Heads, I win; tails, you lose

Requiem for the Super Bowl Indicator.

MarketWatch: In-depth global business coverage Devotees of the Super Bowl Indicator are smiling. That's because the Indicator is saying that, regardless of whether the Chicago Bears or the Indianapolis Colts win the big game today, the stock market is destined to go up in 2007. My advice: Don't bet on it. The Super Bowl Indicator is worthless. Whether the stock market goes up this year has nothing to do with which teams are in the Super Bowl or which of them wins. The Super Bowl Indicator, as many market followers know, holds that stocks will rise over the coming year if the winning team can trace its roots back to the original National Football League and fall if that team's roots are in the old American Football League. Followers claim that the Indicator has an 80% success rate, far better than most other popular stock market timing systems. The reason that followers of this Indicator can be so confident of a bullish outcome this year is that both of this year's teams trace their roots back to the original NFL. Before you rush out to throw more money into the stock market, note that this year is not the first in which the market was supposed to go up regardless of which team won the Super Bowl. The same situation existed in 2001, when the Super Bowl was played between the New York Giants and the Baltimore Ravens. The stock market fell 11% that year, as judged by the Dow Jones Wilshire 5000 index. Of course, no indicator works every year, so perhaps 2001's failure was merely an exception. The bigger problem with the Super Bowl Indicator, at least from a statistical point of view, is that all it has going for it is an 80% success rate. And while you might think that this is good enough to bet on it, you're wrong: Many phenomena that have absolutely nothing to do with each other nevertheless are highly correlated statistically. If you don't believe me, just consider my favorite example. It comes from David Leinweber, a visiting faculty member in Caltech's economics department. Several years ago, wanting to illustrate the perils of confusing correlation in the data with cause and effect, he searched through all the data on a United Nations CD-ROM to find the indicator with the most statistically significant correlation with the S&P 500. His discovery: Butter production in Bangladesh. Don't like that example? Try two more, courtesy of a recent academic working paper by Ben Jacobsen, a visiting financing professor at New Zealand's Massey University and Wessel Marquering of Amsterdam's Erasmus University. The researchers found that, when nevertheless employing apparently rigorously statistical tests, the stock market's seasonal patterns appear to be highly correlated with ice cream production and with airline travel. The lesson of these examples is that it is crucial to subject statistics to a reality check before betting on them: Is there a plausible explanation for why the two events are correlated a believable story, if you will? If one doesn't exist, then you should dismiss the apparent connection as bogus. Does such an explanation exist for the Super Bowl indicator? I'm not aware of any, and I'm not holding my breath that one can ever be found. MarketWatch: In-depth global business coverage