Mark Hulbert: Why Wall Street is rooting for the Giants

CHAPEL HILL, N.C. (MarketWatch) � Can the Super Bowl help you make more money this year?

Yes.

But that profit won�t come from basing your investments on who wins this coming Sunday�s game � even if, per the veritable Super Bowl Predictor, the Giants win on Sunday and the market rises, or the Patriots win and the market falls.

On the contrary, Sunday�s game can help us become better investors only if we use it to understand how almost any data set is vulnerable to being tortured into saying something meaningless.

/quotes/zigman/627449 DJIA 12,425.23, -155.46, -1.24%

The Super Bowl Predictor, for those of you who haven�t heard of it, 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 impressive 79% success rate.

There�s less here than meets the eye, however � even if we overlook the Predictor�s spectacular failure the last time the Giants won the Super Bowl, in 2008. Far from rising, as it the indicator predicted, stocks turned in their worst year since the 1930s.

The problem with the Predictor�s 79% success rate: It is artificially inflated by the indicator�s especially good record in the late 1960s and early 1970s, which came before even the most diehard football fan ever dreamed of correlating the Super bowl winner with the stock market. The earliest mention in a mainstream publication of such a correlation, as far as I can determine, came in 1978 by the late Leonard Koppett, a sports writer for the New York Times.

Reuters New York Giants quarterback Eli Manning.

To properly judge the indicator�s success, we should be focusing only on its record since then. Otherwise, we would be guilty of what statisticians refer to as hindsight bias. To put that in football terms, it�s akin to Monday-morning quarterbacking.

Unfortunately for the indicator, its real-time success since 1978 has been less impressive � 68%, in fact. While that might still seem much better than a coin flip, it isn�t. A simple bet that the market would rise each year would have been right 76% of the time since 1978.

Another essential reality check that can help immunize us from being seduced by spurious correlations: Estimate how many others were tested during the process of �discovering� the one that now appears to be so impressive. We should never forget that, even at the 95% confidence level that statisticians use to conclude that a pattern is genuine, there will be five �false positives� out of every 100 apparent correlations.

This is a crucial point, given the millions of investors who constantly analyze the market�s every squiggle and are correlating those squiggles with everything from sunspots to tea leaves.

To give one example: Consider another sports-related predictor, which also was popular in the 1970s. According to it, if the American League won the World Series in a presidential election year, then the Republican candidate would win the presidency, while the Democrat would win if the National League team won. Based on its �uncanny� success rate from 1952 through 1976, many began singing this indicator�s praises.

It failed in 1980, however, and its record since then has been statistically indistinguishable from 50:50. Now hardly anyone but the old-timers have even heard of it.

Click to Play Why market likes the Giants

For 36 of the 45 Super Bowls, the stock market has gone up after a win by an "original" National Football League team, and gone down when they lose. Does that mean everyone should root for the Giants? (Photo: Getty Images)

To be sure, hope springs eternal, and some believers in the Super Bowl Predictor will undoubtedly continue to believe in it � my calm, reasoned prose notwithstanding. They will find apparently strong support for this belief in an article that appeared in the Spring 2010 issue of the Journal of Investing, written by George Kester, a finance professor at Washington & Lee University.

Kester measured the returns of a portfolio that switched between the market and cash according to the Super Bowl Predictor, and found that even in real time since the indicator was first publicized, this portfolio has beaten a buy-and-hold. So even if it�s not statistically significant, it�s economically significant.

Note carefully, however: Kester does not endorse the indicator. In an interview Monday afternoon, he told me that he thinks it�s ridiculous for anyone to do so. On the contrary, one of his goals in conducting the study was to show how difficult it is to beat a buy-and-hold.

Is Kester nevertheless hoping that the Giants will win on Sunday, I asked him? Laughing, he said �yes,� acknowledging that maybe even he was unconsciously believing that a win by an old NFL team would be better for the stock market than a win by an old AFL team.

Click here to learn more about the Hulbert Financial Digest.