Stocks quoted in this article:
Think there's an edge somewhere handicapping NFL games? The so-called "Super Bowl Quants" got a relatively large degree of publicity ahead of the big game, for picking the last eight championships correctly against the spread.
But alas, all good things must come to an end... if you wanted the Pats to win, that is, for here's what they said ahead of the game:
The New England Patriots will win the Super Bowl by at least three points even though the New York Giants have the appeal of "a cocktail party stock," according to a quantitative money management firm that's correctly picked the team covering the point spread for eight consecutive years.
Analytic Investors LLC in Los Angeles has documented a tendency on the part of Super Bowl bettors to overestimate the chances of the team that rewarded them more during the regular season -- the team with the higher alpha, in investment parlance. In 2008, that was the favored Patriots, who lost to the Giants 17-14. This year, it's New York.
"Everyone thinks the Giants are rolling right now, a lot of people in my office even," said Matthew Robinson, a portfolio analyst for global and Japanese equities at Analytic and the author of this year's analysis. "They like the Giants, but they have faith in the model as well."
Now, does this theory make sense? I suppose. It's almost like the self-reinforcement cycle of a momentum stock, where happy buyers keep buying more until it fails. On the other hand, it almost sounds backwards. We're wired to abhor losses to a much greater degree than we celebrate wins. I'd think a system that finds hated, and ergo underrated, teams would work better. Of course, that would never happen in a Super Bowl, as both teams had to have generated positive alpha. But I digress.
I think the real point here is that they lose the whole concept of the random nature of a small sample size. You know who else went 8-0 picking high-profile sporting events? Paul the Octopus. He went undefeated in the 2010 World Cup.
Now, I doubt anyone would suggest Paul had any special insight into soccer, whereas the Super Bowl Quants possibly have created some alpha. But all they can really do is find an inefficiency in a point spread. And that inefficiency may very well exist. Betting lines get set by relatively irrational market forces, as the fans of one side may put outsized dollars at stake with their hearts, and not their minds. Stocks get overloved by fanboys, too. The difference is that a Super Bowl resolves itself one way or another when the game ends, whereas Apple (AAPL) or whatever can stay loved forever (or close to it, at least).
So, let's say that inefficiency exists in the Super Bowl. It's likely only 1-3 points, but let's just go crazy and say the point spread deviates by an enormous 6 points from "fair." That implies that you can pick the game at the published spread and have a roughly 70% chance of winning. So, let's say analytic investors somehow can pick Super Bowls with 70% certainty. You know the odds of getting eight in a row correct? About 5.75%. In other words, possible, but very improbable.
Now, of course, a 70% likelihood is incredibly unrealistic, but what about a more believable 60% win rate? The likelihood of getting eight straight correct drops to 1.7%.
But really, whatever the answer, it's just an example of randomness fooling whomever uncovered this quant. If 1,000 sports analysts picked the results of eight even bets, odds are that four of them will get all eight correct. If a reporter tracks down one of those four, he would probably have a story about his methodology, and would honestly believe he has uncovered something. Who wouldn't? But it's more likely a function of tiny sample size than any incredible system.
Disclaimer: The views represented on this blog are those of the individual authors only, and do not necessarily represent the views of Schaeffer's Investment Research.