March 14, 2006
Basketball badness
Was part of your March Madness fandom a small, just-for-fun bet on your favorite Division I basketball team? Maybe a wager as a sign of support for the Husky men, the Husky women, or the Gonzaga Bulldogs?
Friendly wagers can be a pleasant, camaraderie-building diversion, and informal betting pools are generally legal in Washington. But nothing ruins the fun of friendly betting quite so fast as knowing that many college basketball games are fixed.
Before we go any further: I’m not suggesting that anything’s amiss with the NCAA finals and I’m certainly not talking about the Huskies or the Zags.
Unfortunately, quite a few NCAA Division I games are fixed. In about six percent of games with strong favorites, someone on the winning team lays back on just enough baskets to blow the point spread.
How do we know? Justin Wolfers, a clever University of Pennsylvania economist, put economic theory together with data on 44,120 basketball games and showed that strong teams win—but miss making the spread—too often to be accounted for by chance.
Why is basketball such a target for cheating? Bets on college games revolve around the “point spread.” Suppose Evil U. is a heavy, 12-point favorite. A bet on EU wins if EU wins by 13 points or more. If EU wins the game by 11, then the bet loses. So an EU player can “shave points” by missing a basket or two and still win the game. That’s the key: A basketball cheat can blow the spread without blowing the game.
Sports where you get odds on who wins the game instead of betting against the point spread, baseball and auto racing for example, are harder to fix. Athletes, especially college athletes, have big incentives to win. A college pitcher with a mediocre won-lost record isn’t headed for the majors, and even a few games might make a difference. In contrast, a basketball forward does little damage to his NBA prospects by missing an occasional basket (and the spread) in a runaway game.
Wolfers found that, overall, teams are likely to cover the spread almost exactly half the time. When he separated out those games where the spread was very large, so that a player could shave points without significant risk of losing the game, all of a sudden favorites were failing to cover the spread noticeably more than half the time.
Wolfers also shot down the alternate explanation that winning teams avoid running up the score to give the bench some playing time or just out of good sportsmanship. Good sportsmanship would mean easing off a little when you’re up six baskets and easing off more when you’re up ten, regardless of the spread. Instead, the data shows that whether the spread is large or very large, favorites miss covering by just a couple of points.
Now the vast majority of games are not fixed. Even the vast majority of games where the favorite fails to cover the spread are completely honest. While point shaving ought to be of greater interest to the FBI, it needn’t ruin the fun of your friendly office betting pool.
Wolfer’s work proves that games are being fixed. It doesn’t identify which games are crooked, but it does tell us to cast a suspicious eye on games where the favorite just misses the spread. Time for local law enforcement to do a little sleuthing.
