Thursday, March 21, 2013
Nate Silver and the NCAA Tournament
In class we discussed Nate Silver and his incredibly accurate use of statistics in predicting the presidential and congressional races in the past. For the past two years he has used statistics to predict the winner of the NCAA basketball tournament. Last year he was successful in predicting Kentucky as the winner of the tournament, giving them a 27% chance of winning. This year he predicts Louisville as the winner with a 23% chance of taking it all. He makes his predictions using a three step process. First he creates a power rating for each team, which is an aggregation of four computer ratings and two human ratings. Each of the ratings he uses has performed well in the past. Next, the ratings are adjusted for injuries, geographical location of the game, and and the team's performance in the tournament after the first round. Finally, he compares two teams' power ratings and estimates the likelihood that they win and move on to the next round. Our Lobos had a .8% chance of winning the tournament.
Subscribe to:
Post Comments (Atom)
Well, as our Lobos lost last night, our 0.8% chance has now been removed from the equation (Harvard...). I assume that Nate Silver's predictions increase in accuracy as more teams have been forced out of the tournament, so it would be interesting if we could see his predictions change over time (Which you can at http://www.nytimes.com/interactive/2013/03/18/sports/ncaabasketball/nate-bracket.html?_r=0). I do find it interesting that there have already been a large amount of upsets even in the second round alone (Kansas State and La Salle *though it isn't over yet*), I wonder if it is statistically significant?
ReplyDelete