Gators In KenPom & Net [Update at #42]

FSU trying to tank our KenPom numbers struggling with Tarleton State.

 

Well FSU pulled away in the 2nd half and won by 20. So the Gator’s KenPom numbers are safe for now

Another team that Florida plays had a good win last night. UK beat Duke on a neutral court in Atlanta. That should help UF’s SOS.

 

I do worry a bit about our computer numbers OOC this year, but Golden and his staff are pretty savvy with this stuff. UNC seems to be a safe bet for the tourney, but besides that you are just sort of hoping Wake and Virginia make it and UVA probably isn't.

 

Not worried at all. Has everyone seen our conference schedule?

 

Kentucky now 20th in KenPom, and Florida down to 21st. Virtually identical ratings.

 

Regardless of liking or hating UK, their computer rankings are way too low after they beat Duke on a neutral court. Duke hardly moved at all in polls.
Mark Pope doesn't have the same talent on his roster that Calimari had year in and year out, but he's so much better at basketball Xs and Os. The SEC is gonna be extremely tough this season.

 

I believe KenPom and most of the other models are trained on preseason factors at the beginning of the season. As more game data rolls in, these preseason factors are slowly phased out. If the Cats keep playing like that, their computer rankings will definitely rise, but no model wants any single game outcome to determine too much of a team’s rating.

 

Yep. Some of the more honest ones admit that they start from last season's results. I would start with all 364 teams in a tie for top ranking and then let the chips fall where they may. After one game you'll have a legitimate ranking, and it will change radically over at least 10 games. It wouldn't be even remotely accurate for a while but it would be a hell of a lot of fun to watch. And the worst player on the worst team in America could one day tell his grandchildren, "I once played for the top-ranked team in the land," and it wouldn't be a lie.

 

That would be fun. Some of the modelers offer multiple versions of their ratings based on different plausible assumptions, so maybe you could even have both. Maybe KenPom Traditional vs KenPom Chaos. And then we could compare them until they converge.

 

Or, here’s a crazy idea, we could just wait until there’s enough data available for it to be useful.

 

Up to 17.

 

Win Tuesday and we'll probably shoot up to 17

 

we are back at 20. our opponent Tuesday, FAMU is currently at 355. We should be about a 40 PT favorite, then add a few more for home court advantage.

to get back to 17, we probably have to win by more than 50 or pitch a shut-out.

https://kenpom.com/

 

Which is why walk-ons don’t get as much play at the end of games anymore. 10 years ago if you were up 30 at the four minute time out the walk-ons would play the rest of the way. Now the walk-ons are lucky if they get a possession or two in a 30 point game.

 

Beating FAMU by 25 or 50 won't matter.

 

Coaches don’t clear benches because of KenPom ratings?

 

I don't like margin of victory as a factor in polling. At all. Some teams' style of play generally produces low-scoring and modest margins. A team that keeps winning that way shouldn't be penalized for it.

 

Well NET more so, but yes. If computers think you should win by 25 and you win by 22 you can drop. Some teams last year took advantage of this by hammering the really bad teams they played.

 

A good model can account for this. 70-40 is a stronger outcome than 100-60, and basically all models appreciate that. Recall when Florida led in the KenPom rankings in 2013. The Gators margin of victory wasn’t crazy because of their glacial pace (largely caused by a suffocating defense). Scoring ratio is closer than scoring margin to what matters, and that metric should be rather impervious to changes in style of play.