The Super Bowl will take place on Sunday. This is the final game for American Football (if you have to ask, then: “No, not real football”). Not only is it a highlight in sports, it is also a financial highlight as it determines the fate of the US stock market for the year.

You can see some history of the indicator on the Wall Street Journal blog.

Testing the prediction

Even better, you can read my paper Permuting Super Bowl Theory.

The moral of the story is that the success rate of a prediction can be a deceiving statistic.  There are more specific morals though:

• success rates are especially deceiving if one choice predominates (this could be a problem for Groundhog Day predictions)
• be careful not to include in-sample observations
• be specific about what success means, and use the most useful objective
• consider how much data snooping was done to find this predictor
• wonder if it makes sense even if everything else checks out

Epilogue

The time you won your town the race
We chaired you through the market-place;

To an Athlete Dying Young by A. E. Housman