|
|
|
|
|
Jens 'n' Frens
Idle thoughts of a relatively libertarian Republican in Cambridge, MA, and whomever he invites. Mostly political.
"A strong conviction that something must be done is the parent of many bad measures." -- Daniel Webster
|
|
|
|
|
Tuesday, November 30, 2010 :::
The election being over, Nate Silver is writing about college football:Since the beginning of the B.C.S. in 1998, the computer ratings have predicted the winners of B.C.S. bowl games correctly 32 times and incorrectly 20 times. That’s marginally better than the 30-22 record put up by the overall B.C.S. formula, using a combination of human polls and computer ratings, and only marginally worse than the 33-19 record the computers would have compiled if they were allowed to consider victory margin and home-field advantage. Teams favored by a composite of human polls are 30-21-1 in the same games. The teams considered the favorites by Las Vegas gamblers are also just 30-21-1.
The Associated Press preseason rankings, at 31-20 (excluding one case in which neither team was ranked at the beginning of the season), turn out to be about as reliable at picking the winners of B.C.S. bowl games as these end-of-the-year rankings are. When you're looking at a sample of 52 binary results, even if they are identically distributed, 33 and 30.5 are essentially the same. I'm curious how he ascertained what the computers "would have" predicted - some of the BCS computer ratings have alternative, non-crippled versions, but I don't think all of them do. I'm also curious to what degree the picks differ. I suspect a lot of games had consensus favorites, even where the favorites didn't end up winning.
I didn't give this a lot of thought before reading the piece, but I think the prior should have been that 52 games wouldn't suffice to provide a statistically significant result. I would have guessed ahead of time that the good computer algorithms might be a little bit better than the humans (i.e., that human bias would slightly outweigh the additional information human experts can incorporate) and both would be better than the BCS computer algorithms and probably the pre-season polls. But any remotely-plausible rankings will be fairly similar, and 52 games isn't much of a sample.Labels: sports
::: posted by Steven at 10:59 AM
|
|
|
|