Skip to main content

Whatever you think about the stock market's August seasonal tendencies, you're probably wrong.

That's because most investors have the wrong belief about August's historical record. And even those few who have it right get tripped up when they try to put that belief into practice.

This discussion is important, because it provides an object lesson in how not to use statistics to improve your investment performance.

Let's start with Wall Street's conventional wisdom, which is that the stock market in August is a below-average performer. That's understandable, since August comes in the middle of the much-vaunted seasonally unfavorable period that goes by the name "Sell in May and Go Away."

However, August's below-average poor performance is a relic of recent decades only. Prior to 1955, the month far and away was the top average performer, as you can see from the accompanying chart.

Image placeholder title

Well, what about years in which the stock market has done particularly well through July -- like this year, for example? Perhaps August is a below-average performer in those years?

No. There is no statistically significant difference in August's average returns depending on how market's year-to-date return through the end of July.

What about the third years of a president's term, like this year? Since other research has found the stock market to be an above-average performer during those years, perhaps August in such years has notable tendencies?

Again the answer is no.

At this point you no doubt begin to detect that we're grasping at straws. And that's a problem, because that grasping inevitably leads us astray. If we search long enough, and torture the data hard enough, we no doubt will eventually "discover" a pattern that appears to be uncannily accurate. But it will stop working the moment we try to exploit it.

This problem is not limited to Wall Street, by the way. Many academics are guilty of the practice too. Campbell Harvey, a Duke University finance professor, several years ago devoted his presidential address to the American Finance Association to this problem. He proposed statistical steps to overcome it.

Suppose, however, that there were a solid scientific basis for believing that August had above-average or below-average seasonal tendencies. How should that change your investment behavior?

The right answer may very well be to still make no changes. That's because seasonal tendencies are nothing more than probabilities, and the only way to confidently translate probabilities into profits is to make numerous bets. So even when a month's seasonal tendency satisfies tests of statistical significance, you still have only minimally increased odds of success when betting on just one month alone.

Your situation in this regard is not dissimilar to the odds facing a blackjack player. A good card counter, you can increase your odds of winning a given hand to slightly more than 50% -- to 54% or 55%, I'm told. While that's statistically significant, notice that your odds of winning when playing just one hand are only slightly better than flipping a coin. Only if you play many hands (at least 30 is a good rule of thumb) do your odds become decent enough to be confident that you will be an eventual winner.

The bottom line? There is no objective reason to believe that the stock market in the coming month will perform particularly poorly (or well) solely because it is August. The best way August's alleged seasonal tendencies can improve your performance is by reminding you of the ways in which statistics can lead you astray.