August, not April, is the cruelest month — for stock investors, that is. Often August has been the best performing month, persuading investors that a hot August is more than a fluke. But there have been long stretches where the market’s return in August has been particularly painful.
Consider the Dow Jones Industrial Average DJIA, -0.14% going back to its creation in 1896. Over the next century, August was in first place for average performance, with the Dow producing an average gain of 1.8% for the month — more than four times the average return of 0.4% in all other months.
Yet for the past 35 years, since 1986, August has been the U.S. stock market’s worst month on average — worse even than September, whose reputation as a terrible month for stocks is widely known. The Dow’s average August return since 1986 is minus 0.67%, slightly worse than the minus 0.64% for September — and versus an average gain of 1.05% for the other months of the calendar.
The difference between August’s average return and that of other months is significant at the 95% confidence level that statisticians often use when determining if a pattern is genuine. In other words, it was statistically significant when August did much better than other months — and when it fell behind.
How can two opposites both be statistically significant? To understand the answer, we need to realize that statistical significance is a necessary, but not sufficient, condition for concluding that a pattern really exists. There also needs to be a plausible theory for why the pattern should exist in the first place. Yet no such plausible theory exists for August, as far as I know.
This discussion already amounts to a cautionary tale about the dangers of data mining. But investors’ appetite for slicing and dicing the data is insatiable. So I can’t ignore other alleged patterns involving August.
One such pattern is that August is a particularly poor month in midterm election years, such as this one. Another is that the odds are against August during economic recessions, which may apply to this year though we won’t know for sure until the recession graders at the National Bureau of Economic Research decide — which could take months. Regardless, neither of these alleged patterns is significant at traditional standards of statistical significance.
The bottom line? How the stock market behaves this month will have nothing to do with it being August.
Mark Hulbert is a regular contributor to MarketWatch. His Hulbert Ratings tracks investment newsletters that pay a flat fee to be audited. He can be reached at email@example.com