the premia based on top-bottom decile spread returns and volatility since the early 1960s, based on Professor Ken French’s data set. As the first chart below shows, risk premia aren’t created equally when viewed on a long-run trailing basis. For this sample, results vary widely, ranging from the high spread/high volatility profile of medium-term price momentum down to the comparatively tame dividend-yield factor.

(Click on image to enlarge)

The research note’s main takeaway is the tendency for mean reversion with respect to risk premia factors. For the analytics, Rumpf begins by measuring the calendar-year mean monthly return spread (long minus short) for every factor over trailing ten-year periods through each December 31. The results for the 13 factors under scrutiny are separated into three groups: the top-four performing factors (highest mean return), the middle five and bottom four (lowest mean).

Next, “we measure their historical mean monthly long-short returns on a forward-looking basis for five forward-looking time horizons: 1-year, 2-year, 3-year, 5-year, and 10-year, as shown in Exhibit 3.”

(Click on image to enlarge)

Note the tendency for the best-performing factors on a trailing basis to underperform in the short-term future (forward 1- and 2-year results). Meanwhile, the worst-performing group tends to bounce back in the near term. By the 5- and 10-year forward periods, a degree of equilibrium returns overall as the mean-reversion effect fades.

The lesson is that the restless rotation that resonates in the stock market overall also finds traction to a degree in factor risk premiums. “It makes sense to expect a reversion to long-term averages”, Rumpf concludes—an observation that finds support in the historical record.