Seasonality, However You Slice It, Isn't Predictive

We think there are a lot of good reasons to be bullish on stocks this year. However, this isn’t one of them: the fact US stocks are on course to finish 2019’s first two months nicely positive. A tweet that went quasi-viral this week suggests otherwise, pointing to the historically strong returns in years where January and February each posted positive gains. An interesting observation! But nothing more, alas. This newfangled twist on the January Effect, which posits full-year returns echo January’s, shares all the flaws typical of seasonal adages.

We don’t dispute that a good January and February have historically accompanied a good year. The data are what they are. But arguing this is predictive strikes us as a bridge too far. Like all seasonal saws, it presumes past performance foretells future returns, which is a statement that just about every investment disclosure and financial regulator disagrees with. If you believe a certain month (or whatever) predicts full-year returns, then you believe stocks are backward-looking and not at all efficient. To the extent any calendar-based forecasting tool ever had any predictive power, efficient markets priced it eons ago, sapping any potential advantage.

Patterns are fickle friends—they always hold until they don’t. The January Effect works often but not always. Ditto for Sell in May, which would have you sell when May starts and sit out for six months. September has the lowest average return of any month, and December is positive far more often than not. But how do you know in advance which trick will work in which year? What if a bad September comes in a positive Sell in May window—what then? What if January is wrong? What if what if what if?

For fun, we did a high-level test of the big four seasonal saws since good S&P 500 data begin. For simplicity, we said the January Effect worked if January and full-year returns went the same direction; Sell in May worked if the six-month stretch from April 30 through Halloween was negative; September stunk if stocks fell; and the Santa Rally worked if December was up. As you will see, it was all over the map. We particularly enjoy the 19 years when September Stinks but Sell in May doesn’t work (there were only three years when the reverse happened). Almost as amusing? The 31 years when a Santa Rally followed a Stinky September, proving the folly of trying to trade around any of this.

Seasonal patterns are fun trivia. Knowing the average 10-month return and frequency of gains after January and February are positive might help you at Finance Pub Trivia, and if you know of a place that actually holds Finance Pub Trivia, do drop us a line through the feedback link at the bottom of this article. But aside from trivia and fun tweets, these fun factoids strike us as not terribly useful.

Exhibit 1: A Lite Brite of Seasonal Patterns

Source: Global Financial Data, Inc., as of 2/27/2019. Based on S&P 500 total return index, monthly, January 1926 – December 2018