KP ratings 11/27

[stever20]

I always find it interesting to look at ratings vs how season started...

for this week:
started/now
Nova 1/2
X 26/21
SH 28/25
PC 25/29
Cre 45/33
SJ 51/48
Marq 49/52
But 42/57
Geo 80/70
Dep 115/125
league avg 46.2/46.2
really not many major changes. Creighton 12 spots better biggest jump up. Butler 15 spots behind biggest drop.

[Edrick]

They are not separate metrics, a team's current rating is a function of their preseason rating, the number of games they've played, and the relative efficiency rating of the games. I forget exactly the number, but lets say its 20 games where the season becomes independent. That would be mean with every game up to 20, 5% of the weight from the preseason goes away and is replaced game-over-game with the actual adj EM. The whole idea is to have a 'whole season' in the numbers at all times.

That said, we are still more or less 2/3 preseason rating. Its kind of like looking at RPI and RPI SOS as distinct metrics. It doesn't really make sense because one is a function of the other. You are just talking colinearity then (for you stat nerds out there)

[stever20]

They are not separate metrics, a team's current rating is a function of their preseason rating, the number of games they've played, and the relative efficiency rating of the games. I forget exactly the number, but lets say its 20 games where the season becomes independent. That would be mean with every game up to 20, 5% of the weight from the preseason goes away and is replaced game-over-game with the actual adj EM. The whole idea is to have a 'whole season' in the numbers at all times.

That said, we are still more or less 2/3 preseason rating. Its kind of like looking at RPI and RPI SOS as distinct metrics. It doesn't really make sense because one is a function of the other. You are just talking colinearity then (for you stat nerds out there)

wouldn't that though mean for a team that has a much different rating(going either way)- that the true impact will be greater? I mean if Butler was 42 preseason and is 57 now, and if the current rating is only about 35% related to this year, wouldn't that 35% have to be much worse than 57 to get the rating to equal 57? The 42 is going to be factored in less and less. And what's happened so far will be factored in more and more?