Pomeroy ratings question...

[Ptzburghcuse]

We are now what I would say a fairly significant way through the season - half way through the schedules of many teams and a start to the conference season at least.

I saw SU was hovering in the low 30's for some time after the back-to-back losses at home. I didn't look after the GT loss but did after the Duke..

SU is currently sitting at #35 - not too bad considering some of our losses and solid road wins. I am looking at some of the teams ahead of us and scratching my head at what it takes for some teams to drop in the rankings. For instance the following teams are currently ranked above SU:

#35 SU 12-5 Off 77 Def 13
#32 Indiana 12-5 Off 41 Def 33
#28 Texas 10-7!!! Off 47 Def 18
#24 Florida 9-7!!! Off 80 Def 8
#23 Louisville 11-5 Off 15 Def 61
#16 Wisconsin 11-6!! Off 30 Def 14
#13 Purdue 11-6!! Off 6 Def 70

I know strength of schedule and many other factors go into and it looks like offense is clearly valued over defense but just wondering what does it take for some of these teams to drop - eventually you have to win some games right - I mean 9-7?

 

[Cusefannotindc]

Wins and Losses don't matter in Pomeroy's model, it is a measure of efficiency. A loss by two points and a win by two points aren't going to look much different for your efficiency, but your efficiency is going to hold more predictive value than you W/L record.

A team's rank in offense or defense is not the same thing as their efficiency. I believe if you were to dump offense and defense efficiency of teams into a regression it comes out something like 48% offense/52% defense.

 

[malpearl31]

I don't think the actual W-L record really factors in to the kenpom rankings - I believe it's designed to be more predictive, based on what you've done and who you've done it against. (Except to the extent that the more often that you lose, the more that your points-per-possession are automatically less than your opponent's).

For Florida (to pick a random example), my guess would be that their performance against Butler on 12/29 is probably a huge driver of their current overall Efficiency Margin (ranking). That night, they held Butler - the 32nd best offense in the kenpom rankings and 41st ranked team overall - to 0.72 points-per-possession. And Florida put up 1.28 PPP themselves on the offensive end (Butler's defense is by no means elite, but that's still a pretty good offensive performance by the Gators).

SU doesn't have any performances to date that are that good (our season-best PPP differential is 0.47 in game #1 against Eastern Washington... and they're all the way down at 280th best in kenpom's rankings).

 

[moqui]

pomeroy rankings = offensive efficiency - defensive efficiency

 

[Brooky03]

pomeroy rankings = offensive efficiency - defensive efficiency

Also, Pomeroy rankings = poop.

But the math sounds fancy so it's treated like gold.

 

[kirbivore]

Also, Pomeroy rankings = poop.

But the math sounds fancy so it's treated like gold.

 

[Cusefan0307]

Also, Pomeroy rankings = poop.

But the math sounds fancy so it's treated like gold.

Nope. Best predictor of point spreads we have and the NET was modeled after it.

 

[Brooky03]

Nope. Best predictor of point sorrads we have and the NET was modeled after it.

I just died reading this post, i was so bored