2017 has seen equity markets steadily moving higher, volatility remaining at historically low levels, and trend-following managed futures strategies continuing to languish. Amid this backdrop, investors naturally ask if this has created a buying opportunity, with the expectation that markets, volatility, and trend-following should revert to longer-term averages. This is a reasonable question and expectation, but what really matters is whether investors can predict when this long-awaited mean-reversion will occur. Conventional wisdom says they can’t, but according to a recent article at ValueWalk, it appears the author believes timing managed futures allocations may be possible. In a recent whitepaper, they address “some of the potential benefits, challenges and opportunity costs” seen for investors who are “seeking to time managed futures allocations.”
They approach the question of timing by showing statistical evidence that managed futures strategies tend to mean revert. To show mean reversion they compare 12-month trailing Sharpe ratios to subsequent 12-month forward Sharpe Ratios, as is shown in the following scatter-plot.
The trend line of this plot is downward sloping, indicating some level of mean reversion. However, to my eye the scatter plot looks awfully random even if the trend line is sloping downward. Since I don’t tend to trust optics when conducting statistical analysis, I attempted to replicate their results to gain more statistical insight.
As can be seen, I was able to create a very similar looking plot with a very similar looking trend line using data from January 2000 thru September 2017 (I was unable to find return data prior to 2000 when searching the Societe Generale website or via Bloomberg). Additionally, I added one key piece of information: R2. For this analysis, this is a very important statistic as it tells the analyst “how well the regression line approximates the real data points.” In this case, a value of 0.018 indicates that the regression (trend) line is very ineffective in approximating the data points. Translation: There is very little to no statistical relationship between the last 12-month Sharpe ratio and the next 12-month Sharpe ratio.
The point of my comments is not to disagree that managed futures strategies tend to mean-revert. In fact, I believe they do mean-revert, as do most investment strategies. The challenge is knowing when mean-reversion is most likely to happen. Is 12-months the right lookback? Is the cutoff for mean reversion a sharpe ratio of 0.5, 1.5, or 2.3? Is sharpe ratio even the right metric? Quantitative researchers spend their entire working lives trying to answer these questions and at best, they gain a slight edge by discerning approximations for these various, mostly random parameters.
So what is a prudent investor to do? Give up and buy an S&P 500 Index ETF? Certainly not, as it will likely mean-revert within hours of you doing so! The answers are strikingly simple, if you are a disciplined investor, mean-reversion investing is likely already embedded in your process. The arduous investment work was done when you determined your optimal portfolio; selecting strategies and managers you believe will provide value over the long-term (5, 10, 20+ years; not months, quarters or even years) and allocating the appropriate amount to each investment. Once this has been accomplished, you are taking advantage of the mean-reversion that is likely to occur across all your investments every time you rebalance your portfolio and stay the course with your original plan. You are forcing yourself to do what most investors find painstakingly difficult; buying more of your underperforming investments while selling what has performed best. In other words, doing what all attempt but few succeed at, buying low and selling high.