Stats Made Easy

Practical Tools for Effective Experimentation

Sunday, October 01, 2006

Economists shave hairs on whether basketball games are fixed: Any bets on who wins?

In my March 26 blog I reported that 'forensic economist' Justin Wolfers, a professor at the University of Pennsylvania, inferred point-shaving from his statistical analysis of 44,120 NCAA Division I basketball games between 1989 and 2005. This new study by University of Illinois economist Dan Bernhardt disputes Wolfer's contention that statistics indicate point-shaving on college basketball. Perhaps it's only natural that superior teams fall short of expectations on their winning margin. According to Professor Bernhardt "the statistical properties that Wolfers identified in his paper seem to be intrinsic to the game of basketball itself, occurring independently of whether there are incentives to point shave, and are not indicative of an epidemic of gambling-related corruption."
It's good that this new analysis dissipates the cloud of suspicion about point-shaving raised by the first study.


  • At 5:44 AM, Blogger Wayne said…

    I had to weigh in on this one, with stuff you probably already know.

    "Combining these two observations, the researchers believe that a strong team is more likely to win by a little less than the spread than by a little more than the spread."

    The last sentence can be taken to confirm Wolfers conclusion. Favored teams fail to cover the spread more often than not.

    So, if strong, historical correlation exists, one can establish a predictive model, so basically you always bet against the favored team. Of course the problem here is the book makers change the point spread based upon how people are actually betting. The bookmakers always strive for a 50% for and 50% against bet structure, and they do a good job of it.

    So how to bookies make money? All they really do is move loser money to winners. What the article fails to mention is the practice of the "vigorish" (a.k.a. vig) - that little 10% fee the bookmaker adds. So betting $100 and winning at even odds will get you back $190.

    But I agree with the tone of the article; in this case I think it is more a matter of correlation than causation.


Post a Comment

<< Home