Thursday, June 5, 2008

An interesting concept, somewhat marred by an essential flaw

Recent list work: #5 (done), #9 (done), #32, #42 (done), #44 (done), #69, #70, #83, #85 (done).

Slate has an interesting article about color/racial bias potentials in professional athletics. It discusses some cases of white athletes getting higher pay than black athletes with comparable stats, whether that's economically justified based on the effects racial makeup of the team has on ticket sales, whether baseball umpires alter their calls on ball v strike based on the race of the batter/pitcher (and how omitting an important variable such as time of day can lead noise to be mistaken for signal), etc.

But there's a big flaw in the article.

The hook of the article is that the Celtics are going to be playing the Lakers in the 2008 NBA finals. On paper, the Celtics are a better team--better win/loss record, better point differential, long winning streaks against teams the Lakers themselves were playing. The Vegas oddsmakers have, on the other hand, favored the Lakers on the point spread. The article attempts to discuss whether Vegas has caught on to referee racial bias which tends to operate on the whole in favor of white players. Most likely no, but not for any reason discussed in the article. Fundamentally, oddsmakers are not really concerned with whether team A will beat team B by 5 points. They are concerned with whether 50% of the betting dollars will be on team A if you give team B an artificial boost of 5 points. Gambling establishments essentially want an equal amount of money on each possible outcome, so they can minimize their risks and make their profit from the house taking whatever cut it takes. It frequently amazes me how many people don't recognize this distinction.