So a couple of days ago, Bloomberg’s Cameron Crise gave those looking for a reason to “sell in May” a … well … he gave them a reason. And that reason was China. You can read the entire post (which includes a walk down memory lane) here.
Frankly, the only reason we ran that piece was because of the China commentary. That is, we’re no fan of what we consider to be old wives’ tales and/or urban legends like “the January effect” and the most famous of them all “sell in May and go away.”
Now sure, there’s “evidence” to support those silly adages and indeed, the fact that there’s evidence means they might not be “silly” after all. But damned if we want to be the idiots that end up having to explain poor performance after the fact by admitting that we made an investment decision based on what amounts to a Farmers Almanac thesis and that decision ended up being a bad one.
Well anyway, Crise is back with more on the whole “sell in May and go away” cliche and based on the historical evidence he reviewed, US investors should exercise “caution” next month. More below…
I am a sucker for the seasonal and behavioral tendencies of financial markets, which offer evidence that life is not seasonally adjusted. Perhaps the most famous of these is the equity market admonition to “sell in May and go away.” But is there any way to say whether to stay, or stray, in May? I decided to take a look.
- I compiled a list of variables that we traditionally think of as drivers of economic and market performance. Among these were the real Fed funds rate, the 10y yield, payroll employment, and a rolling 12m forward expectation for S&P 500 earnings. I also looked at year-to-date performance of the SPX through April to capture the underlying market narrative and momentum.
- I calculated changes in the relevant variables and regressed them against both the change in the subsequent May alone and versus the seasonally fallow May-October period. The correlations were decent, with r-squareds of roughly 0.30 for both iterations of the model. Given the blunt construction of the framework (none of the inputs were optimized) and the randomness of monthly price action, this was a pretty solid result.
- The results suggest caution for U.S. equity investors next month, with a forecast return of -1.3%. Since 1991, there have been seven prior negative forecasts for May. The average return of those months has been -1.29% with a median of -0.91%. Four of the seven delivered negative returns.
- In contrast, when the model forecasts a positive return for May, the average monthly gain has been 1.48% with a median of 2.08%. Fifteen of the nineteen positive forecasts were followed by SPX gains in May.
- What about for the entirety of the May-October period? The model is more sanguine, forecasting a return of 3.96%. In prior years with a positive forecast, the average return from May to October was 3.7% with a median of 3.63%. 13 of the 18 years with positive forecast returns actually delivered them.
- When the model forecast is negative, subsequent market returns are far worse. The average return is 0.03% and the median is -0.70%. Five of the eight years with negative forecasts subsquently produced subzero returns.
- Whether investors choose to sell in May is, of course, their own choice. Based on this analysis, however, those who do may not wish to stay away the whole summer.