(09-09-2021 06:52 PM)nyc Wrote:
That's not how any of this works. Taking the average rating of teams moving from one conference to another is pointless. And he should know better.
Every rating system is schedule dependent. There's no way for it not to be. Because any statistic you use, you're posting that stat against an opponent. The conference schedule is 60% or more of a team's schedule. Change conference, change the schedule, change the ratings. It's that simple.
And his stuff is flawed because there is zero way for any system to account for the effect conference play has on every metric. If you build a system that's totally schedule neutral, it's going to tell you that the 30-5 Little Rock team that made the NCAA second round is the 5th best team in the country, when everyone believes they aren't quite that good; and it's going to tell you that a 15-17 Indiana team is terrible when everyone knows they're a top 80 program who just happens to play 18 games against other top 80 programs.
So KenPom builds in the conference expectations into the model, because if it consistently said Loyola or Stephen F Austin or Belmont were Top 10 teams... no one is subscribing for an inside edge in gambling.
Predictive measures have ZERO BUSINESS being involved in NCAA Tournament selection. When the NCAA let Greg Shaheen go, it was a BCS takeover of the NCAA Tournament process so that the Committee got to decide "Who was the best team" instead of being told to invite "who was the most deserving." And since the committee is 50% BCS members, it was just a money grab by the cartel.