The Super Bowl Indicator: A Lesson on Correlation Versus Causation
Years ago in my Investments course, I got my students excited about an attractive investment strategy. (Spoiler alert: If you know me, you know I don’t give investment advice, and so this was a set-up.) I told them about some fascinating research that was recently published in the prestigious Journal of Finance. Researchers investigated a market timing strategy for U.S. stocks. Each January, they concocted a model to predict whether the market, as measure by the S&P 500, would be up or down that year. If the model predicted up, then they’d invest in the market; if the model predicted down, then they’d short the market.
The researchers tested the model over 22 years. The model correctly predicted the direction of the market 20 out of 22 times, an incredible accuracy rate of 91 percent. Moreover, the average annual return when the market was signaling up was 15.2 percent, and when the market was signaling down was negative 10.9 percent. The results were statistically significant. Of course they had to be — this was published in the friggin’ Journal of Finance!
After whetting their appetite, I asked for a show of hands, to see who would invest in this strategy? But first, I warned them that I couldn’t reveal the nature of the strategy until they made their decision — in other words, until they (hypothetically) gave me their money. Invariably, I’d get a majority of hands shooting up.
The Big Reveal
Then the big reveal. The entire strategy was predicated on whether an original NFL (or subsequent NFC) team won the Super Bowl, in which case the market was predicted to go up; or whether an original AFL (or subsequent AFC) team won, in which case the market was predicted to go down. I would wait for the groans as the class realized they had been duped. Then I would poll them again — would they still invest? Amazingly, a few hands would again shoot up for investing in the strategy. Maybe they were just trying to be provocative. Maybe a few really believed in it.
Of course, the teaching moment was to talk about correlation versus causation. Just because two things are related doesn’t mean that one caused the other. This is particularly…