Editors' pick: Originally published Feb. 2.
What would you pay for an indicator that predicts the annual direction of the stock market with 76% accuracy? Most market pundits could only hope to be correct on the direction of the market over three-quarters of the time. Yet remarkably, the Super Bowl Indicator of the Stock Market has done just that over the past 50 years.
The Super Bowl Indicator was originally proposed somewhat tongue in cheek by New York Times sportswriter Leonard Koppett in 1978. While legend has it that the theory was memorialized in a New York Times article, no one has been able to locate it. The phenomenon was the subject of a 1978 column in, of all places, The Sporting News entitled "Carrying Statistics to Extremes." Every year, around this time of year, the Super Bowl Theory of the Stock Market is revisited.
What Koppett observed was that in 10 out of the 11 years the direction of the Dow Jones Industrial Average was "predicted" by the outcome of the Super Bowl. Specifically, if an old (pre-merger) NFL team won the Super Bowl the market closed up for the year and if an old AFL team won the Super Bowl, the market closed down for the year. When Koppett first proposed this relationship, the only time that it wasn't consistent was in 1970 when the Kansas City Chiefs won the Super Bowl and the Dow advanced a scant 4.8%.
Now, not only has the Super Bowl Indicator consistently "predicted" the direction of the market, but the average returns when the old NFL wins and when the AFL wins are dramatically different. The Dow has averaged a healthy 11.4% return in years when the old NFL wins and has declined by an average of 0.85% in years when the old AFL prevailed.
Now, pre-merger is important to note. The Pittsburgh Steelers are an old NFL team that won six Super Bowls after migrating to the American Football Conference when the NFL and AFL merged in 1970, and the market has advanced all six of those years by an average of 18.4%. Mr. Market is clearly a Steelers fan.
The Atlanta Falcons
Looking at history, there is a 70% probability that an old NFL team wins the Super Bowl and there is a 74% chance that the market rises in a year. Given these probabilities, by simple chance the Super Bowl Indicator should be correct about 60% of the time. This is estimated by calculating the statistical likelihood of the two circumstances in which the indicator will be correct: (1) NFL wins and market goes up; and (2) AFL wins and market goes down.
Even so, there is a tremendous statistical difference between 60% and 76%. Should equity investors be Atlanta Falcons' fans this weekend?
That said, the Super Bowl Indicator is an example of how some individuals can confuse the concepts of correlation and causality. Two series of data are correlated if they move together. However, simply because two series are correlated does not mean there is a cause-and-effect relationship between them, or more importantly, that the relationship will continue into the future. When two series are statistically correlated but there is no reason to infer causality, there is a false correlation.
If you look long and hard enough -- an exercise academics pejoratively refer to as data mining -- you will find data series that are highly correlated but have no causal relationship between them. One of my favorites: annual U.S. spending on science, space and technology correlates with suicides by hanging, strangulation and suffocation.
In the song Superstition, Stevie Wonder wrote "when you believe in things that you don't understand then you suffer." Anyone who thinks Tom Brady or Matt Ryan has anything to do with the direction of the stock market is likely to suffer. Enjoy the game.