kenpom OMG

[CusefanATL]

http://kenpom.com/

1 thing jumps out at me. per usual with kenny.

 

[Knicks411]

http://kenpom.com/

1 thing jumps out at me. per usual with kenny.

Haha yeah I mentioned that when the rankings first came out. Blind spot for the Badgers, I guess.

 

[HoustonCuse]

Haha yeah I mentioned that when the rankings first came out. Blind spot for the Badgers, I guess.

Maybe the whole Big 10. 4 of the top 5 are BiG teams, including one that is 0-1.

 

[IthacaMatt]

Yeah, how is a team that lost to someone who didn't even make his top 40 still ranked 5th?

 

[HOFCeluck]

It's not a poll.

 

[Knicks411]

It's not a poll.

Yeah, it's not going to react too much to one game, good or bad. Just looking at it, UConn was 45 coming into the game. Pomeroy has Michigan State as a 10 point favorite at home against Purdue in January; so if you use that as a rough proxy and then change it from a home court game to a neutral site game, Michigan State was probably expected to beat UConn by about 6 or 7. They lost by 4 instead; but it isn't the kind of result that is going to massively change their ranking. (They moved from 4 to 5)

That said, it does seem unlikely that 4 of the 5 best teams in the country are from the same conference.

 

[CusefanATL]

It's not a poll.

call it whatever you want, he considers it an indication of who is going to win a game.

wisconsin FINISHED the season in the top 5 last year. they lost 10 times.

 

[Knicks411]

call it whatever you want, he considers it an indication of who is going to win a game.

wisconsin FINISHED the season in the top 5 last year. they lost 10 times.

Wisconsin was pretty clearly a miss last year. They blew out a ton of really bad teams and that probably artificially boosted their rating. Good news is they are doing the same thing this year!

 

[Capt. Tuttle]

Maybe the whole Big 10. 4 of the top 5 are BiG teams, including one that is 0-1.

he is a Purdue guy, I believe.

Sent from my Vortex using Tapatalk 2

 

[Knicks411]

Don't think so. Via Wiki, so take it for what its worth

Pomeroy earned his undergraduate degree at Virginia Tech, and received a graduate degree in atmospheric science from Wyoming.[citation needed] He currently resides in Salt Lake City, Utah, working as a meteorologist for the U.S. government

 

[jordoo]

Its every year with Wisc and pom. His system always has them higher than they should be IMo even when they are very good.

 

[HOFCeluck]

call it whatever you want, he considers it an indication of who is going to win a game.

wisconsin FINISHED the season in the top 5 last year. they lost 10 times.

Even with that one outlier it's one of the better predictors available IMO.

What would you prefer? The RPI?

 

[pearl31]

'parently Kenny doesn't do any tweaking to his formula in the offseason—despite it being obvious that tweaking is needed

 

[HOFCeluck]

'parently Kenny doesn't do any tweaking to his formula in the offseason—despite it being obvious that tweaking is needed

Changing the whole system for one outlier is usually not a good idea.

 

[pearl31]

one outlier?

 

[HOFCeluck]

one outlier?

I can't think of any more...and it's still too early to go off this years rankings.

 

[pearl31]

His rankings, through the years, have mostly been accurate—but seemingly could be better if he refined his criteria, weighting, etc. By his rationale, a team like Wisc should have been national champs multiple times.
(there have been other temas, like Wisc, that he's over-valued significantly, as well, but I can't list them at the moment)

 

[Cusefannotindc]

The best predictor of a team's offensive and defensive efficiency is the previous years. Therefore teams who finished last year ranked highly will typically begin the next year ranked highly. Wisconsin plays efficient basketball which his system rewards and the Big 10 has been strong top to bottom the last couple of seasons, no Rutgers or Depauls, which keeps them ranked high. Losses are worth less, wins are worth more.

No need for everyone to freak out, the season JUST began, the largest weight on efficiency will still be the preseason scores until new data starts becoming a greater factor.

 

[CusefanATL]

The best predictor of a team's offensive and defensive efficiency is the previous years. Therefore teams who finished last year ranked highly will typically begin the next year ranked highly. Wisconsin plays efficient basketball which his system rewards and the Big 10 has been strong top to bottom the last couple of seasons, no Rutgers or Depauls, which keeps them ranked high. Losses are worth less, wins are worth more.

No need for everyone to freak out, the season JUST began, the largest weight on efficiency will still be the preseason scores until new data starts becoming a greater factor.

wisconsin lost 10 games last year. 1 of which to us. they were ranked higher than us.

LAST YEAR.

 

[pearl31]

glancing at his rankings from the end of last season...
Wisconsin, Memphis, Wichita St, Florida and St. Louis, for example, are all ranked ridiculously high.
While Florida St, Michigan and UNLV, for example, are all ranked way too low.

He's generally accurate, but his system is flawed.

 

[Cusefannotindc]

wisconsin lost 10 games last year. 1 of which to us. they were ranked higher than us.

LAST YEAR.

Michigan St and Ohio St each lost 8.

It's not the end of the world.

 

[Cusefannotindc]

glancing at his rankings from the end of last season...
Wisconsin, Memphis, Wichita St, Florida and St. Louis, for example, are all ranked ridiculously high.
While Florida St, Michigan and UNLV, for example, are all ranked way too low.

He's generally accurate, but his system is flawed.

What would be really impressive is if he could devise a way to forecast how many internet freakouts his rating system will cause.

 

[pearl31]

Or, how many internet fawnings his rating system will inspire.

 

[moqui]

This is an email that I sent to Pomeroy this afternoon. I'll report back if I get an answer:

Ken,
Am i missing something here?. With only one game played, matched pairs should have matching numbers. That is, Syracuse's Offensive Efficiency should have the same value as SDSU's Defensive Efficiency, and vice versa. If SU scores an average of 111 points per 100 possessions, then SDSU's defense must give up the same amount, no?

Instead, we get:

SU offensive efficiency 111.0, SDSU defensive efficiency 91.4
SU defensive efficiency 87.8, SDSU offensive efficiency 106.5

What am I missing here? I don't see this explained in your help section.

Regards,
Moqui

 

[Knicks411]

I believe the efficiency numbers are adjusted for competition, so it isn't just straight raw totals. Also, I believe he continues to use his pre-season projections for the first month or so of the season.

For instance, he has us scoring 62 points on 74 possessions yesterday. That's obviously not 1.11 points per possession. Conversely, we allowed about .66 points per possession, which also is not 91.4, which is our defensive efficiency. Pretty sure he takes the pre-season projections, then uses yesterdays total (adjusted for competition) as another data point. At some point in December I believe the projections are done away with and he just takes the actual season results, still adjusted.

glancing at his rankings from the end of last season...
Wisconsin, Memphis, Wichita St, Florida and St. Louis, for example, are all ranked ridiculously high.
While Florida St, Michigan and UNLV, for example, are all ranked way too low.

It's not like it's just Pomeroy. Sagarin had Memphis 18 as opposed to 19 for Pomeroy. They had Wisconsin 7th. St. Louis 17. Wichita 15. UNLV was 33. I don't really know a lot of the other computer rankings, did they differ a lot on those other teams?