Tuesday, March 19, 2013

Who's Hot? And Does It Matter?

Dan Hanner has done the work of providing the adjusted Pythagorean winning percentage for almost all of the tourney teams over the last ten games. The idea is that this can show us who is hot heading into the tournament, and who is not.

It was easy enough for me to to rejigger the Bracket Generator technology and run 1000 simulations to see how using this different data would change the results for each team.


In the Midwest, the hot team is St. Louis and the cold team is Oklahoma State. As shown in the chart below,  St. Louis made the Sweet Sixteen 23.2% more of the time using Hanner's numbers than it did using Pomeroy's full-season numbers, and Oklahoma State made the Sweet Sixteen 14.6% fewer times. (These are raw percentages. In other words, if St. Louis made the Sweet Sixteen 50% of the time under the original simulation, it would have made the Sweet Sixteen 73.2% of the time in the new simulation. The number shown is the difference between the team's chance using the full-season numbers and the team's chance using the last 10 games.)


Gonzaga was the huge winner in the West, and overall. In fact, using just the last 10 games, Gonzaga is the clear number two favorite to win it all, just behind Louisville. New Mexico also makes big gains, particularly in the early rounds. Pitt and Arizona suffer mightily.


Michigan and Florida get killed in the South. Georgetown and Kansas rise.


The East doesn't have any changes quite as dramatic as the other regionals. Cal, Davidson, and Illinois would be much more likely to get to the second weekend, at the expense of Syracuse and Miami. The big story would be Davidson, which over its last ten games is playing like a legitimate Elite Eight contender. It's looking like pretty much anyone could come out of the bottom half of that bracket to earn the privilege of losing to Indiana.

I'm actually suspicious that this new data means much. After all, why disregard the first 25 games of the season? But it's fun to play around with.

Note: I updated the Midwest numbers to account for an input error regarding Michigan State. Same gist except that Michigan State is better now.

No comments:

Post a Comment