January Effect is a phrase many investors know, the possible phenomenon that stocks tend to perform better than normal during the first month of the year. The concept has been debated widely for decades, with different caveats proposed around small-cap stocks or large caps, wash-sale rules, liquidity, and trading costs. A quick summary of the research suggests that any trading advantages January might provide are at best slight and fleeting.
A related concept is the January Barometer, a claim that a positive first month indicates a positive full year to come, and vice versa. This barometer, too, is more myth than reality. But together these dubious phenomena raise a more interesting question: Which month actually serves as the best bellwether for investors? Does a specific month stand out as the most predictive of yearly returns? And are there months with no predictive power at all?
Looking at more than 85 years of data on the S&P 500 allows easy comparisons in the correlation between monthly returns and annual returns. January performance has a correlation of .30 to annual returns over that span—making it a very mediocre month for predicting performance, ranked six out of 12. So much for the “January Effect.”
What stands out even more, however, is the complete irrelevance of February returns. The correlation is basically zero and, in fact, slightly negative. February should be considered a throwaway month as far as any predictions, patterns, and relationships go.
The summer correlations are also notably weak: July and August are notorious for low trading volumes, Wall Street vacations, and a lack of interest in market movement. That’s all proven in the data here, as those two months have no bearing on the movements of the full year. Following summer vacation, however, is the most correlated month: September. At .58, it is the month most correlated to annual stock market returns.
Now turning our attention to small-cap stocks, which tend to have a higher percentage of individual investors, we see some similar patterns. Using the full 35 years of data on Russell 2000 performance, we see again that January is an average month in terms of yearly predictive power:
The irrelevance of February is similar—again, the month proves slightly negative. March is nearly as useless. And once again, September stands out as a highly correlated month.
Looking at both these indices, the effect of January is mild at best, and the predictive powers of September might come too late in the year to prove much use for the calendar-inclined investor. If you want to gauge annual stock market returns using a single month, the earliest month that seems to matter in any reliable way is April.
