If you’re busily filling out your brackets for the office NCAA Tournament pool, you might want to look at Georgia Tech.
No, it’s not a favorite for the Final Four or even in the tournament. But it does have a computer that’s able to calculate the odds of who might win.
The computer model, called the Logistic Regression Markov Chain, crunches data that includes home court advantage, previous scores, teams competing and margin of victory. It also uses statistical methods to determine potential upsets.
Based on the latest info, LRMC predicts this year’s Final Four match-ups as Florida vs. Indiana and Gonzaga vs. Louisville. Florida is favored to beat Gonzaga for the championship.
The record of the computer system is pretty good: it correctly predicted the winners of 51 out of 64 NCAA games last year, and in 2008, it even called some upsets accurately. You can see the entire set of brackets predicted here.
The computer was developed by Georgia Tech professors Joel Sokol, Paul Kvam and George Nemhauser, with assistance from Mark Brown, a math professor at City College of New York.
While the LRMC has been more accurate than chance and better than a lot of betting sites, its predictions are no guarantee. As Nate Silver, of the FiveThirtyEight blog at the New York Times has noted, if you have a 75 percent chance of winning a poker hand, there’s still a 25 percent chance you’ll lose.
Silver himself has a set of percentages for the winner of the NCAA Tournament, which is bigger than it has ever been with 68 teams. For his part, Silver has Louisville as a favorite, with a 23 percent chance of winning, followed by Indiana at 18 percent and Florida at 13 percent. On his blog he noted that the relative parity of the teams this year means there isn’t any single heavy favorite — and that increases the chances a given pick will be wrong.
Update: an earlier version of this story had links to last year’s LRMC predictions. Ohio State will still do well, but won’t win.
via Georgia Tech
Photo: Tipoff between Duke University and the University of Virginia, January 12, 2012. Credit: Wikimedia Commons / D. Myles Cullen