PDA

View Full Version : Pomeroy Strength of Schedule



camion
11-27-2012, 06:48 AM
Here are the Strength of Schedule rankings for the Pomeroy (http://kenpom.com/) top ten ranked teams. Can you guess which one is Duke.

268
189
195
185
130
49
294
6
173
171

I know it's early yet for the computer polls, but still that's pretty impressive. It doesn't look like the other top teams have been working all that hard.

MarkD83
11-27-2012, 07:33 AM
Not to pik nits but....

Isn't 6 too low for an SOS after playing the 2nd and 3rd ranked team in the nation in two of six games.

BD80
11-27-2012, 07:55 AM
Here are the Strength of Schedule rankings for the Pomeroy (http://kenpom.com/) top ten ranked teams. Can you guess which one is Duke.

268
189
195
185
130
49
294
6
173
171

I know it's early yet for the computer polls, but still that's pretty impressive. It doesn't look like the other top teams have been working all that hard.

The only team with a more difficult SOS in the top 100 is Deleware, whose schedule becomes MORE difficult this weekend ;)

The SOS at this point is horse hockey anyway, the North Florida Ospreys are "ahead" of us in SOS, and they have played Edward Waters.

Playing OSU should move us up a bit, but the ACC schedule might bring us down :(

JBDuke
11-27-2012, 08:09 AM
Not to pik nits but....

Isn't 6 too low for an SOS after playing the 2nd and 3rd ranked team in the nation in two of six games.

One of the things that impacts the SOS early in the season is that most top teams feast on some Little Sisters of the Poor in November and December. Then, as conference play picks up, they tend to face stronger opponents and their SOS picks up. Meanwhile, the Little Sisters of the Poor, who played a bunch of tough teams early in "revenue" games, now are playing their weaker conference opponents.

Duke's SOS is really good for a top team early in the year. The numbers are brought down a bit by playing Georgia State (#177) and FGCU (#135). Meanwhile, Mississippi Valley State has the #2 SOS because they've played Ole Miss (#28), Cincinnati (#16), LSU (#116), and Northwestern (#56). They're winless so far, so they're #333 in Pomeroy's rankings, but they've only played one opponent in triple digits, and Duke has played two weaker teams.

pfrduke
11-27-2012, 08:49 AM
The only team with a more difficult SOS in the top 100 is Deleware, whose schedule becomes MORE difficult this weekend ;)

The SOS at this point is horse hockey anyway, the North Florida Ospreys are "ahead" of us in SOS, and they have played Edward Waters.

Playing OSU should move us up a bit, but the ACC schedule might bring us down :(

Because Pomeroy only ranks DI teams, games against Non-DI teams simply don't count for SOS - it's like it never happened. So teams like North Florida don't get dinged for playing teams like Edward Waters. It's obviously an imperfect system.

patentgeek
11-27-2012, 09:29 AM
It's also worth remembering that the Pomeroy ratings put no stock whatsoever into actual wins and losses - it's simply about points per possession scored on offense and points per possession given up on defense. Thus, a 1 point win vs a 1 point loss against a particular opponent are treated almost identically, whereas a 1 point win vs. a 20 point win against a particular opponent are treated much differently. I'm not suggesting that Pomeroy is either right or wrong about this type of treatment - only that it yields results that can cause you to scratch your head if you focus only on wins and losses.

sagegrouse
11-27-2012, 09:29 AM
Because Pomeroy only ranks DI teams, games against Non-DI teams simply don't count for SOS - it's like it never happened. So teams like North Florida don't get dinged for playing teams like Edward Waters. It's obviously an imperfect system.

I was in a discussion here a while ago, where those on the other side wanted to rank a recruiting class by averaging the rankings of the recruits. Therefore, a team with three top ten recruits and one rated #100 would not have as good a recruiting class as a team that recruited three players ranked between 10 and 20. No guys, it really is about how many top recruits you get, and you shouldn't get penalized for one or two marginal players.

In the context of SOS, would you say that a team that plays three top ten teams and three ranked 250 to 300 (average ranking = 140) would have a weaker strength of schedule than a team that plays six teams ranked between 40 and 60 (average ranking = 50)? I wouldn't, although I would agree that the logic is somewhat different than in the previous example.

sagegrouse

Kedsy
11-27-2012, 09:54 AM
I was in a discussion here a while ago, where those on the other side wanted to rank a recruiting class by averaging the rankings of the recruits. Therefore, a team with three top ten recruits and one rated #100 would not have as good a recruiting class as a team that recruited three players ranked between 10 and 20. No guys, it really is about how many top recruits you get, and you shouldn't get penalized for one or two marginal players.

In the context of SOS, would you say that a team that plays three top ten teams and three ranked 250 to 300 (average ranking = 140) would have a weaker strength of schedule than a team that plays six teams ranked between 40 and 60 (average ranking = 50)? I wouldn't, although I would agree that the logic is somewhat different than in the previous example.

Well, you're right about the recruiting rankings, but the reason you're right is the lower rated recruits sit at the end of the bench. That's not a good analogy for SOS, because all the games count.

I think you also have to consider the purpose of the SOS ranking. If its purpose is to chat over a couple beers about who ran the more brutal gauntlet, then perhaps your way is correct, because the first team had to play three super-hard opponents and the second team didn't play any. If, however, you're the RPI and you're trying to decide who had a harder time compiling their 21-4 record, a team who played five top 5 teams and 20 teams ranked between 250 and 300 has played a much easier schedule than a team who had to try and beat 25 teams rated between 40 and 60.

I believe the purpose of Pomeroy's SOS (to the extent that it has a purpose and is not just a by-product of his other calculations) is to allow his efficiency rankings to be adjusted for schedule. And in that case, similar to the RPI, the second team in your example has played a more difficult schedule, because it's sooooo much easier to rack up points per possession against #275.

sagegrouse
11-27-2012, 10:13 AM
Well, you're right about the recruiting rankings, but the reason you're right is the lower rated recruits sit at the end of the bench. That's not a good analogy for SOS, because all the games count.

I think you also have to consider the purpose of the SOS ranking. If its purpose is to chat over a couple beers about who ran the more brutal gauntlet, then perhaps your way is correct, because the first team had to play three super-hard opponents and the second team didn't play any. If, however, you're the RPI and you're trying to decide who had a harder time compiling their 21-4 record, a team who played five top 5 teams and 20 teams ranked between 250 and 300 has played a much easier schedule than a team who had to try and beat 25 teams rated between 40 and 60.

I believe the purpose of Pomeroy's SOS (to the extent that it has a purpose and is not just a by-product of his other calculations) is to allow his efficiency rankings to be adjusted for schedule. And in that case, similar to the RPI, the second team in your example has played a more difficult schedule, because it's sooooo much easier to rack up points per possession against #275.

Actually, I didn't take a position -- I just raised a question. But you make good points re the quant applications.

sage

Kedsy
11-27-2012, 10:49 AM
In the context of SOS, would you say that a team that plays three top ten teams and three ranked 250 to 300 (average ranking = 140) would have a weaker strength of schedule than a team that plays six teams ranked between 40 and 60 (average ranking = 50)? I wouldn't, although I would agree that the logic is somewhat different than in the previous example.

sagegrouse


Actually, I didn't take a position -- I just raised a question.

It hardly matters, but the bolded section above makes it appear to me as if you took a position. Not that there's anything wrong with that. ;)

Wander
11-27-2012, 10:58 AM
I believe the purpose of Pomeroy's SOS (to the extent that it has a purpose and is not just a by-product of his other calculations) is to allow his efficiency rankings to be adjusted for schedule. And in that case, similar to the RPI, the second team in your example has played a more difficult schedule, because it's sooooo much easier to rack up points per possession against #275.

This is exactly right. The key point to remember is that margin of victory matters in Pomeroy rankings. If straight win-loss was all that mattered, I agree playing a few top ten teams and a few bottom ten teams might be harder than playing all 40-60 teams. But if you're going to measure per possession and not per game, the latter option definitely makes more sense to count as the more difficult schedule.

sagegrouse
11-27-2012, 11:26 AM
It hardly matters, but the bolded section above makes it appear to me as if you took a position. Not that there's anything wrong with that. ;)

There you go again. Reading my posts in detail. What nerve!

As a Duke fan, I would rate the playing of three top ten teams and three patsies harder than a schedule with six teams just below the NCAA tournament qualifying level. Basically, I would be really mad as a Duke fan to lose any games in the latter case, and I would be overjoyed to go undefeated in the former case.

I am hoping for a little more joy tomorrow.

sagegrouse

Olympic Fan
11-27-2012, 12:23 PM
I see that the rankings have Duke at "only" No. 6, but does anybody read Pomeroy's blog?

In his wrapup of the previous week's games, he picked the three most significant games of the week in order:

http://kenpom.com/blog/index.php/weblog/entry/weekend_in_review_bruinses_collapses

The 3 most appealing games (according to FanMatchTM):
1) Duke 76, Louisville 71 [74]. (Saturday) The Cardinals were playing without Gorgui Dieng—who Coach K wisely referred to as “the big guy” or something like that after the game, presumably deferring on the pronunciation – and Louisville could have used him. The Krzyzewski/Pitino reunion produced a compelling tight game, even though Louisville led for just one possession of the second half.

2) Duke 89, Minnesota 71 [72]. (Thursday) This is the game that forced me to take notice of Duke. The Gophers would go on to beat Memphis and Stanford, so it’s quite possible they’re a top 25 team themselves.

3) Duke 67, VCU 58 [65]. (Friday) Hey, it’s not my fault the Battle 4 Atlantis organizers decided to make Duke run a gauntlet of teams that has to be one of the most difficult in the history of preseason tournaments. We’ll see more diversity in this space on Friday. In the meantime, know that no other human-ranked team has played anything close to the schedule Duke has thus far.

He comes back to pick gthe most appealing games for next week ... and apologizes for picking Ohio State at Duke No. 1:

1) #7 Ohio State at #8 Duke. (Wednesday, 9 ET, ESPN) OK, I swear there will be more diversity here eventually. Duke’s in the middle of brutal stretch, you know? I haven’t run the kPOY calculations yet because it’s kind of early, but I expect Deshaun Thomas would be in a close race with Cody Zeller at the top. Duke has more of a balanced attacked, and interestingly, they so far have refused to rebound. Let’s see if that continues.


BTW: According to RPI, Duke's SOS is No. 7 right now.

ChillinDuke
11-27-2012, 12:39 PM
I see that the rankings have Duke at "only" No. 6, but does anybody read Pomeroy's blog?

In his wrapup of the previous week's games, he picked the three most significant games of the week in order:

http://kenpom.com/blog/index.php/weblog/entry/weekend_in_review_bruinses_collapses

The 3 most appealing games (according to FanMatchTM):
1) Duke 76, Louisville 71 [74]. (Saturday) The Cardinals were playing without Gorgui Dieng—who Coach K wisely referred to as “the big guy” or something like that after the game, presumably deferring on the pronunciation – and Louisville could have used him. The Krzyzewski/Pitino reunion produced a compelling tight game, even though Louisville led for just one possession of the second half.

2) Duke 89, Minnesota 71 [72]. (Thursday) This is the game that forced me to take notice of Duke. The Gophers would go on to beat Memphis and Stanford, so it’s quite possible they’re a top 25 team themselves.

3) Duke 67, VCU 58 [65]. (Friday) Hey, it’s not my fault the Battle 4 Atlantis organizers decided to make Duke run a gauntlet of teams that has to be one of the most difficult in the history of preseason tournaments. We’ll see more diversity in this space on Friday. In the meantime, know that no other human-ranked team has played anything close to the schedule Duke has thus far.

He comes back to pick gthe most appealing games for next week ... and apologizes for picking Ohio State at Duke No. 1:

1) #7 Ohio State at #8 Duke. (Wednesday, 9 ET, ESPN) OK, I swear there will be more diversity here eventually. Duke’s in the middle of brutal stretch, you know? I haven’t run the kPOY calculations yet because it’s kind of early, but I expect Deshaun Thomas would be in a close race with Cody Zeller at the top. Duke has more of a balanced attacked, and interestingly, they so far have refused to rebound. Let’s see if that continues.


BTW: According to RPI, Duke's SOS is No. 7 right now.

This is really cool. Thanks for posting.

- Chillin

loran16
11-27-2012, 02:09 PM
I see that the rankings have Duke at "only" No. 6, but does anybody read Pomeroy's blog?

In his wrapup of the previous week's games, he picked the three most significant games of the week in order:

http://kenpom.com/blog/index.php/weblog/entry/weekend_in_review_bruinses_collapses

The 3 most appealing games (according to FanMatchTM):
1) Duke 76, Louisville 71 [74]. (Saturday) The Cardinals were playing without Gorgui Dieng—who Coach K wisely referred to as “the big guy” or something like that after the game, presumably deferring on the pronunciation – and Louisville could have used him. The Krzyzewski/Pitino reunion produced a compelling tight game, even though Louisville led for just one possession of the second half.

2) Duke 89, Minnesota 71 [72]. (Thursday) This is the game that forced me to take notice of Duke. The Gophers would go on to beat Memphis and Stanford, so it’s quite possible they’re a top 25 team themselves.

3) Duke 67, VCU 58 [65]. (Friday) Hey, it’s not my fault the Battle 4 Atlantis organizers decided to make Duke run a gauntlet of teams that has to be one of the most difficult in the history of preseason tournaments. We’ll see more diversity in this space on Friday. In the meantime, know that no other human-ranked team has played anything close to the schedule Duke has thus far.

He comes back to pick gthe most appealing games for next week ... and apologizes for picking Ohio State at Duke No. 1:

1) #7 Ohio State at #8 Duke. (Wednesday, 9 ET, ESPN) OK, I swear there will be more diversity here eventually. Duke’s in the middle of brutal stretch, you know? I haven’t run the kPOY calculations yet because it’s kind of early, but I expect Deshaun Thomas would be in a close race with Cody Zeller at the top. Duke has more of a balanced attacked, and interestingly, they so far have refused to rebound. Let’s see if that continues.


BTW: According to RPI, Duke's SOS is No. 7 right now.

Again, Duke is ranked 8th right now in Pomeroy for several reasons:
1. Preseason rankings are still being weighed in -> Duke was preseason ranked 8th or 9th (I forget which).
2. Lack of Cupcake beatdowns - Pomeroy's system doesn't care about wins or losses. It cares about efficiency. And it - admittedly by the man himself - can get a bit skewed by meaningless blowouts over crappy teams. Duke's only played two bad teams, neither of which was truly horrific. By Contrast Indiana has played 3 teams way worse than our worst opponent (Georgia State) at home and won each by 40-50 points. This is also known as the Wisconsin effect. Ohio State has one game like this against UMKC which has skewed their rankings slightly (which is why they jumped us after their win on Sunday).