Wednesday, March 7, 2007

ACC Tournament Predictions

The ACC tournament starts tomorrow, and many sites are awash with predictions. The best conference in the land is rife with parity, and this weekend will truly be Madness. One way to predict the tournament scientifically is to use the offensive and defensive efficiencies you've read about here all season. Ken Pomeroy has done just that, using the data from each team's entire season of work.

The principle is based on the Pythagorean Winning Percentage. To explain again, each team can be expected to win a certain percentage of games based on its offense and defense. In baseball this is done with runs scored/allowed; here we are using offensive and defensive efficiencies. I wanted to look at whether the results would be different if only ACC games were considered. That data is available here, thanks to the great work of Paul Rugani. After getting each team's Pythag WinPct, we can then predict how team A would fare against team B: (A-AB)/(A+B-2AB). Then based on potential matchups in the ACC tournament, we can calculate the chances that each team reach each round:




Again, this is based entirely on ACC games. Duke's chances of winning the tournament are severely depressed compared to Pomeroy's prediction, due to Duke's relatively worse defense in ACC play. The increased chances of UVA, VT, and Duke compared to the rest of the bracket have to do with how long each team has to wait to play the overwhelming favorite- UNC.

This seems like a good place for my predictions. Maryland is one of the most under-seeded teams in this tournament, and has the best chances of knocking of UNC. I don't see any of the slower-pace, short-bench teams (like FSU, Clemson, or BC) upsetting UNC, so I'll pick Maryland to reach the finals. Duke, meanwhile, is similarly under-seeded, and could make quick work of NCSU before revenging themselves upon UVA. A matchup with either Virginia Tech or Georgia Tech likely sits next. If it's Georgia Tech, I like Duke- Virginia Tech is a much tougher game. Duke will have trouble against either UNC or Maryland in the final, and so I think whoever emerges from the 1-4 side of the bracket should win the tournament.

What I think these stats show more than anything is that Duke is not to be counted out of this tournament simply because of a low seed. Whatever happens, it will be fun to watch. Any other predictions?

1 comment:

Anonymous said...

I hope you pick your real bracket better than you do your ACC bracket, because Maryland is DONE. Here's hoping you didn't similarly jinx Duke...