Pomeroy and other Statistical Rankings - 2010-2011 Season

[loran16]

So for his site and College Basketball Prospectus 2010, Ken Pomeroy created his own projections of the NCAA Teams for this season.

The rankings of every team are currently up on his website, kenpom.com. On twitter, he confirmed that he'll be weighting the preseason projections such that they still will have an impact on the rankings for a while.

Just as interestingly, if you click on a team, such as say DUKE (http://kenpom.com/team.php?team=Duke), you get the expected results for that team.

Duke is projected by the system to go 26-4 (13-3). The projections do not include the Gonzaga/K-State game, because who that is is yet to be determined.

Duke is currently favored in every game, but it's two most likely losses are 2/26 @Virginia Tech (64% winning percentage) and 3/5 @UNC (62% Winning Percentage).

[flyingdutchdevil]

Duke is projected by the system to go 26-4 (13-3). The projections do not include the Gonzaga/K-State game, because who that is is yet to be determined.

13-3 in the ACC? Same record as last year? When we're by far the strongest team? And the ACC is weaker than last year? I would think 14-2 would be the most accurate.

26-4 sounds about right, but I would put my money on 28-2 before the ACC tournament. Call me an optimist, but this team is stacked from head to toe and our most difficult games (at least on paper) come before the ACC starts and Duke always shines during these games.

[-jk]

Don't get too worked up with 14-2 or 13-3. That's a rounding error.

We have three upperclassmen. Ok, two are to die for and the third is no slouch. We have a lot of talented but unproven players, we have a completely new system to learn and implement, and enormous hype to endure.

I'm going to enjoy the ride. I expect it'll be a wild one, regardless of the Ws and Ls.

I'm certainly not going to get worked up over someone's system making projections based on seeding data.

-jk

[flyingdutchdevil]

Don't get too worked up with 14-2 or 13-3. That's a rounding error.

We have three upperclassmen. Ok, two are to die for and the third is no slouch. We have a lot of talented but unproven players, we have a completely new system to learn and implement, and enormous hype to endure.

I'm going to enjoy the ride. I expect it'll be a wild one, regardless of the Ws and Ls.

I'm certainly not going to get worked up over someone's system making projections based on seeding data.

-jk

While we're all going to enjoy the ride, part of the ride is the media, polls, analysis, rankings, ratings, and projections. It turns a fun up-and-down roller-coaster ride into the cork-screw, multiple loop, dangling-feet, upside-down roller-coaster ride.

[Duvall]

13-3 in the ACC? Same record as last year? When we're by far the strongest team? And the ACC is weaker than last year? I would think 14-2 would be the most accurate.

Worth noting that with unbalanced schedules a team can play a tougher schedule against a weaker league simply by playing more tough games. Which is true of Duke this year, I think - this season their eight road games include trips to each of the four teams picked to finish in 2nd-6th place in preseason.

[airowe]

We're going to lose a game or two nobody can predict. Happens every year.

I just hope they come before the calendar says March.

[Gthoma2a]

We're going to lose a game or two nobody can predict. Happens every year.

I just hope they come before the calendar says March.

And not to UNC! I like pitching a shutout against them for the year.

[Jderf]

Does anybody else notice that Kenpom gives us an 81% chance of winning at home against Mich St, but only a 66% chance at FSU? Does anybody else think that's giving a LOT of credit to the home court advantage? I guess it is Cameron...

[airowe]

Does anybody else notice that Kenpom gives us an 81% chance of winning at home against Mich St, but only a 66% chance at FSU? Does anybody else think that's giving a LOT of credit to the home court advantage? I guess it is Cameron...

I don't know, does Pomeroy factor in Michigan State's reputation for starting off slow?

[Jderf]

I don't know, does Pomeroy factor in Michigan State's reputation for starting off slow?

That would be some serious mathematical rigor.

[Wander]

I guess it is Cameron...

I'm pretty sure every arena gets the same home court advantage bonus in his system.

Also, I think it's been proven that college basketball has the biggest home court factor of all American sports.

[COYS]

I don't know, does Pomeroy factor in Michigan State's reputation for starting off slow?

Well, Kenpom wasn't nearly as high on Michigan State last year, either. Even after their run through the tournament to the Final Four, he had them at ranked at 23 in adjusted efficiency stats . . . one spot ahead of FSU. For next season, he has them rated 6th in the preseason, which is much lower than any of the human polls. I think his computational methods view the Spartan's run through the tournament last year to be predicated on a bit of luck that allowed an otherwise strong but unspectacular team to appear like an elite team.

[COYS]

After looking at Duke's projected record a bit more, I also realized that Pomeroy expects Duke to play at a similar pace to last season. Only the games against the always fast-paced Tar Heels are projected to reach 70 possessions. So obviously most of his predictions for this season are based on last season and do not take into account how replacing Scheyer, Zoubek, and Thomas in the lineup with Kyrie, Miles, and Mason will dramatically alter the pace at which Duke plays.

[loran16]

Does anybody else notice that Kenpom gives us an 81% chance of winning at home against Mich St, but only a 66% chance at FSU? Does anybody else think that's giving a LOT of credit to the home court advantage? I guess it is Cameron...

The Pomeroy system really likes FSU's defense (Projecting it to be #1 Defensively, for the 2nd year running even without Alabi).

FSU is ranked 20, while MSU is ranked 6 overall, in general....and yes it seems to give a big bonus for Home Court Edge.

(As for the pace being the same as last year, that's a good point).

[nocilla]

Well, Kenpom wasn't nearly as high on Michigan State last year, either. Even after their run through the tournament to the Final Four, he had them at ranked at 23 in adjusted efficiency stats . . . one spot ahead of FSU. For next season, he has them rated 6th in the preseason, which is much lower than any of the human polls. I think his computational methods view the Spartan's run through the tournament last year to be predicated on a bit of luck that allowed an otherwise strong but unspectacular team to appear like an elite team.

I tend to agree with Ken here. For all the talk we hear about how easy of a path Duke had to the Final Four, MSU had an even easier path. We don't hear much about it because they lost to Butler anyway. They do get some credit for beating the teams in front of them and doing so without Lucas for a couple of them, but they avoided the top 3 seeds and their toughest opponent was Maryland. They should be very good this year with the returning talent, but they are still a notch below Duke, IMHO.

MSU's path to the Final Four; (12) New Mex St, (4) Maryland, (9) N Iowa, (6) Tenn.

[timmy c]

While we're all going to enjoy the ride, part of the ride is the media, polls, analysis, rankings, ratings, and projections. It turns a fun up-and-down roller-coaster ride into the cork-screw, multiple loop, dangling-feet, upside-down roller-coaster ride.

I will also enjoy the ride. Last year's squad was so solid. They consistantly got better as the season went along while limiting the typical lows.
This years squad will have more raw talent and youth. I expect the in-season roller-coaster will be a high-flying adventure!

[SCMatt33]

After looking at Duke's projected record a bit more, I also realized that Pomeroy expects Duke to play at a similar pace to last season. Only the games against the always fast-paced Tar Heels are projected to reach 70 possessions. So obviously most of his predictions for this season are based on last season and do not take into account how replacing Scheyer, Zoubek, and Thomas in the lineup with Kyrie, Miles, and Mason will dramatically alter the pace at which Duke plays.

I wouldn't worry about the pace predictions. It is much more difficult for a computer to guess how new players and increased minutes will affect pace compared to efficiencies. It's pretty easy for the computers to take Scheyer out of the equation and guess what Irving will do compared to how fast.

As for the FSU-Mich St. thing, It's not just that he's low on MSU, but high on FSU, all the way up at 20, with the number 1 defense in the country. With Duke looking to be a more offensive oriented team compared to last year, it wouldn't surprise me to see @FSU become a candidate for a surprisingly tough road game and perhaps a loss. I know this stuff isn't in the computers, but Duke will have played all but one game in the month before FSU in Cameron, with the Away game being @Greensboro. The last game before FSU outside the state of North Carolina will have been Butler and that will be a pretty friendly environment. Duke will also be only 3 days removed from what is always an emotional home game against Maryland. Tallahassee has been a tough place for Duke to play in recent years, and I expect a tough game this year.

[timmy c]

Ken Pomeroy was spot on last year in predicting Duke as the number one team in the land. In early February, when Coach K was tempering expectations by describing the team as very good but not great, Kenpom moved Duke into the #1 spot, thanks in large part to defensive efficiency and offensive rebounds. By tournament time most of the play-by-play guys were finally discussing Duke’s huge advantages in those areas. Although Ken’s analysis isn’t perfect, he’s beat some of the best prognosticators with his mathematical genius.

I wouldn’t bet against his predictive models.

[COYS]

I wouldn't worry about the pace predictions. It is much more difficult for a computer to guess how new players and increased minutes will affect pace compared to efficiencies. It's pretty easy for the computers to take Scheyer out of the equation and guess what Irving will do compared to how fast.

As for the FSU-Mich St. thing, It's not just that he's low on MSU, but high on FSU, all the way up at 20, with the number 1 defense in the country. With Duke looking to be a more offensive oriented team compared to last year, it wouldn't surprise me to see @FSU become a candidate for a surprisingly tough road game and perhaps a loss. I know this stuff isn't in the computers, but Duke will have played all but one game in the month before FSU in Cameron, with the Away game being @Greensboro. The last game before FSU outside the state of North Carolina will have been Butler and that will be a pretty friendly environment. Duke will also be only 3 days removed from what is always an emotional home game against Maryland. Tallahassee has been a tough place for Duke to play in recent years, and I expect a tough game this year.

I completely agree. FSU has a team full of long athletes that can probably keep up with Duke's offensive weapons as well as anyone in the land. If I'm a neutral observer asked to predict the three games most likely to be L's for Duke this season, the game in Tallahassee definitely makes the list.

As for pace, I should have been more specific. I wasn't worried about the pace rating for Duke, I just wanted to point out that the Pomeroy's predicted scores do not reflect a likely change in pace for Duke, making some of the predicted results seem a little less accurate. I think it likely that after a few games this season, the predicted pace and scoring margins for many of the games will shift upward very quickly, even if the predicted difference in efficiency between the two teams remains more or less the same.

I think the fact that Duke's offensive efficiency drops a bit from last year (from 123-120) serves as a reminder that for all the talk about how great our defense was last year, the offense was even better . . . even without the highlight reel fast breaks. The starting five could execute our half court motion with ice cold efficiency. I can't wait to see if this year's offense can achieve the same success by entirely different means . . . by running teams into oblivion.

[gam7]

After looking at Duke's projected record a bit more, I also realized that Pomeroy expects Duke to play at a similar pace to last season. Only the games against the always fast-paced Tar Heels are projected to reach 70 possessions. So obviously most of his predictions for this season are based on last season and do not take into account how replacing Scheyer, Zoubek, and Thomas in the lineup with Kyrie, Miles, and Mason will dramatically alter the pace at which Duke plays.

This must be related to the pace projection: it's interesting to me that his model predicts that we will not score more than 86 points in any game. I predict that we'll exceed that total about a dozen times this year. In the same vein, he predicts that we will give up more than 70 points just once. I predict that we will give up more than 70 points a dozen times as well.

[Kedsy]

This must be related to the pace projection: it's interesting to me that his model predicts that we will not score more than 86 points in any game. I predict that we'll exceed that total about a dozen times this year. In the same vein, he predicts that we will give up more than 70 points just once. I predict that we will give up more than 70 points a dozen times as well.

Both these issues are entirely related to pace. If Pomeroy is projecting us to have a similar pace to last year, then his pre-season projections (at least with regards to Duke) are meaningless. Once the season gets underway and he gets enough data points, his ratings will model the current year's teams, and his ratings will once again be "must read" for serious college hoops fans.

[SCMatt33]

Right now, the "player stats" section at the bottom of each team's page is blank with a note that they will be coming in December. When KenPom has enough info to post player efficiency stats, I would start paying serious attention to his team info as well.

[NSDukeFan]

Ken Pomeroy was spot on last year in predicting Duke as the number one team in the land. In early February, when Coach K was tempering expectations by describing the team as very good but not great, Kenpom moved Duke into the #1 spot, thanks in large part to defensive efficiency and offensive rebounds. By tournament time most of the play-by-play guys were finally discussing Duke’s huge advantages in those areas. Although Ken’s analysis isn’t perfect, he’s beat some of the best prognosticators with his mathematical genius.

I wouldn’t bet against his predictive models.


Both these issues are entirely related to pace. If Pomeroy is projecting us to have a similar pace to last year, then his pre-season projections (at least with regards to Duke) are meaningless. Once the season gets underway and he gets enough data points, his ratings will model the current year's teams, and his ratings will once again be "must read" for serious college hoops fans.

Ken is certainly popular on this board and I am very interested in what he has to say and enjoy his ratings model during the year. But, as Kedsy says, I don't know if he is any better at predicting how teams will do when he doesn't have any data to use in his models. I am not going to take too much from his analysis right now, but will become much more interested once teams have played some games.

[sagegrouse]

Ken is certainly popular on this board and I am very interested in what he has to say and enjoy his ratings model during the year. But, as Kedsy says, I don't know if he is any better at predicting how teams will do when he doesn't have any data to use in his models. I am not going to take too much from his analysis right now, but will become much more interested once teams have played some games.

I agree, NSDukeFan.

Fact is, KenPom would certainly rather start his analysis when he has ten or so games to crank into his model. However, he has to make a living, and he can't skip the preseason and early season forecasts (or, if he did, no one would be able to read about it him later). So he goes with what he has.

One benefit of a preseason analysis is to later give teams stars and kudos for doing better than expected, with those expectations based on last year's results, additions, subtractions, opposition, etc. And it's sure more thoughtful than the preseason polls.

sagegrouse

[badgerbd]

This must be related to the pace projection: it's interesting to me that his model predicts that we will not score more than 86 points in any game. I predict that we'll exceed that total about a dozen times this year. In the same vein, he predicts that we will give up more than 70 points just once. I predict that we will give up more than 70 points a dozen times as well.

His model predicts that in any given game he wouldn't bet on us to score more than 86.
Not that it doesn't expect us to score more than 86 at some point in the season.

Sort of like how it predicts we win every individual game, but doesn't predict the team will go undefeated.

[MChambers]

His model predicts that in any given game he wouldn't bet on us to score more than 86.
Not that it doesn't expect us to score more than 86 at some point in the season.

Sort of like how it predicts we win every individual game, but doesn't predict the team will go undefeated.

My guess is that he (Pomeroy) came up with a way to roughly predict a team's quality, based on its losses and new players, but didn't come up with a way to predict a team's style. (Not surprising at all, in my view.) So he's basing tempo on last year, or maybe the last three years, or something like that.

Let's see what the tempo numbers look like in mid-December.

[loran16]

My guess is that he (Pomeroy) came up with a way to roughly predict a team's quality, based on its losses and new players, but didn't come up with a way to predict a team's style. (Not surprising at all, in my view.) So he's basing tempo on last year, or maybe the last three years, or something like that.

Let's see what the tempo numbers look like in mid-December.

I bought college bball prospectus myself. His system is based upon:
Adjusted efficiencies from the past two seasons,
Returning Personnel,
Recruiting Information.

Value is adjusted based upon usage of returning and non-returning players. He admits it's a first try at a system and is hardly set in stone (he even refuses to name the system, like the prospectus sites have named all their pro sports projections systems.)

I don't see tempo mentioned anywhere in his introduction to the system.

[COYS]

I don't see tempo mentioned anywhere in his introduction to the system.

Maybe we are to assume that tempo is lumped in with efficiency stats from the past two seasons? 67.4 is our adjusted tempo for 2009 and 65.5 is our tempo for 2010. Our projected tempo for 2011 is . . . 66.5. He must average the two tempos.

[DukieTiger]

Ken usually starts the year with all zeros across the board. I like that he's stepping out on a limb and giving us something to chew on and talk about.

Another thing I'm excited about is that he said he will be archiving his predictions this year, so that we can compare them to the actual results and future predictions. In years past, we would see his projections but it was always hard to see how they stacked up, because they changed every time new data was entered. It will be interesting to see if his predictions become more accurate as the season goes on.

[Troublemaker]

Don't get too worked up with 14-2 or 13-3. That's a rounding error.

We have three upperclassmen. Ok, two are to die for and the third is no slouch. We have a lot of talented but unproven players, we have a completely new system to learn and implement, and enormous hype to endure.

I'm going to enjoy the ride. I expect it'll be a wild one, regardless of the Ws and Ls.

I'm certainly not going to get worked up over someone's system making projections based on seeding data.

-jk

Didn't want to sidetrack the nerdy statistical model talk, but I liked this post by jk a lot. I, too, will really enjoy the ride this season. And if Duke happens to "only" go 26-4/13-3 this season, I'm sure there will be good reasons for that, that are unrelated to effort or underperformance, and I'm sure I will enjoy even the losses.

This team is going to play a fun and aesthetically-pleasing style of basketball, and more importantly, this season will be a post-national-championship honeymoon season. The players won't dwell on the previous season's accomplishments, but the fans certainly can! I'm going to smile every time an announcer refers to Duke as the defending champs, and I'm going take comfort in knowing that the current team has an outstanding blend of talent and senior leaders. Realistically, there's almost nothing that can happen (besides injury) during the regular season that would make me think Duke isn't a top 5 team. Even if they lose like 6 times. I mean, who cares? The roster has Singler, Smith, Irving, guard depth, and athletic, skilled post players. Give me that roster healthy in March and we'll rock and roll.

To summarize, this is going to be SUCH an enjoyable season to watch game by game. Because (a) it's post-championship and (b) we, the fans, can take comfort in knowing that the roster is loaded and that a loss here or there won't really take away from that. Just knock on wood for health.

[Troublemaker]

Pomeroy preseason ratings that were interesting to me:

Ohio State at 2: I could see this being correct
Kansas at 3: Very interesting. Probably too high, but Kansas is going to be a top 10 team this season despite their losses.
Syracuse at 5: Ditto
Michigan St at 6: Too low. I buy the hype. Factor in a slow start for injury recovery, though
Illinois at 9: Dead on. Big 10 to be loaded

Bleh. Everything past the top 10 looks good/possible to me.