Reply to thread

Message: [QUOTE="Sherman20, post: 4888786, member: 6357"] Everyone should listen to Mike Waters' podcast interview with Ken Pomeroy to understand a bit more how KenPom works and what I think are some of the faults. I've been a big KenPom devotee, but I'm more skeptical of the ratings now and how much to take them as gospel. [URL unfurl="true"]https://www.syracuse.com/orangebasketball/2023/10/ken-pomeroy-explains-his-college-basketball-ratings-why-is-syracuse-at-no-105-podcast.html[/URL] His preseason ratings still play an important role in the algorithm even 10 games into the season!...And it sounds like the preseason ratings play [B]some[/B] role even [I][B]after[/B][/I] 10 games, which is crazy. Some other caveats about how Ken's algorithm spits out its pre-season rankings: —"It's hard for a team that has played poorly the last few years to jump up in the [pre-season] ratings simply from adding a bunch of transfers." —“Coaching changes on average are a negative. They have a more negative impact on the better programs." —Pomeroy estimated the change from Boeheim to Autry might have cost the Orange “10 to 15 spots or so.’' —"My [preseason] ratings are calibrated on past season ratings." —"The previous season ratings is probably the most important thing. It looks at five seasons worth of ratings, but the most recent season is the most important." [Which is why we keep dropping the past few seasons in KenPom] —"The NET rating essentially duplicates my ratings...My ratings should not really be used for NCAA tourney selection, except for in maybe very specific circumstances. If you're trying to measure accomplishment, my ratings are not the way to do it." —This was KenPom statement that really gobsmacked me: "There's actually not much difference between how a team is at the end of the season and how they are at the beginning of the season. Obviously some teams are an exception there. ...There isn't that much difference between beginning of the season quality of a team and end of the season." He acknowledges that there are exceptions every year. But, to me, this is a stunning statement and where I think previous season baselines give KenPom's ratings lots of continued statistical noise throughout a season that the algorithm can't sift out. I imagine most coaches would be stunned by that statement. [B]I just don't think his algorithm is all that good at discerning much of a difference between a 50th rated team and a 100th rated team or a 35th and a 65th or a 20th and a 40th. Yet the NCAA tournament selection committee uses a team's NET in a significant way. Yes, they say that Q1 and Q2 wins are the most important, but guess what? Whether a win is Q1 or Q2 is based almost wholly on the NET! So you have a self-reinforcing system[/B]. [/QUOTE]

Verification: What is a Syracuse fan's favorite color?