Monday, October 21, 2013

Solving the Wisconsin Problem

The big news among college hoops nerds today is that Ken Pomeroy has changed his formula to deemphasize the effect of blowouts over bad teams:
The result is that games perceived by the system as big upsets get the most weight, while the influence of expected lopsided wins is minimized. For instance, last season’s non-conference games involving Grambling would be largely ignored. Whether a team beat the Tigers by 30 or 60 would make little difference in its rating.
This issue of how to treat non-conference blowouts is sometimes called "the Wisconsin problem," because it is one of the explanations for why the Pomeroy system always seems to overrate the Badgers. (Other explanations for this phenomenon include: the Badgers are usually actually better than people are willing to believe.)

An interesting aspect of Kenpom 2.0 is that he's gone back and retroactively changed all the ratings to from 2003 to present to reflect his new formula. Since my recent project of producing a model-based preseason top-40 relied heavily on historical Kenpom data, I decided to go ahead and rerun the numbers with the new data.

No surprise, Wisconsin drops. In my most recent update to the Top 40, Wisconsin had been projected 9th. Using the new Kenpom data, they drop six spots to 15th. (I will do another post with the "final" top 40 right before the season—I'm still tweaking it with more info about injuries, transfers, fixing mistakes, etc.)

I also put the new data into my model for predicting the final Big Ten standings, and they are now projected to win 0.5 fewer games. (My previous post on the Big Ten projections is actually obsolete because my Wisconsin data originally double-counted Gasser's return. After fixing that, Wisconsin was projected to win 12.1 games; now they're projected to win 11.6 games and tie with Michigan for third place.)

Wisconsin is by no means the only team affected. Here are the upper-level teams whose projected ranking changed the most:

Belmont: -10
Indiana: -9
Syracuse: -8
Marquette: -8
Iowa State: -8
Kansas: -7
Alabama: -7

Akron: +10
Harvard: +6
Oklahoma St.: +6
No. Dak. St.: +6
Boise St.: +6

As you can see, most of the "winners" are mid-majors and most of the "losers" (with the notable exception of Belmont) are major conference teams who probably had their previous kenpom ratings juiced by non-con romps over really bad teams.

No comments:

Post a Comment