internetter
Postdoc
Posts: 3400
Loc: Los Angeles
Reg: 11-21-04
|
11-15-17 01:36 PM - Post#236392
Given how volatile the various rankings are, as illustrated by one game, what significance are they?
Players' progress or lack thereof seems much more important.
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
11-15-17 02:13 PM - Post#236397
In response to internetter
They were of little consequence in the era before informed priors. Back when the ratings just floated (i.e. everyone started at the same place), it took a month or so before there was enough data to understand who you beat, how good they were and what that meant about how good you are.
With the preseason rankings solidly solving for much of that "unknown," the movement based on results early in the season is actually far more stable than it used to be. You'd still want to wait a few weeks if a team seems to be outperforming its preseason rank to see if that outperformance holds up over a longer period of time, but under the new systems, we can feel more confident about team quality earlier than we could, say, a decade ago.
|
cc66
Postdoc
Posts: 2204
Reg: 10-09-09
|
11-15-17 03:05 PM - Post#236403
In response to mrjames
I am not making particularly large claims for KP, especially after only two games. Nevertheless, Penn is our primary competition for the 4th spot in the IL tournament. All I was highlighting was that after just two games, the trend was noticeably in our direction.
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
11-15-17 07:25 PM - Post#236453
In response to cc66
I mean, I think what Columbia gets out of its freshmen could be the difference relative to Penn...
|
SomeGuy
Professor
Posts: 6415
Reg: 11-22-04
|
Re: rankings now 11-15-17 09:20 PM - Post#236486
In response to internetter
Does it have to be perfectly predictive in order to be of any use? It seems to me that things like kenpom do a remarkably good job of projecting outcomes over a number of games (though if i was going to be unfair and focus on one game, I'd point out that it was 2 points off of the differential in the Longwood game).
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
11-15-17 09:42 PM - Post#236495
In response to SomeGuy
All models are wrong. Some are useful.
|