Reply to thread

Message

[QUOTE="businesslawyer, post: 3425588, member: 55"]
So that shows the problem with the KenPoms, and many similar ratings services.  They measure success possession by possession, but not by wins and losses.  Let's say we have a 2 game homestand against Temple and Tulsa and they are ranked similarly.  As I understand the KenPoms, the computer is largely indifferent to whether we go 2-O, winning both games by 5 points, or go 1-1, losing one game by 1 and winning the other by 11.  Now, rating those two sets of results similarly may make sense as a pure predictor of future results (although that assumes that winning isn't a skill separate from how you play over the entire game, which I think this year's team has totally disproven -- note that the KenPoms view our failure to win close games as bad luck rather than a problem with our performance), but in reality 2-0 is totally different from 1-1 over those two games.

I think it's absurd to rank teams based on how "good" they are, meaning how likely they are to beat a given team on a given court in the next game, without giving far more weight to whether they are getting it done based on wins and losses.
[/QUOTE]

Verification

First name of men's bb coach