Dork Polls: Men's Bball 2013-14 Edition

[Troublemaker]

Wanted to start this thread for this season because...

Well, go to Pomeroy this morning if you haven't already. I just about fell out of my chair.

[CDu]

Giving up 90 points and almost losing at home to a 1-4 team early in the year will do that. Especially when that follows another all-too-close game at home against another team outside of the top-150.

In 1/3 of our games so far, we have struggled to beat bad teams at home. Hard to argue we're playing like a top-25 team right now.

[pfrduke]

Wanted to start this thread for this season because...

Well, go to Pomeroy this morning if you haven't already. I just about fell out of my chair.

I'm going to mention this again in the weekly roundup, but there's been a lot more variance in the early season numbers than in the last few seasons (and not just in connection with our ranking). I wonder if there's less weight on pre-season as part of the new algorithm that was rolled out.

That said, it's amazing to see us ranked 1st in offense but 176th in defense and out of the top 25, which might be the first time ever under his system (even in 2007 we hung around the top 25 because we lost several close games).

[sagegrouse]

Wanted to start this thread for this season because...

Well, go to Pomeroy this morning if you haven't already. I just about fell out of my chair.

I'm not sure why we have mysteries here, but Duke is no. 1 on offense and no. 176 on defense. Overall ranking is 28.

Sounds about right. At least, the coaches know what to work on.

Sagegrouse

[tbyers11]

I'm going to mention this again in the weekly roundup, but there's been a lot more variance in the early season numbers than in the last few seasons (and not just in connection with our ranking). I wonder if there's less weight on pre-season as part of the new algorithm that was rolled out.


I too noticed the increased variance in Pomeroy. If the system was tweaked to limit positive gains in blowing out vastly inferior teams is a side effect that playing poorly against said teams affects you more negatively. That's not how my brief understanding of the tweaks worked out. Could be small sample size or as you mentioned less weight on pre-season rankings. Interesting to monitor.

[Troublemaker]

I'm going to mention this again in the weekly roundup, but there's been a lot more variance in the early season numbers than in the last few seasons (and not just in connection with our ranking). I wonder if there's less weight on pre-season as part of the new algorithm that was rolled out.

That said, it's amazing to see us ranked 1st in offense but 176th in defense and out of the top 25, which might be the first time ever under his system (even in 2007 we hung around the top 25 because we lost several close games).

Yes, it was the 176 that was so striking to me.

To others, I'm not contesting the sensibility of it all, once I digested it. But, this morning, I was momentarily a little bit shocked to see 176 pop up when loading Pomeroy. I think there will be others with the same reaction. (Perhaps my own lead-in to this thread would shockproof them though by having them expect something noteworthy).

[CLW]

Knew the # would be bad but didn't realize it would be THAT bad on D. I think before last night we were at roughly 58 or so on D so last night's performance had a huge impact on the ranking. Then again given how bad we looked last night on D perhaps it should have.

[Ggallagher]

I'm not sure why we have mysteries here, but Duke is no. 1 on offense and no. 176 on defense. Overall ranking is 28.

Sounds about right. At least, the coaches know what to work on.

Sagegrouse

Our ranking really startled me when I saw it this morning - wish I could argue with it, but we definitely have a lot to work on at the defensive end. Out of curiosity, I sent an email Kenpom asking if we might have set a record with our current offense to defense gap. I have no idea whether he can determine that - or will be interested in replying - but it would have to be gut wrenching to watch any other team with an even greater difference in their offense/defense capabilities.

[flyingdutchdevil]

Yes, it was the 176 that was so striking to me.

To others, I'm not contesting the sensibility of it all, once I digested it. But, this morning, I was momentarily a little bit shocked to see 176 pop up when loading Pomeroy. I think there will be others with the same reaction. (Perhaps my own lead-in to this thread would shockproof them though by having them expect something noteworthy).

176 sounds about right to me. Our defense has been poor all season (yes, I know it's only been 6 games).

This is a great incentive for the players and coaches to inject a little fire into these youngins.

LET'S GO DUKE!

[hurleyfor3]

People!

They're called DORK polls.

And they won't be fully connected (where a line can be drawn between any two teams via a series of common opponents) for another couple of weeks.

Feh. Dork wannabees.

[bob blue devil]

...And they won't be fully connected (where a line can be drawn between any two teams via a series of common opponents) for another couple of weeks.

Interestingly, the same system KP uses to normalize for the lack of connectedness and small sample size (a pre-season ranking that is weighted in his in-season rankings) almost definitely has had the impact of improving our defensive rating in his system. I.e. our on the court results are very likely worse than 176th!

[jipops]

I actually thought the defensive rating would be even worse. This means we've been more efficient defensively than 175 other div 1 teams, and I wouldn't have confidently said that last night following after the first set of games we've seen.

Even after last night, Vermont is still 121 in offense. sigh...

Worst defense of the K era?

[Duvall]

I actually thought the defensive rating would be even worse. This means we've been more efficient defensively than 175 other div 1 teams, and I wouldn't have confidently said that last night following after the first set of games we've seen.

Even after last night, Vermont is still 121 in offense. sigh...

As bob blue devil noted, the ratings still include some influence from the preseason projection that had Duke as a top ten team with a top 30 defense. So the reality is probably even worse.

[Duvall]

Worst defense of the K era?

Tough to say without a way to adjust the stats for the 82 and 83 teams for the quality of their opponents. Worst of the last fifteen years or so, almost certainly thus far.

[Troublemaker]

Interesting blog post from Pomeroy this week about how charges may be making a comeback: http://kenpom.com/blog/index.php/weblog/entry/are_charges_coming_back

Summary:

First, turnovers are significantly lower this season, but they are rising as the season progresses, unlike previous seasons when they declined with more games played.


But the trend in turnovers is zigging when previous seasons zagged, increasing since the beginning of the season which is the opposite of what normally happens.

Next, he breaks down turnovers into two categories: Steals and Non-Steals.

He shows that Steals are lower this season and holding steady. BUT, Non-Steal Turnovers (NSTs), of which charges are a major component, have been rising back towards previous seasons' levels.


The interesting thing is that the difference in NST’s between this season and previous seasons is rapidly disappearing. The entire difference over the past few days could explain just under one charge per team per game. It also could be that teams are traveling more or making other unforced mistakes more often, but given that the trend in steals isn’t changing, that seems a little more difficult to believe. This is far from a conclusive study, but the charge may be sneaking its way back into the game.

There are graphs showing this stuff. Pretty cool.

[nocilla]

Could it be that offensive players are attacking without regards in a sole attempt to draw fouls since the fouls are being called tighter? It could be a temporary thing as the players adjust to figuring out what will be called and what they can get away with, the same way defenses are adjusting on their end.

[Newton_14]

Interesting blog post from Pomeroy this week about how charges may be making a comeback: http://kenpom.com/blog/index.php/weblog/entry/are_charges_coming_back

Summary:

First, turnovers are significantly lower this season, but they are rising as the season progresses, unlike previous seasons when they declined with more games played.



Next, he breaks down turnovers into two categories: Steals and Non-Steals.

He shows that Steals are lower this season and holding steady. BUT, Non-Steal Turnovers (NSTs), of which charges are a major component, have been rising back towards previous seasons' levels.



There are graphs showing this stuff. Pretty cool.

This fits with my eye test. I had noticed without checking any stats or boxscores, that over our last 3 or 4 games, we have drawn more charges than the earlier games. Which I feel is a good thing. You can't eliminate the charge from the game. It is very much a legal defensive play. Like I have stated for years, when a guy flops, then call nothing, call a block, or T him up, but if it is a legitimate charge, even by a help defender then the ref should call it.

[ChillinDuke]

Don't look now but Duke climbed from KenPom's #101 Defense to #66 after last night's performance.

Keep it going, boys.

- Chillin

[Kedsy]

Don't look now but Duke climbed from KenPom's #101 Defense to #66 after last night's performance.

Keep it going, boys.

- Chillin

Which all by itself answers the question of how much KenPom's ratings mean at this time of year.

[Duvall]

Which all by itself answers the question of how much KenPom's ratings mean at this time of year.

Or how good Duke's defensive success was last night.

[vick]

Or how good Duke's defensive success was last night.

It's both, right? As well as the fact that the difference between teams just isn't that great at that level. Looking at last year (so even with a full season of data), the difference between 1 and 10 was 3.6 points per 100 possessions. That's the same as the difference between 10 and 33, or between 33 and 82. Even later in the season, it's not that hard to move up a number of ranks if you aren't near the top, because those teams aren't very differentiated (sort of puts into perspective the level of outrage about tournament snubs every year).

[MChambers]

Or how good Duke's defensive success was last night.
Or both, showing that this team has a chance to be really good.

[uh_no]

It's both, right? As well as the fact that the difference between teams just isn't that great at that level. Looking at last year (so even with a full season of data), the difference between 1 and 10 was 3.6 points per 100 possessions. That's the same as the difference between 10 and 33, or between 33 and 82. Even later in the season, it's not that hard to move up a number of ranks if you aren't near the top, because those teams aren't very differentiated (sort of puts into perspective the level of outrage about tournament snubs every year).

this is the key...it's like running the boston marathon a minute faster and finishing one thousand places better....the rankings themselves are largely irrelevent, it's only the ratings that are relevant and used in calculations...the ranking numbers are just applied for us fans to have something to argue about

[TexHawk]

Which all by itself answers the question of how much KenPom's ratings mean at this time of year.

Does anyone know if Kenpom aggregates and trends his numbers across the year? I know a lot of people like spouting things like "the national champ has been in Kenpom's top 10 every year!", well, it would also be nice to see WHEN those ratings were made. Kenpom saying that Louisville is #1 in his ratings before the tournament starts is interesting, sure. But it would also be nice to see that they've been in his Top 3 since November. Or am I crazy?

The data is readily available, it *seems* like it should be easy.

[Kedsy]

Does anyone know if Kenpom aggregates and trends his numbers across the year? I know a lot of people like spouting things like "the national champ has been in Kenpom's top 10 every year!", well, it would also be nice to see WHEN those ratings were made. Kenpom saying that Louisville is #1 in his ratings before the tournament starts is interesting, sure. But it would also be nice to see that they've been in his Top 3 since November. Or am I crazy?

The data is readily available, it *seems* like it should be easy.

It might be interesting, but I doubt it would tell us much. I expect Duke and Kansas this season will be exhibits A and B for the argument that teams can be very different in the beginning and end of the season.

[MChambers]

Duke is improving, not just based on the eye test, but also in the Dork polls. Up to #5 in the Sagarin predictor (but 11 in the blended rating) and #14 in Pomeroy (#2 in offense and #48 on defense). Let's hope the team continues to improve, especially on defense.

[CLW]

Duke is improving, not just based on the eye test, but also in the Dork polls. Up to #5 in the Sagarin predictor (but 11 in the blended rating) and #14 in Pomeroy (#2 in offense and #48 on defense). Let's hope the team continues to improve, especially on defense.

Yep the defensive performance of late has been much better and I'm pretty sure yesterday's performance jumped us 10-15 spots in the D rankings. That Vermont game is just going to haunt the defensive rating for the team for the remainder of the year I'd imagine.

[loran16]

One thing of note is that dukes rankings aren't improving as much in years past due to a tweak in pomeroys formula to account for the"Wisconsin effect". As a result games against elon and emu do not have nearly the same impact.

A game on the road vs notre dame would of course

[MChambers]

Anyone know what the Sagarin "golden mean" measure is and how it is calculated? It's based purely on scores, like the predictor, but it seems to often vary significantly from the predictor. Since Sagarin doesn't explain it, and since Duke rates poorly on it (42), I am choosing to disregard it.

[77devil]

Anyone know what the Sagarin "golden mean" measure is and how it is calculated? It's based purely on scores, like the predictor, but it seems to often vary significantly from the predictor. Since Sagarin doesn't explain it, and since Duke rates poorly on it (42), I am choosing to disregard it.

When I looked today, Duke is ranked 11 on the golden mean scale. In the top 15 overall ranking, Duke is one of but three along with Kansas and Creighton whose golden mean and predictor ranking deviate more three positions(most are one, two or the same). Kansas differs by 17 positions while Duke and Creighton are 6. I didn't see a common thread between the three schools from available data.

[Indoor66]

When I looked today, Duke is ranked 11 on the golden mean scale. In the top 15 overall ranking, Duke is one of but three along with Kansas and Creighton whose golden mean and predictor ranking deviate more three positions(most are one, two or the same). Kansas differs by 17 positions while Duke and Creighton are 6. I didn't see a common thread between the three schools from available data.

Can ANYONE explain what any of the above means? Strikes me as mental masturbation. (Nothing aimed at the reporter, 77devil, just at the data/rankings/B.S.)

[Duvall]

Can ANYONE explain what any of the above means? Strikes me as mental masturbation. (Nothing aimed at the reporter, 77devil, just at the data/rankings/B.S.)

As opposed to pointless invective? If you don't like the dork poll thread, don't read the dork poll thread.

[Des Esseintes]

Can ANYONE explain what any of the above means? Strikes me as mental masturbation. (Nothing aimed at the reporter, 77devil, just at the data/rankings/B.S.)

I mean, sports fandom IS mental masturbation. That's literally all it is. I'm fine with that. You've been here longer than I have, so I assume you are, too. Right?

[77devil]

Can ANYONE explain what any of the above means? Strikes me as mental masturbation. (Nothing aimed at the reporter, 77devil, just at the data/rankings/B.S.)

I suppose one person's mental masturbation is another's quant puzzle. If I had the data, I would enjoy deciphering why Duke, Creighton, and Kansas are outliers.

[sporthenry]

Duke has jumped up from 23rd before the UVA game to 12th in Kenpom. They were 20th after UVA and 18th going into Miami. 14th in Sagarin, not sure what they were before.

Absolute beatdowns of two top 100 teams including one on the road must have really helped. Also, apparently Kenpom updates his algorithm to ignore expected blow outs so there is more movement as there isn't as much data. Add in Michigan going on a run and easy to see how we've jumped up again.

[77devil]

Duke has jumped up from 23rd before the UVA game to 12th in Kenpom. They were 20th after UVA and 18th going into Miami. 14th in Sagarin, not sure what they were before.

Absolute beatdowns of two top 100 teams including one on the road must have really helped. Also, apparently Kenpom updates his algorithm to ignore expected blow outs so there is more movement as there isn't as much data. Add in Michigan going on a run and easy to see how we've jumped up again.

Duke's SOS, 20, and defensive efficiency, 73, have moved up considerably. Both were in the mid to high 100s before Virginia. In addition to Michigan, UCLA is in the top 25 in the dork polls as another quality win. Our SOS will continue to move up, likely top 10, over the coming week. FS, Pitt, and Cuse are 21, 2, and 4, respectively, representing a brutal stretch that is being discussed in another thread.

So what does all this mean? Not much but it is below 10 degrees outside so it's something to noodle on while lying in bed and babysitting this guy.

3821

[uh_no]

Duke has jumped up from 23rd before the UVA game to 12th in Kenpom. They were 20th after UVA and 18th going into Miami. 14th in Sagarin, not sure what they were before.

Absolute beatdowns of two top 100 teams including one on the road must have really helped. Also, apparently Kenpom updates his algorithm to ignore expected blow outs so there is more movement as there isn't as much data. Add in Michigan going on a run and easy to see how we've jumped up again.

the big difference was the defense against miami....we had a 102 defense going into the game (which means we give up 102 points per 100 pos against an average offense). miami was putting up a 100.....so for all intents and purposes, we should have given up ~100 points efficiency...we put up an 81....so 20 points better than it was supposed to be....the fact that we "beat them down" didn't apply to this game, since they in fact held our offense in check (relative to average) and our defense vs their offense were evenly matched.

what we need to remember is that offensive and defensive efficiencies for each team are largely arrived at independently...therefore the fact that we blew them out in the end isn't considered in isolation, but only as a consequence of our O outplaying their D, and our D outplaying their O. AFAIK, the "wisconsin effect" considerations are applied on each side separately....so if our O is significantly better than their D coming in, and we score a ton of points, then it will be mitigated by anti-wisconsin effect (same vice versa).

Therefore it shouldn't have applied in either case against miami...since our O was close to the expected value for the game and our D was not supposed to outperform their O.

[CLW]

Should see another nice bump tomorrow after beating #21 in Kenpom by twenty. O put up 127.5 against the #17 D and held the #46 O to 90.0 (which is well below our average)

[uh_no]

Should see another nice bump tomorrow after beating #21 in Kenpom by twenty. O put up 127.5 against the #17 D and held the #46 O to 90.0 (which is well below our average)

he's been updating real time throughout the day, so we should see the results in an hour or so

[DukeHLM'13]

Up to number 9 in the KenPom. Offense still number 2 and the defensive rating continues to rise, moving up to number 65 (with a 98.1 PER). FSU also dropped a few spots to number 24 after the game today.

[hurleyfor3]

Intraday Pomeroy updates. Just in case we didn't have enough things to obsess over.

[uh_no]

Up to number 9 in the KenPom. Offense still number 2 and the defensive rating continues to rise, moving up to number 65 (with a 98.1 PER). FSU also dropped a few spots to number 24 after the game today.

we were a 98.8 before....i was hoping we'd see a slightly larger bump....especially after holding a team averaging 112 to a 90....but alas, that's why i'm an armchair kenpom and not the real thing!

[sporthenry]

the big difference was the defense against miami....we had a 102 defense going into the game (which means we give up 102 points per 100 pos against an average offense). miami was putting up a 100.....so for all intents and purposes, we should have given up ~100 points efficiency...we put up an 81....so 20 points better than it was supposed to be....the fact that we "beat them down" didn't apply to this game, since they in fact held our offense in check (relative to average) and our defense vs their offense were evenly matched.

what we need to remember is that offensive and defensive efficiencies for each team are largely arrived at independently...therefore the fact that we blew them out in the end isn't considered in isolation, but only as a consequence of our O outplaying their D, and our D outplaying their O. AFAIK, the "wisconsin effect" considerations are applied on each side separately....so if our O is significantly better than their D coming in, and we score a ton of points, then it will be mitigated by anti-wisconsin effect (same vice versa).

Therefore it shouldn't have applied in either case against miami...since our O was close to the expected value for the game and our D was not supposed to outperform their O.

I'm not really sure what you are arguing but when you beat a team by 21 that a system predicted you to win by 3, it is a beatdown. One side of the ball obviously outplayed its predicted total by a lot. I didn't really get into which side of the ball, just the fact that Duke won 2 games by significantly more than Kenpom predicted which is why you've seen the recent jump.

The other thing with the Miami game is that his system gives bumps for playing home/away so our offense playing up to efficiency would in actuality, be better than expected. I forget the formula.

ETA: Kenpom adds 1.4% to the home team on each side of the ball and likewise subtracts that from the away team, hence why Duke's jump after FSU isn't as large as some anticipated.

[sagegrouse]

we were a 98.8 before....i was hoping we'd see a slightly larger bump....especially after holding a team averaging 112 to a 90....but alas, that's why i'm an armchair kenpom and not the real thing!


I'm not really sure what you are arguing but when you beat a team by 21 that a system predicted you to win by 3, it is a beatdown. One side of the ball obviously outplayed its predicted total by a lot. I didn't really get into which side of the ball, just the fact that Duke won 2 games by significantly more than Kenpom predicted which is why you've seen the recent jump.

The other thing with the Miami game is that his system gives bumps for playing home/away so our offense playing up to efficiency would in actuality, be better than expected. I forget the formula.

ETA: Kenpom adds 1.4% to the home team on each side of the ball and likewise subtracts that from the away team, hence why Duke's jump after FSU isn't as large as some anticipated.

Taking a broad view of KenPom and other rating systems, they essentially come up with ratings by curve-fitting against past results. If the team's or its opponents' capabilities or performances change -- Duke's performance is improving IMHO (where the H is silent) -- then it has no particular predictive value.

Moreover, to celebrate a KenPom ranking seems kinda meaningless, since it is just an embodiment of past results, but whatever floats your boat!

Kindly, Sage

[sporthenry]

Taking a broad view of KenPom and other rating systems, they essentially come up with ratings by curve-fitting against past results. If the team's or its opponents' capabilities or performances change -- Duke's performance is improving IMHO (where the H is silent) -- then it has no particular predictive value.

Moreover, to celebrate a KenPom ranking seems kinda meaningless, since it is just an embodiment of past results, but whatever floats your boat!

Kindly, Sage

Well, sure things are constantly changing and Kenpom tries to take that into account with emphasis on L10 played. And sure, for each team you can probably point to certain things like Duke going 2 line ups as changing things up, but Kenpom is still a very good predictor.

Beyond that, Kenpom does a good job of indicating whether the changes are real or not. You call this an embodiment of past results but it certainly puts it into perspective. Yeah, we can see with the naked eye Duke's defense has been better but now we have numbers to back up just how better they are especially when you consider to other team. Some look at the Clemson loss as probably a bad loss b/c its Clemson but they are very good defensively and that is a better loss than ND.

And the last 11 champions have all been in the top 20 offensively and defensively. I guess you could argue that maybe this pattern has emerged by chance but I'd disagree with that.

[uh_no]

I'm not really sure what you are arguing but when you beat a team by 21 that a system predicted you to win by 3, it is a beatdown.

Nobody is arguing it's not a beatdown...the argument was that his "anti wisconsin effect" adjustments only apply for a team which you were supposed to already beat by a lot, which was not the case in either of our past two games. We had two games that ought to have been relatively competitive which were not....that carries a lot of predictive value, and is not mitigated as such.

[sagegrouse]

Well, sure things are constantly changing and Kenpom tries to take that into account with emphasis on L10 played. And sure, for each team you can probably point to certain things like Duke going 2 line ups as changing things up, but Kenpom is still a very good predictor.

Beyond that, Kenpom does a good job of indicating whether the changes are real or not. You call this an embodiment of past results but it certainly puts it into perspective. Yeah, we can see with the naked eye Duke's defense has been better but now we have numbers to back up just how better they are especially when you consider to other team. Some look at the Clemson loss as probably a bad loss b/c its Clemson but they are very good defensively and that is a better loss than ND.

And the last 11 champions have all been in the top 20 offensively and defensively. I guess you could argue that maybe this pattern has emerged by chance but I'd disagree with that.

There are no "projections" in any of the systems, merely a different representation of past results.

KenPom, which is very insightful, is provides an analytic view of past performance. It may do a good job of predicting results, but only in the sense that "past is prologue."

[Kedsy]

And the last 11 champions have all been in the top 20 offensively and defensively. I guess you could argue that maybe this pattern has emerged by chance but I'd disagree with that.

I won't get into the rest of your claims, but this one is a myth that just won't go away and is patently untrue. If you look at Pomeroy's rankings pre-tournament, as opposed to the post-tournament numbers he posts, then the numbers tell a different story. I don't have Pomeroy's pre-tournament rankings for all of the past 11 years, but I know for a fact that in 2011 (UConn -- ranked #21 in offense and #31 in defense pre-tournament), 2009 (UNC --ranked #1 in offense and #35 in defense, pre-tournament), and 2005 (UNC, ranked #7 in offense, #30 in defense, pre-tournament), the eventual champions were NOT ranked in the top 20 offensively and defensively going into the tournament. I don't have numbers for 2006 or anything from 2004 backward, but my guess is some of those champions didn't meet your criteria either. So you don't have to argue random chance. The basis of your argument is false.

[Troublemaker]

Up to number 9 in the KenPom. Offense still number 2 and the defensive rating continues to rise

One quick note about that O-ranking. Duke's Offensive Efficiency is currently 123.3, which would be the 4th-best offense of the KenPom era if we maintained it through the end of the season. It just so happens that we rank #2 because right now Creighton has THE best offense of the KenPom era if they maintain their ridiculous 127.2 number (which, of course, they won't. I think Duke might maintain or even improve on our number, though.).

[CLW]

Should see another large bump after winning big on the road at Kenpom #5 Pitt.

[Ichabod Drain]

Should see another large bump after winning big on the road at Kenpom #5 Pitt.

Duke jumps all the way to #4... Pitt falls to #12.

Might be adjusted again before tomorrow morning

[CDu]

Duke jumps all the way to #4... Pitt falls to #12.

Might be adjusted again before tomorrow morning

And when we knock off Syracuse this weekend, we'll probably jump up to #2.

[Troublemaker]

This was an interesting jump.

Our D-ranking actually only improved a little bit from 63 to 60 because we gave up 1.05ppp even though it probably felt better than that while watching the game. But heck, 1.05ppp at Pitt is pretty darn good, kenpom computer! Oh well. Shut down Cuse on Saturday (very possible), and we'll get the big bump we desire here.

Our Offensive Efficiency, though, it soared. Thanks to Dre, we were able to put up 1.29 ppp on Pitt's defense, which is superb. For the season, our adjusted OE is now 125.3, which would be the best offense in the entire kenpom era, except for that little matter that this year's Creighton is still at 127+, and so we're stuck with the #2 ranking not just for the era but for this season as well until Creighton sinks a bit.

[FerryFor50]

This was an interesting jump.

Our D-ranking actually only improved a little bit from 63 to 60 because we gave up 1.05ppp even though it probably felt better than that while watching the game. But heck, 1.05ppp at Pitt is pretty darn good, kenpom computer! Oh well. Shut down Cuse on Saturday (very possible), and we'll get the big bump we desire here.

Our Offensive Efficiency, though, it soared. Thanks to Dre, we were able to put up 1.29 ppp on Pitt's defense, which is superb. For the season, our adjusted OE is now 125.3, which would be the best offense in the entire kenpom era, except for that little matter that this year's Creighton is still at 127+, and so we're stuck with the #2 ranking not just for the era but for this season as well until Creighton sinks a bit.

Hard to imagine Creighton slipping too much. Their remaining schedule is full of mediocre defenses..

@ St Johns (kenpom 63, 59th defense)
@Butler (kenpom 111, 81st defense)
vs Nova (revenge game for Nova) (kenpom 14, 32nd defense)
@Marquette (kenpom 56, 51st in defense)
vs Seton Hall (kenpom 87, 112th defense)
@ Xavier (kenpom 30, 65th defense)
@ Georgetown (kenpom 65, 70th defense)
vs Providence (kenpom 45, 55th defense)

[uh_no]

This was an interesting jump.

Our D-ranking actually only improved a little bit from 63 to 60 because we gave up 1.05ppp even though it probably felt better than that while watching the game. But heck, 1.05ppp at Pitt is pretty darn good, kenpom computer! Oh well. Shut down Cuse on Saturday (very possible), and we'll get the big bump we desire here.

Our Offensive Efficiency, though, it soared. Thanks to Dre, we were able to put up 1.29 ppp on Pitt's defense, which is superb. For the season, our adjusted OE is now 125.3, which would be the best offense in the entire kenpom era, except for that little matter that this year's Creighton is still at 127+, and so we're stuck with the #2 ranking not just for the era but for this season as well until Creighton sinks a bit.

the defensive rating improved .6 I think....that's nothing to shake a stick at regardless of how many teams we happened to pass in the rankings.....plus the defense wasn't spectacular tonight...good yes, but not jaw dropping like the O....i'd imagine we were supposed to put up a 110 or something, and we put up a 107.

by the logic that the defense "only improved from 63 to 60," we could say that the offense didn't improve at all! since we're still #2

[TexHawk]

Hard to imagine Creighton slipping too much. Their remaining schedule is full of mediocre defenses..

@ St Johns (kenpom 63, 59th defense)
@Butler (kenpom 111, 81st defense)
vs Nova (revenge game for Nova) (kenpom 14, 32nd defense)
@Marquette (kenpom 56, 51st in defense)
vs Seton Hall (kenpom 87, 112th defense)
@ Xavier (kenpom 30, 65th defense)
@ Georgetown (kenpom 65, 70th defense)
vs Providence (kenpom 45, 55th defense)

Creighton's AdjO slipped around a full point after their win over Georgetown, and the Hoya D ain't great.. The 128+ they posted after that ridiculous Villanova performance likely isn't sustainable.

[Olympic Fan]

I just checked kenpom at 1:30 am Eastern time and it had Duke at No. 5, not No. 4. Arizona, Creighton, Syracuse and Iowa 1-thu-4.

On the other hand, Jabari Parker is now No. 2 in his POY standings ... behind only McDermott

[loran16]

I just checked kenpom at 1:30 am Eastern time and it had Duke at No. 5, not No. 4. Arizona, Creighton, Syracuse and Iowa 1-thu-4.

On the other hand, Jabari Parker is now No. 2 in his POY standings ... behind only McDermott

We're .0001 in pythag away from #4 Iowa - practically identical. In other words, tiny changes due to changes in opponents caused the "drop", which really only looks like a drop because of how close the two teams are ranked.

[MChambers]

Amazingly, up to #2 in Sagarin (#3 in the Predictor). Amazing what a good five games will do.

By Sagarin, the game at Syracuse is a tossup, taking into account the location.

[77devil]

We are # 2 in Sagarin and moved up 4 or 5 spots to # 12 in the RPI. The next human ranking will be the ESPN power index later today. I believe there is still skepticism in voting community and expect all the voting polls to continue to lag the computers even after we beat Cuse on Saturday.

[jv001]

We are # 2 in Sagarin and moved up 4 or 5 spots to # 12 in the RPI. The next human ranking will be the ESPN power index later today. I believe there is still skepticism in voting community and expect all the voting polls to continue to lag the computers even after we beat Cuse on Saturday.

I love your way of thinking. Beat Cuse and GoDuke!

[Troublemaker]

For the season, our adjusted OE is now 125.3, which would be the best offense in the entire kenpom era, except for that little matter that this year's Creighton is still at 127+, and so we're stuck with the #2 ranking not just for the era but for this season as well until Creighton sinks a bit.

Question: Is it possible that this season's new rules/emphases have helped 3-pt shooting teams somehow? Couldn't help but notice that the top 2 offenses in the kenpom era are currently this season's Creighton and Duke, two teams that are heavily reliant on the 3-ball.

BTW, I don't really believe these are the two best offenses in college basketball of the past 11.5 years:
* Love kenpom, but not married to it. Eye test still part of any eval, of course.
* Rising tide raises all ships. With the new rules/emphases, maybe 125 adjusted OE is the new 120.
* Still only late Jan. If we give it another month and half, maybe Creighton and Duke's OE drop into more reasonable ranges.

Still, I am wondering if 3-pt shooting reliant teams are benefitted somehow. It would be nice, as the changes have hurt our defense WRT forcing turnovers.

[vick]

Question: Is it possible that this season's new rules/emphases have helped 3-pt shooting teams somehow? Couldn't help but notice that the top 2 offenses in the kenpom era are currently this season's Creighton and Duke, two teams that are heavily reliant on the 3-ball.

BTW, I don't really believe these are the two best offenses in college basketball of the past 11.5 years:
* Love kenpom, but not married to it. Eye test still part of any eval, of course.
* Rising tide raises all ships. With the new rules/emphases, maybe 125 adjusted OE is the new 120.
* Still only late Jan. If we give it another month and half, maybe Creighton and Duke's OE drop into more reasonable ranges.

Still, I am wondering if 3-pt shooting reliant teams are benefitted somehow. It would be nice, as the changes have hurt our defense WRT forcing turnovers.

Point 2 is very relevant--this year the NCAA average is 104.2, last year 100.4. Big difference--100.4 would rank 246th this year but 179th last year. So we're good, but we're not that good, not yet anyway.

[uh_no]

We're .0001 in pythag away from #4 Iowa - practically identical. In other words, tiny changes due to changes in opponents caused the "drop", which really only looks like a drop because of how close the two teams are ranked.

shenanigans! WE GOT SNUBBED. we didn't even play and his polls dropped us a spot!

[loran16]

Point 2 is very relevant--this year the NCAA average is 104.2, last year 100.4. Big difference--100.4 would rank 246th this year but 179th last year. So we're good, but we're not that good, not yet anyway.

I'm not sure this is real- offense drops as the season goes on I believe and you're comparing current #s to end of season numbers

[vick]

I'm not sure this is real- offense drops as the season goes on I believe and you're comparing current #s to end of season numbers

It may drop, but I'd be highly surprised if it dropped enough to make it as low as last year. We're about 2/3 of the season already (last year the average team played about 32 games, this year they've already played 20). So in order for the average to fall to 100.4, teams would have to have an efficiency around 94.1 from here on out I think, which strikes me as very unlikely. And of course, even if you are right it would just underscore the problem of directly comparing Creighton or Duke's number right now to previous seasons.

[Des Esseintes]

I'm not sure this is real- offense drops as the season goes on I believe and you're comparing current #s to end of season numbers


It may drop, but I'd be highly surprised if it dropped enough to make it as low as last year. We're about 2/3 of the season already (last year the average team played about 32 games, this year they've already played 20). So in order for the average to fall to 100.4, teams would have to have an efficiency around 94.1 from here on out I think, which strikes me as very unlikely. And of course, even if you are right it would just underscore the problem of directly comparing Creighton or Duke's number right now to previous seasons.

Actually, offense improves (http://kenpom.com/blog/index.php/weblog/entry/our_brave_new_world) as the season goes on. (Table is near the bottom of the blog entry.) We might not notice the increased potency as much because, though efficiency rises, tempo falls. Maybe this year will be different because foul calls will keep drifting back toward historic norms, but that may mean offense just plateaus. Duke has a decent chance to maintain its efficiency through the end of the season and perhaps even improve it. I agree, though, that any discussion of best offenses has to be in terms that measure performance against the environment rather than raw potency.

[hurleyfor3]

Check it out, obsessive dorks! We passed Creighton!

[kAzE]

Check it out, obsessive dorks! We passed Creighton!

Best offense we've had since 1999.

[Des Esseintes]

Best offense we've had since 1999.

2001 and 2002 were pretty insane. I'm not saying this year is not better. But 2001 and 2002 were pretty insane.

[Troublemaker]

Luke Winn's power rankings column is out, and Duke checks in at #11: http://sportsillustrated.cnn.com/college-basketball/news/20140130/power-rankings-arizona-syracuse-wichita-state-florida-kansas/index.html

Duke's blurb deals with Jabari's post play. According to Synergy, Jabari has the 10th highest ppp among major conference players on "post possessions" (which I think means postups), minimum 50 possessions.

[Billy Dat]

Luke Winn's power rankings column is out, and Duke checks in at #11: http://sportsillustrated.cnn.com/college-basketball/news/20140130/power-rankings-arizona-syracuse-wichita-state-florida-kansas/index.html

Duke's blurb deals with Jabari's post play. According to Synergy, Jabari has the 10th highest ppp among major conference players on "post possessions" (which I think means postups), minimum 50 possessions.

Yeah, and I like how the graphic shows that among the players on that list, his post possessions as a % of his overall possessions are the lowest on the list. Even with the potential for him to clog the lane, dare I say....get yee to the post more often, son!!!!!!!

[bob blue devil]

for all of those who harp on the fact that the kenpom rating #s are more important than the ranking #s, here is an nice illustration of your point.

this am, duke is ranked #2 in offense and #60 in defense. for comparison, michigan is ranked #3 in offense and #61 in defense. with 351 teams, that is oddly similar. however, duke is ranked #3 overall, where is michigan is ranked #10. why? because, while duke is only one slot better on offense, it is by a wide margin (125.3 vs. 122.8). as an aside, our offense is simply incredible this year.

[JasonEvans]

for all of those who harp on the fact that the kenpom rating #s are more important than the ranking #s, here is an nice illustration of your point.

this am, duke is ranked #2 in offense and #60 in defense. for comparison, michigan is ranked #3 in offense and #61 in defense. with 351 teams, that is oddly similar. however, duke is ranked #3 overall, where is michigan is ranked #10. why? because, while duke is only one slot better on offense, it is by a wide margin (125.3 vs. 122.8). as an aside, our offense is simply incredible this year.

Duke, Michigan, and Creighton look almost unnaturally similar.

Duke is #2 on O, #60 on D
Creighton is #1 on O and #62 on D
Michigan is #3 on O and #61 on D

That's crazy!

-Jason "I wish Pom could publish rankings from just the past 10 games... as I think recency really matters in figuring out how a team will perform going forward" Evans

[flyingdutchdevil]

Duke, Michigan, and Creighton look almost unnaturally similar.

Duke is #2 on O, #60 on D
Creighton is #1 on O and #62 on D
Michigan is #3 on O and #61 on D

That's crazy!

-Jason "I wish Pom could publish rankings from just the past 10 games... as I think recency really matters in figuring out how a team will perform going forward" Evans

I'd be happy to face Creighton in the tourney again. Also, if we aren't going to be a 2 seed (seems very, very unlikely), can we please have Wichita St. as our 1 seed?

[Des Esseintes]

I'd be happy to face Creighton in the tourney again. Also, if we aren't going to be a 2 seed (seems very, very unlikely), can we please have Wichita St. as our 1 seed?

Because we're a lock for a 1?

[flyingdutchdevil]

Because we're a lock for a 1?

Whoops. I meant 1 seed. I think we'll be a 2 seed. But what's of right now. Hopefully, we'll get a lot more information about our seeding in 36 hours...

[77devil]

Duke, Michigan, and Creighton look almost unnaturally similar.

Duke is #2 on O, #60 on D
Creighton is #1 on O and #62 on D
Michigan is #3 on O and #61 on D

That's crazy!

-Jason "I wish Pom could publish rankings from just the past 10 games... as I think recency really matters in figuring out how a team will perform going forward" Evans

Presumably he gives more weight to recent results in his algorithm but I've never seen an explanation one way or another

[Troublemaker]

Whoops. I meant 1 seed. I think we'll be a 2 seed. But what's of right now. Hopefully, we'll get a lot more information about our seeding in 36 hours...

Yeah, a lot depends on tomorrow's game. If Duke wins, I'd say we're a favorite at that point to get one of the four 1 seeds, given our relatively easier 2nd-half ACC schedule.

[sagegrouse]

Whoops. I meant 1 seed. I think we'll be a 2 seed. But what's of right now. Hopefully, we'll get a lot more information about our seeding in 36 hours...


Yeah, a lot depends on tomorrow's game. If Duke wins, I'd say we're a favorite at that point to get one of the four 1 seeds, given our relatively easier 2nd-half ACC schedule.

I have no predictions about tomorrow or the rest of the season, but I have some thoughts on what it takes to be a #1 seed.


Become the ACC champion through winning the tournament. I can only remember one time when we were #1 and weren't conference champions, which was 1998, when both Duke and UNC were #1 seeds.

Finish near the top of the ACC regular season standings.

[toooskies]

Yeah, a lot depends on tomorrow's game. If Duke wins, I'd say we're a favorite at that point to get one of the four 1 seeds, given our relatively easier 2nd-half ACC schedule.

I'd still rank Arizona, Syracuse, Wichita State, Michigan State, Florida, and San Diego State ahead of us even if we win at Syracuse. And Michigan, Villanova, Creighton, and Kansas are all in the discussion around our seed. Our problem is that we don't have a lot of top-tier wins. We'd really need Michigan, Virginia, UCLA, and Pitt to play above expectations for the human polls to acknowledge our previous wins as top-tier wins. It wouldn't hurt if ND and Clemson make our losses look a little better, too.

We are certainly playing like a top-4 team right now, but a lot of teams can look the part. Otherwise we're just one of a handful of streaky teams.

[flyingdutchdevil]

I'd still rank Arizona, Syracuse, Wichita State, Michigan State, Florida, and San Diego State ahead of us even if we win at Syracuse. And Michigan, Villanova, Creighton, and Kansas are all in the discussion around our seed. Our problem is that we don't have a lot of top-tier wins. We'd really need Michigan, Virginia, UCLA, and Pitt to play above expectations for the human polls to acknowledge our previous wins as top-tier wins. It wouldn't hurt if ND and Clemson make our losses look a little better, too.

We are certainly playing like a top-4 team right now, but a lot of teams can look the part. Otherwise we're just one of a handful of streaky teams.

Right now, you are correct. We are probably a 3 seed.

However, a lot can change in the next 5 weeks. First and foremost, Michigan State and Michigan and the rest of the B1G teams will beat the crap out of each other. Florida will get tested by a few teams in the SEC. Nova and Creighton will lose a few more games. Syracuse will face a tough second half in the ACC.

I see Arizona as a guaranteed 1 seed already (PAC10 is better than most years but still pretty bad). Given that Wichita St. has been playing cupcakes all long and nothing is changing, they will probably be a 1 seed. SDSU is in the same boat as Wichita St, although they have already beaten Kansas (at Kansas) and had a close one with Arizona. I see them as a 2 seed, however.

A lot can change, and a lot will. Let's hope that we see constant improvement from this team. That, and a minimum 2-seed.

[77devil]

I have no predictions about tomorrow or the rest of the season, but I have some thoughts on what it takes to be a #1 seed.


Become the ACC champion through winning the tournament. I can only remember one time when we were #1 and weren't conference champions, which was 1998, when both Duke and UNC were #1 seeds.

Finish near the top of the ACC regular season standings.


2004 when we lost to MD in the tournament final but your point still stands. Hard to see a scenario this year where we are a #1 seed without winning the tournament.

The officiating in the MD game was almost as bad as it was against CT - bad memories.

[Kedsy]

I'd still rank Arizona, Syracuse, Wichita State, Michigan State, Florida, and San Diego State ahead of us even if we win at Syracuse. And Michigan, Villanova, Creighton, and Kansas are all in the discussion around our seed. Our problem is that we don't have a lot of top-tier wins. We'd really need Michigan, Virginia, UCLA, and Pitt to play above expectations for the human polls to acknowledge our previous wins as top-tier wins. It wouldn't hurt if ND and Clemson make our losses look a little better, too.


I see Arizona as a guaranteed 1 seed already (PAC10 is better than most years but still pretty bad). Given that Wichita St. has been playing cupcakes all long and nothing is changing, they will probably be a 1 seed.

I remember in mid-January, 2010, lot of people on these boards were arguing that Duke had absolutely no chance for a #1 seed, in part because Texas (at that time 17-0 and #1 in the country) had one of the top seeds all locked up.

Of course, Duke did get a #1 seed. Texas managed a #8 seed.

[TexHawk]

Right now, you are correct. We are probably a 3 seed.

However, a lot can change in the next 5 weeks. First and foremost, Michigan State and Michigan and the rest of the B1G teams will beat the crap out of each other. Florida will get tested by a few teams in the SEC. Nova and Creighton will lose a few more games. Syracuse will face a tough second half in the ACC.

I see Arizona as a guaranteed 1 seed already (PAC10 is better than most years but still pretty bad). Given that Wichita St. has been playing cupcakes all long and nothing is changing, they will probably be a 1 seed. SDSU is in the same boat as Wichita St, although they have already beaten Kansas (at Kansas) and had a close one with Arizona. I see them as a 2 seed, however.

A lot can change, and a lot will. Let's hope that we see constant improvement from this team. That, and a minimum 2-seed.

I think the opposite is going to happen regarding Florida. I think they are a scary good team, but they aren't going to impress a lot of people beating up on the SEC, especially with UK looking worse and worse by the day.

On the topic of dork polls, teamrankings (http://www.teamrankings.com/ncaa-tournament/bracketology/) has Duke tied with Nova for probability of getting the last #1 seed. They have been down on Cuse all year, just recently bumping them up from a 3 seed to a 2.

[wk2109]

I remember in mid-January, 2010, lot of people on these boards were arguing that Duke had absolutely no chance for a #1 seed, in part because Texas (at that time 17-0 and #1 in the country) had one of the top seeds all locked up.

Of course, Duke did get a #1 seed. Texas managed a #8 seed.

Yeah and even just last year, Louisville was ranked #12 in the 1/28/13 AP poll -- I'm guessing no one had them pegged as a #1 seed, let alone the overall #1. Michigan was ranked #1 in that same poll and ended up with a #4 seed.

After Duke beat Miami and demolished Carolina last March, they were considered a #1 seed. Then came one bad game against Maryland in the ACC tournament and we end up as Louisville's #2 seed. Teams can have their seeding majorly affected by a game or two at the end of the regular season or their conference tournaments. It's WAY too early to declare anyone a lock for a #1 seed in January.

[NYBri]

Beat FSU - check
Beat Pitt - check
Syracuse - next game. focus on the zone
After that - not applicable

[ice-9]

I'm sure Wichita State is a good team (#9 on KenPom) but can someone explain to me the love fest for the Shockers? I looked at their record and their highlight wins are against Tennessee, Alabama and St Louis. The three are good teams, but none of them are ranked and for good reason. They're still nice wins, sure, but how are we talking about this as the profile of a #1 seed? I just don't see the resume to justify Wichita State's lofty ranking -- #4 AP and #3 Coaches -- much less a #1 seed.

I can at least understand San Diego State's ranking. They beat Creighton, Marquette and Kansas and lost only to Arizona. I get their #5 ranking.

Wichita State...I don't get. Even taking a step back from records and rankings, just from a basketball point of view, why are they supposed to be so good? Why are they more highly regarded than Michigan State or Duke or San Diego State? Do people truly believe that Wichita State would beat those teams on a neutral court? Is there Damien Lillard type of talent on the Shockers?

I realize a Duke forum may not have the answer... just venting my confusion. :)

Just feels like some kind of massive groupthink party that nobody invited me to.

[Kedsy]

I'm sure Wichita State is a good team (#9 on KenPom) but can someone explain to me the love fest for the Shockers? I looked at their record and their highlight wins are against Tennessee, Alabama and St Louis. The three are good teams, but none of them are ranked and for good reason. They're still nice wins, sure, but how are we talking about this as the profile of a #1 seed? I just don't see the resume to justify Wichita State's lofty ranking -- #4 AP and #3 Coaches -- much less a #1 seed.

I can at least understand San Diego State's ranking. They beat Creighton, Marquette and Kansas and lost only to Arizona. I get their #5 ranking.

Wichita State...I don't get. Even taking a step back from records and rankings, just from a basketball point of view, why are they supposed to be so good? Why are they more highly regarded than Michigan State or Duke or San Diego State? Do people truly believe that Wichita State would beat those teams on a neutral court? Is there Damien Lillard type of talent on the Shockers?

I realize a Duke forum may not have the answer... just venting my confusion. :)

Just feels like some kind of massive groupthink party that nobody invited me to.

You already know the answer: because that's how these polls work. A team rarely drops in the rankings unless they lose and Wichita State hasn't lost yet. On the other hand, we're talking about a team that made the Final Four last season and is undefeated this season. On some level they ought to get some love from the polls.

[ice-9]

On the other hand, we're talking about a team that made the Final Four last season and is undefeated this season. On some level they ought to get some love from the polls.

Ah, I forgot about that element -- that explains some of the love fest.

I wondered if it was basically the same team this year as last, and looked at the roster. Of the six players who played over 20 minutes per game the Shockers get three back this year; so not the same, but not wholly different. That said the Shockers do play a deep bench (11 in double figures last year, 9 this year) so they are probably less immune to roster changes year-to-year than most.

OK, I understand the love fest a little bit more. Still, I'd put money on Duke against them any day.

[throatybeard]

I'm sure Wichita State is a good team (#9 on KenPom) but can someone explain to me the love fest for the Shockers? I looked at their record and their highlight wins are against Tennessee, Alabama and St Louis. The three are good teams, but none of them are ranked and for good reason.

Kedsy is on it. SLU has been ranked, above Duke even one week, I think. Alabama is not good. Not even close.

I can't tell how good SLU is, and I read the paper in the morning.

[Troublemaker]

Not that this is any consolation whatsoever, but we did pull ahead of Syracuse in the KenPom rankings.

We're #2 now with the #1 offense (ridiculous 127+ OE) and #81 defense.

[Kfanarmy]

Not that this is any consolation whatsoever, but we did pull ahead of Syracuse in the KenPom rankings.

We're #2 now with the #1 offense (ridiculous 127+ OE) and #81 defense.

That first number is awesome. That second number is REALLY worrisome though!

[COYS]

That first number is awesome. That second number is REALLY worrisome though!

While our defensive ranking dropped after losing to Cuse tonight, we've otherwise been playing much better defense. Also, Cuse did their worst damage to our defensive efficiency after Jabari and Amile fouled out and k made the choice to go all out on offense (the right call, in my opinion). Also, Cuse hit a lot of contested shots. I'm satisfied with the defensive improvement and chalk up tonight's minor setback to Cuse being really hot even when we played good defense and then, of course, Cuse getting freebie twos and free throws in overtime after we lost our front court.

[MChambers]

Not that this is any consolation whatsoever, but we did pull ahead of Syracuse in the KenPom rankings.

We're #2 now with the #1 offense (ridiculous 127+ OE) and #81 defense.
Now we're #3. (Maybe Arizona's late night loss dropped us down a little?)

We #3 behind Syracuse, by the slimmest of margins, in Pomeroy, and #2 in Sagarin's predictor (#3 in the blended category).

Sure wish we could improve our defense, however (#82).

[Olympic Fan]

I was going to post about the odd fluctuation in the Pomeroy ratings -- we briefly moved ahead of Syracuse into the No. 2 spot late last night, but were back at No. 3 this morning. I also think Arizona's loss may have done that -- it's damn close -- in fact, both Syracuse and Duke have the exact Pyth number: .9468.

I'm not going to complain about being ranked behind a team that has a better record and beat us head-to-head -- even if it was at home and based on a couple of questionable calls.

One other note: As of this morning, we jumped to No. 6 in the RPI.

It will be interesting to see what happens in the human polls -- I'll bet we gain a place or two in the coaches poll and drop a place or two in the AP poll.

[wk2109]

It will be interesting to see what happens in the human polls -- I'll bet we gain a place or two in the coaches poll and drop a place or two in the AP poll.

I can see Duke actually rising in the AP poll. Six of the seven teams directly ahead of Duke lost a game, including Wisconsin which lost two. It also helps that Duke's game was super hyped and the media storyline is that it was an instant classic. I can't see Saint Louis (ranked #20 with wins over Richmond and George Mason) or Creighton (ranked #21 with a 3 point home win against St. John's) jumping Duke in the polls.

I like that Duke is ranked #3 on kenpom, but the fact that Creighton is ranked #4 kind of takes the shine off Duke's high ranking for me. Hopefully Duke can put together a nice stretch of defensive games and possibly move into the top spot.

[Ichabod Drain]

Was looking at Kenpom this morning and I found it interesting that it gives us a better chance of beating Syracuse at home (65%) than UNC on the road (61%). The UNC road game is our lowest percent chance to win a game the rest of the season. The UNC home game jumps up to 85% chance to win. I know the numbers don't really mean anything it's just interesting to think about.

[toooskies]

I like that Duke is ranked #3 on kenpom, but the fact that Creighton is ranked #4 kind of takes the shine off Duke's high ranking for me. Hopefully Duke can put together a nice stretch of defensive games and possibly move into the top spot.

What do you have against Creighton? They're a member of the Big East, they've played a harder schedule than Syracuse, and they have the soon-to-be NPOY in McDermott. This is not the mid-major Creighton of yesteryear.

[hurleyfor3]

Y'all realize that a difference of .0016 is equivalent to the effect on batting average of an MLB player getting one more or fewer hit during an entire season, right? (I mean, I know *I'm* right, cuz it's math.)

[TruBlu]

Y'all realize that a difference of .0016 is equivalent to the effect on batting average of an MLB player getting one more or fewer hit during an entire season, right? (I mean, I know *I'm* right, cuz it's math.)

OK, what would one more HB three pointer a game work out to? (In case Jay Bilas reads DBR.)

[wk2109]

What do you have against Creighton? They're a member of the Big East, they've played a harder schedule than Syracuse, and they have the soon-to-be NPOY in McDermott. This is not the mid-major Creighton of yesteryear.

I honestly don't have anything against Creighton, and truth be told I haven't seen them play this year. But I like to think of kenpom as a somewhat legitimate ranking of all the teams in the nation, and I don't think Creighton is the 4th best team in the country. Because of that, I don't know how much value to give to the #3 'ranking' that the system gives Duke.

[Troublemaker]

I honestly don't have anything against Creighton, and truth be told I haven't seen them play this year. But I like to think of kenpom as a somewhat legitimate ranking of all the teams in the nation, and I don't think Creighton is the 4th best team in the country. Because of that, I don't know how much value to give to the #3 'ranking' that the system gives Duke.

What about other surrounding teams like Arizona at #1, Syracuse at #2, Florida at #5, etc.

One outlier, if Creighton is indeed an outlier, doesn't sink the whole system. Bathwater bad, baby good.

[wk2109]

What about other surrounding teams like Arizona at #1, Syracuse at #2, Florida at #5, etc.

One outlier, if Creighton is indeed an outlier, doesn't sink the whole system. Bathwater bad, baby good.

True. And if Creighton is an outlier, hopefully Duke isn't one too.

[Ichabod Drain]

I honestly don't have anything against Creighton, and truth be told I haven't seen them play this year. But I like to think of kenpom as a somewhat legitimate ranking of all the teams in the nation, and I don't think Creighton is the 4th best team in the country. Because of that, I don't know how much value to give to the #3 'ranking' that the system gives Duke.

No offense but saying you haven't seen them play this year and then saying anything about how good or not good they are is a little ironic.

[toooskies]

Creighton is legit. They beat Villanova by 28, and Villanova is a really good team. They might be overrated on KenPom, but when they're playing their best basketball they are absolutely one of the top teams in the country. Teams with strong post play might give them problems, but you can say the same about us.

That said, I'm not sold that the tweaks the KenPom system made in the offseason were to the system's benefit. Teams are still changing ratings very, very quickly. And not just in ranking terms-- it wasn't all that long ago that Duke was below .9000 in the rankings, and we're already up to .9453. That's a big shift, to move from being rated just above Florida State to even with Syracuse.

[tbyers11]

No offense but saying you haven't seen them play this year and then saying anything about how good or not good they are is a little ironic.

Valid point. I've seen Creighton play parts of three games and I think they are a good team but nowhere near #2 (as they currently stand in Ken Pom). I think their Pom rating is a bit inflated by their 28 point beat down of Villanova on the road 2 weeks ago. I honestly think Creighton was better last year with Echenique in the middle. They are extremely dangerous when they are hot from 3 (@ Nova) but extremely vulnerable when they are not (@ Providence).

I think they also benefit from the move to the new Big East from the Missouri Valley. One, the league is sort of down as Georgetown, Xavier, Butler and Marquette are having down years and only Villanova is playing better than projected. In the Valley, Creighton was also kind of like Duke in the sense that when they played on the road they were everyone's Super Bowl. I don't see Villanova or Marquette fans (or players) getting excited for a Creighton game the way that Indiana St or Northern Iowa did.

I think the polls are actually a bit more accurate (AP #12) on this one as I kind of see Creighton as a 3-5 seed when things shake out. After a likely easy home game against DePaul, they have (@ St Johns, @ Butler, and Villanova at home). I think we'll know more about them that after that stretch. That being said, I don't want to see them in the Tourney b/c McDermott is that good and could single-handedly win a game they probably shouldn't in a one-and-done scenario.

[uh_no]

Creighton is legit. They beat Villanova by 28, and Villanova is a really good team. They might be overrated on KenPom, but when they're playing their best basketball they are absolutely one of the top teams in the country. Teams with strong post play might give them problems, but you can say the same about us.

That said, I'm not sold that the tweaks the KenPom system made in the offseason were to the system's benefit. Teams are still changing ratings very, very quickly. And not just in ranking terms-- it wasn't all that long ago that Duke was below .9000 in the rankings, and we're already up to .9453. That's a big shift, to move from being rated just above Florida State to even with Syracuse.

but it matches a big shift in how well the team has been playing.....so i'm not wholly concerned...

i'm not sure anecdotal evidence from an eye test will help us evaluate whether there is actually more volatility or not...

[wk2109]

No offense but saying you haven't seen them play this year and then saying anything about how good or not good they are is a little ironic.

There are plenty of teams I haven't seen, but I think I can still legitimately say something about how good or not good they are by looking at their schedule and results. Creighton strikes me as a solid team built around a really great player, but not the #4 team (or today the #2 team) in the nation. I understand there are flaws to my reasoning, but that's how I feel. Nothing about what they've done screams top-5 in America to me. This includes the Villanova game, which seems like the fluke result of having 4 guys combine to shoot 20/30 from three. But, of course, I could be wrong.

[SirBlueDevil]

Just throwing this out there blue nation, where do some of you think the Orangemen will incur their first "L" of the season?

Hosting Clemson, going on the road two games later to both Pitt then Nc State or when they meet up with our Devils? In trying to be both fair and rational, i'm thinking either Clemson at home or Pitt on road trip either way i'm definitly seeing at least two if not three "L" before our rematch!

Not dissing the BC program or it's fan base, i just don't see them putting up a big match when the Orangemen host them after the boys of NC State. I could be wrong because i'm no svengali when it comes to predicting the outcome of any of these last stance games but wanted to get some of your thoughts all the same.

Duke Blue Or Bust!

[TexHawk]

I honestly think Creighton was better last year with Echenique in the middle.

This is a good point. Aren't you all uniquely qualified to talk about the talent on Creighton? Their Top 8 players in their rotation this year (by mpg) are all seniors or juniors, and Duke beat them pretty handily last March. Granted, experience means something, players get better, and overall CBB talent can be quite different year-to-year, but they were a 7 seed in 2013. I guess it's not completely crazy to see them jumping to a 3-5 seed this year, but I don't see the jump to a #1 seed like some are suggesting.

[Ichabod Drain]

There are plenty of teams I haven't seen, but I think I can still legitimately say something about how good or not good they are by looking at their schedule and results. Creighton strikes me as a solid team built around a really great player, but not the #4 team (or today the #2 team) in the nation. I understand there are flaws to my reasoning, but that's how I feel. Nothing about what they've done screams top-5 in America to me. This includes the Villanova game, which seems like the fluke result of having 4 guys combine to shoot 20/30 from three. But, of course, I could be wrong.

Exactly! But why stop at their schedule and results? Let's look at points per possession, defensive rebounding percentages, player usage, and all other kinds of basketball data analysis. Which is exactly what Kenpom does. Personally I don't know how good Creighton is and they very well may not be a top five team. But Ken Pomeroy has data from every possession of every game to back up his rankings. It is perfect? Absolutely not... but he's trying to get as accurate as humanly possible.

[johnb]

I've seen Creighton play multiple times this season and have a fairly clear perspective on their strengths and weaknesses. My summary: they haven't missed a three pointer all season, and they continued to make them even while crushing an elite team.

Admittedly, I've only seen them play via Sportscenter highlights, so I might be missing something, but if they can continue to never miss their 30 footers, they belong in the top 5.

[Olympic Fan]

Just throwing this out there blue nation, where do some of you think the Orangemen will incur their first "L" of the season?

Hosting Clemson, going on the road two games later to both Pitt then Nc State or when they meet up with our Devils? In trying to be both fair and rational, i'm thinking either Clemson at home or Pitt on road trip either way i'm definitly seeing at least two if not three "L" before our rematch!

Not dissing the BC program or it's fan base, i just don't see them putting up a big match when the Orangemen host them after the boys of NC State. I could be wrong because i'm no svengali when it comes to predicting the outcome of any of these last stance games but wanted to get some of your thoughts all the same.

Duke Blue Or Bust!

They ain't losing to Clemson in the dome.

And they ain't losing to NC State, BC or Georgia Tech in the Dome either.

The next realistic chance they have to lose is at Pitt on Feb. 12.

Then they have a three-game road stretch at Duke (Feb. 22), at Maryland (Feb. 24) and at Virginia (Mar. 1) where I expect them to lose at least one and probably two.

They finish at FSU, which could be a problem for the 'Cuse if FSU regains the form it showed two weeks ago.

[Troublemaker]

They ain't losing to Clemson in the dome.

And they ain't losing to NC State, BC or Georgia Tech in the Dome either.

The next realistic chance they have to lose is at Pitt on Feb. 12.

Then they have a three-game road stretch at Duke (Feb. 22), at Maryland (Feb. 24) and at Virginia (Mar. 1) where I expect them to lose at least one and probably two.

They finish at FSU, which could be a problem for the 'Cuse if FSU regains the form it showed two weeks ago.

Yeah, the Syracuse at UVA game will be interesting from a rooting standpoint (and likely basketball standpoint, too). I find it likely that I'll be rooting for Cuse in that game because I'm a greedy Blue Devil fan who is still dreaming about a piece of the regular season title.

UVA can get to 3 losses by losing to Syracuse and one of: @Clemson and @Maryland

Cuse can get to 3 losses by losing @Pitt, @Duke, and one of: @Maryland and @FSU

If Duke wins out, I think that's the path to a share. Pretty sure that gives Duke the 1 seed in the ACCT as well.

Incidentally, the Terps could be Duke's biggest friend in the 2nd half of the ACC schedule. Starting tonight with their game in Chapel Hill.

[jv001]

I've seen Creighton play multiple times this season and have a fairly clear perspective on their strengths and weaknesses. My summary: they haven't missed a three pointer all season, and they continued to make them even while crushing an elite team.

Admittedly, I've only seen them play via Sportscenter highlights, so I might be missing something, but if they can continue to never miss their 30 footers, they belong in the top 5.

You do realize that Sportscenter highlights only show their makes(most of the time) and not their misses. :cool: GoDuke!

[gotoguy]

You do realize that Sportscenter highlights only show their makes(most of the time) and not their misses. :cool: GoDuke!

That Creighton 3-point expertise extends to the NBA where the Hawks Kyle Korver has set the NBA record this year for consecutive games making a three.


As to never missing, I think JohnB's comment was tongue in cheek.;)

[Kedsy]

Since Pomeroy altered his formula this season, supposedly in an effort to remedy the so-called "Wisconsin Effect," I've been thinking about what's really going on with the Badgers and their ilk. To briefly recap, for 7 of the past 8 years, Pomeroy ranked Wisconsin in the top 12. But he took a lot of flack because people looked at a bunch of slow, geeky-looking white guys who lost at least 9 games in each of the past 5 seasons and said, "how come you rank them so high?" It got so bad that Pomeroy felt the need to post a "Wisconsin FAQ (http://kenpom.com/blog/index.php/weblog/on_wisconsin_the_faq/)" to explain.

So the first thing I wondered is was there really a Wisconsin problem to begin with. Here's the basics on Wisconsin for the past 7 seasons:



Year Pom Rk AP Rk Record Big 10 NCAA Result
2013 12 18 23-12 12-6 5 seed, lost to 12 seed Mississippi in round of 64
2012 5 14 26-10 12-6 4 seed, lost to 1 seed Syracuse in round of 16
2011 7 16 25-9 13-5 4 seed, lost to 8 seed Butler in round of 16
2010 9 16 24-9 13-5 4 seed, lost to 12 seed Cornell in round of 32
2009 29 unr 19-12 10-8 12 seed, beat 5 seed Florida State, lost to 4 seed Xavier in round of 32
2008 5 6 31-5 16-2 3 seed, lost to 10 seed Davidson in round of 16
2007 8 6 30-6 13-3 2 seed, lost to 7 seed UNLV in round of 32


Certainly no "problem" in 2007 and 2008 -- Pomeroy's rank was very close to the AP rank and Wisconsin had a great overall and Big 10 record -- or 2009 when Pomeroy didn't think much of them anyway. But from 2010 to 2013, Pomeroy's rank diverged from the AP and Wisconsin lost a lot of games, although their Big 10 record was reasonably impressive considering the Big 10 was generally considered to be the best conference in college basketball. So what was the problem? Besides Wisconsin being slow and mostly white, the problem would appear to be the Badger's NCAA performance. Although I'm not sure it really was. Despite Pomeroy's high rating, the NCAA seeding committee gave Wisconsin a 4 (or 5) seed in each of the past four seasons, so the Sweet 16 was their expected outcome. Granted they only achieved that twice, but losing to NCAA finalist Butler in 2011 shouldn't have been cause for embarrassment, and Cornell was pretty hot in 2010 (having also beaten Pomeroy's #15 team, Temple). And in 2012, the year in which the phrase "Wisconsin Effect" hit dorkdom, Wisconsin lost in the Sweet 16 to the #2 team in the final AP poll (Syracuse, #6 in Pomeroy's rankings) by one point. Which to me says, (a) if they lost by one point to the #2 team in the country (or even the #6), who's to say Wisconsin wasn't a legitimate #5; and (b) if Wisconsin had hit one more basket and beaten Syracuse and moved on to play Ohio State (a team they'd split with during the regular season) for the Final Four, would there even be a "Wisconsin Problem"?

Whatever. Assuming the problem exists, my next question was did Pomeroy solve it? Supposedly the issue was Wisconsin inflated its ranking by beating up on weak teams. But here's a table showing Wisconsin's final ranking under the old Pomeroy formula, compared to the new formula:



Year Old Pom Rk New Pom Rk
2013 12 13
2012 5 7
2011 7 8
2010 9 11
2009 29 28
2008 5 5
2007 8 8


Hard to imagine that Wisconsin was outrageously overranked at #5 in 2012 but fine at #7, right? So if there was a problem, I'm thinking Mr. Pomeroy didn't fix it. So maybe the problem isn't that the Badgers beat up too greedily on hapless opponents, maybe it's something else.

Which raised my current question: again pre-supposing a problem exists, what else could it be? What is it about Wisconsin that could fool Pomeroy's otherwise excellent system?

Well, the two things Wisconsin is famous for are playing a very slow tempo and playing a pretty weak non-conference schedule. So I looked through Pomeroy's historical rankings for teams with a tempo in the bottom 50 of the country and a NCSOS in the bottom 150. Since the "problem" seems to date back to 2010, I looked at the last four seasons:



Year Team Pom Rk Tempo Rk NCSOS rk Seed Result
2013 Georgetown 12 340 298 2 Lost to 15 seed in round of 64
2013 Wisconsin 13 318 241 5 Lost to 12 seed in round of 64
2013 Kansas St 21 299 232 4 Lost to 13 seed in round of 64
2013 Notre Dame 37 319 322 7 Lost to 10 seed in round of 64
2012 Wisconsin 7 345 271 4 Lost to 1 seed in round of 16
2012 Michigan 23 329 204 4 Lost to 13 seed in round of 64
2012 Notre Dame 37 328 239 7 Lost to 10 seed in round of 64
2011 Pitt 3 318 269 1 Lost to 8 seed in round of 32
2011 Wisconsin 8 344 225 4 Lost to 8 seed in round of 16
2011 Cincinnati 21 299 335 6 Lost to 3 seed in round of 32
2011 Texas A&M 49 326 327 7 Lost to 10 seed in round of 64
2010 Pitt 25 322 229 3 Lost to 6 seed in round of 32
2010 Marquette 30 304 308 6 Lost to 11 seed in round of 64
2010 Notre Dame 31 312 244 6 Lost to 11 seed in round of 64


So, in the last four tournaments, there were 14 teams seeded 7 or better with tempo in the lowest 50 and NCSOS in the bottom 150. Of those 14 teams, twelve of them lost to teams seeded 3 or more spots below them, including 9 first round upsets and a #8 beating a #1. Ironically, the only team in the group to even get to the Sweet 16 was Wisconsin, which did it twice.

Teams that didn't quite make the cut:

2012 Georgetown (300 tempo; 152 NCSOS), 3 seed lost to 11 seed in round of 32
2012 St. Mary's (261 tempo; 233 NCSOS), 7 seed lost to 10 seed in round of 64
2011 Notre Dame (273 tempo; 263 NCSOS), 2 seed lost to 10 seed in round of 32
2010 Wisconsin (340 tempo; 131 NCSOS), 4 seed lost to 12 seed in round of 32

I used the new Pomeroy formula and end-of-tournament rankings, which might skew things a bit, but probably not too much since teams don't change tempo or non-conference schedule all that much in the tournament. Also, full disclosure, if you go back the 7 seasons from 2003 to 2009, there are only 7 such teams and only 1 of them lost to team seeded 3 spots or more below them (and none of the 7 teams went past the Sweet 16, but 5 of the 7 reached the Sweet 16). Still, Wisconsin doesn't appear to have been a problem before 2010, and there were so few teams that met the criteria in those years (1.0 per year as opposed to 3.5 per year in the later sample), so I'm running with it.

Now just because they all lost in the NCAAT doesn't mean they're overrated in Pomeroy, but it might. And either way it's pretty amazing, right? I don't know about you, but when I'm filling out my bracket this year I'll be looking for teams with slow tempos and bad non-conference schedules as potential upset victims. For what it's worth, there are three teams currently in danger of meeting my criteria in the upcoming NCAA tournament: Syracuse (345 tempo; 218 NCSOS), Cincinnati (318 tempo; 315 NCSOS), and Pittsburgh (290 tempo; 285 NCSOS). You heard it here first.

[Wander]

Now just because they all lost in the NCAAT doesn't mean they're overrated in Pomeroy, but it might. And either way it's pretty amazing, right? I don't know about you, but when I'm filling out my bracket this year I'll be looking for teams with slow tempos and bad non-conference schedules as potential upset victims. For what it's worth, there are three teams currently in danger of meeting my criteria in the upcoming NCAA tournament: Syracuse (345 tempo; 218 NCSOS), Cincinnati (318 tempo; 315 NCSOS), and Pittsburgh (290 tempo; 285 NCSOS). You heard it here first.

I've always suspected it was a tempo issue, but never actually took the time to look into it. One suggestion I would have is to use Pomeroy's "Luck" rankings. Luck is a measure of how his predicted records differ from actual records. It could be due to: (a) teams and coaches that are genuinely better or worse at winning close games, (b) actual statistical randomness, or what I am hypothesizing here, (c) flaws in the kenpom rankings. If there's a correlation between tempo and luck, ie slow teams are consistently unluckier over a few years, then I think we've got a good case.

It might make sense - you won't find anyone who likes kenpom more than I do, but I'm not sure the concept of tempo-free is perfect for every statistic. This is already taken into account in fouling, as he calculates foul rates per unit time rather than per possession, but maybe there are other subtler examples.

[Kedsy]

I've always suspected it was a tempo issue, but never actually took the time to look into it. One suggestion I would have is to use Pomeroy's "Luck" rankings. Luck is a measure of how his predicted records differ from actual records. It could be due to: (a) teams and coaches that are genuinely better or worse at winning close games, (b) actual statistical randomness, or what I am hypothesizing here, (c) flaws in the kenpom rankings. If there's a correlation between tempo and luck, ie slow teams are consistently unluckier over a few years, then I think we've got a good case.

It might make sense - you won't find anyone who likes kenpom more than I do, but I'm not sure the concept of tempo-free is perfect for every statistic. This is already taken into account in fouling, as he calculates foul rates per unit time rather than per possession, but maybe there are other subtler examples.

Unfortunately, it doesn't look that way. Here's the table with luck rank added in. Wisconsin seems consistently "unlucky," but for most of the others it looks like their luck ran out when they hit the NCAAT.



Year Team Pom Rk Tempo Rk NCSOS rk Luck rk Seed Result
2013 Georgetown 12 340 298 80 2 Lost to 15 seed in round of 64
2013 Wisconsin 13 318 241 266 5 Lost to 12 seed in round of 64
2013 Kansas St 21 299 232 50 4 Lost to 13 seed in round of 64
2013 Notre Dame 37 319 322 58 7 Lost to 10 seed in round of 64
2012 Wisconsin 7 345 271 214 4 Lost to 1 seed in round of 16
2012 Michigan 23 329 204 42 4 Lost to 13 seed in round of 64
2012 Notre Dame 37 328 239 86 7 Lost to 10 seed in round of 64
2011 Pitt 3 318 269 236 1 Lost to 8 seed in round of 32
2011 Wisconsin 8 344 225 192 4 Lost to 8 seed in round of 16
2011 Cincinnati 21 299 335 79 6 Lost to 3 seed in round of 32
2011 Texas A&M 49 326 327 55 7 Lost to 10 seed in round of 64
2010 Pitt 25 322 229 76 3 Lost to 6 seed in round of 32
2010 Marquette 30 304 308 312 6 Lost to 11 seed in round of 64
2010 Notre Dame 31 312 244 318 6 Lost to 11 seed in round of 64

[tommy]

Since Pomeroy altered his formula this season, supposedly in an effort to remedy the so-called "Wisconsin Effect," I've been thinking about what's really going on with the Badgers and their ilk. To briefly recap, for 7 of the past 8 years, Pomeroy ranked Wisconsin in the top 12. But he took a lot of flack because people looked at a bunch of slow, geeky-looking white guys who lost at least 9 games in each of the past 5 seasons and said, "how come you rank them so high?" It got so bad that Pomeroy felt the need to post a "Wisconsin FAQ (http://kenpom.com/blog/index.php/weblog/on_wisconsin_the_faq/)" to explain.

So the first thing I wondered is was there really a Wisconsin problem to begin with. Here's the basics on Wisconsin for the past 7 seasons:



Year Pom Rk AP Rk Record Big 10 NCAA Result
2013 12 18 23-12 12-6 5 seed, lost to 12 seed Mississippi in round of 64
2012 5 14 26-10 12-6 4 seed, lost to 1 seed Syracuse in round of 16
2011 7 16 25-9 13-5 4 seed, lost to 8 seed Butler in round of 16
2010 9 16 24-9 13-5 4 seed, lost to 12 seed Cornell in round of 32
2009 29 unr 19-12 10-8 12 seed, beat 5 seed Florida State, lost to 4 seed Xavier in round of 32
2008 5 6 31-5 16-2 3 seed, lost to 10 seed Davidson in round of 16
2007 8 6 30-6 13-3 2 seed, lost to 7 seed UNLV in round of 32


Certainly no "problem" in 2007 and 2008 -- Pomeroy's rank was very close to the AP rank and Wisconsin had a great overall and Big 10 record -- or 2009 when Pomeroy didn't think much of them anyway. But from 2010 to 2013, Pomeroy's rank diverged from the AP and Wisconsin lost a lot of games, although their Big 10 record was reasonably impressive considering the Big 10 was generally considered to be the best conference in college basketball. So what was the problem? Besides Wisconsin being slow and mostly white, the problem would appear to be the Badger's NCAA performance. Although I'm not sure it really was. Despite Pomeroy's high rating, the NCAA seeding committee gave Wisconsin a 4 (or 5) seed in each of the past four seasons, so the Sweet 16 was their expected outcome. Granted they only achieved that twice, but losing to NCAA finalist Butler in 2011 shouldn't have been cause for embarrassment, and Cornell was pretty hot in 2010 (having also beaten Pomeroy's #15 team, Temple). And in 2012, the year in which the phrase "Wisconsin Effect" hit dorkdom, Wisconsin lost in the Sweet 16 to the #2 team in the final AP poll (Syracuse, #6 in Pomeroy's rankings) by one point. Which to me says, (a) if they lost by one point to the #2 team in the country (or even the #6), who's to say Wisconsin wasn't a legitimate #5; and (b) if Wisconsin had hit one more basket and beaten Syracuse and moved on to play Ohio State (a team they'd split with during the regular season) for the Final Four, would there even be a "Wisconsin Problem"?

Whatever. Assuming the problem exists, my next question was did Pomeroy solve it? Supposedly the issue was Wisconsin inflated its ranking by beating up on weak teams. But here's a table showing Wisconsin's final ranking under the old Pomeroy formula, compared to the new formula:



Year Old Pom Rk New Pom Rk
2013 12 13
2012 5 7
2011 7 8
2010 9 11
2009 29 28
2008 5 5
2007 8 8


Hard to imagine that Wisconsin was outrageously overranked at #5 in 2012 but fine at #7, right? So if there was a problem, I'm thinking Mr. Pomeroy didn't fix it. So maybe the problem isn't that the Badgers beat up too greedily on hapless opponents, maybe it's something else.

Which raised my current question: again pre-supposing a problem exists, what else could it be? What is it about Wisconsin that could fool Pomeroy's otherwise excellent system?

Well, the two things Wisconsin is famous for are playing a very slow tempo and playing a pretty weak non-conference schedule. So I looked through Pomeroy's historical rankings for teams with a tempo in the bottom 50 of the country and a NCSOS in the bottom 150. Since the "problem" seems to date back to 2010, I looked at the last four seasons:



Year Team Pom Rk Tempo Rk NCSOS rk Seed Result
2013 Georgetown 12 340 298 2 Lost to 15 seed in round of 64
2013 Wisconsin 13 318 241 5 Lost to 12 seed in round of 64
2013 Kansas St 21 299 232 4 Lost to 13 seed in round of 64
2013 Notre Dame 37 319 322 7 Lost to 10 seed in round of 64
2012 Wisconsin 7 345 271 4 Lost to 1 seed in round of 16
2012 Michigan 23 329 204 4 Lost to 13 seed in round of 64
2012 Notre Dame 37 328 239 7 Lost to 10 seed in round of 64
2011 Pitt 3 318 269 1 Lost to 8 seed in round of 32
2011 Wisconsin 8 344 225 4 Lost to 8 seed in round of 16
2011 Cincinnati 21 299 335 6 Lost to 3 seed in round of 32
2011 Texas A&M 49 326 327 7 Lost to 10 seed in round of 64
2010 Pitt 25 322 229 3 Lost to 6 seed in round of 32
2010 Marquette 30 304 308 6 Lost to 11 seed in round of 64
2010 Notre Dame 31 312 244 6 Lost to 11 seed in round of 64


So, in the last four tournaments, there were 14 teams seeded 7 or better with tempo in the lowest 50 and NCSOS in the bottom 150. Of those 14 teams, twelve of them lost to teams seeded 3 or more spots below them, including 9 first round upsets and a #8 beating a #1. Ironically, the only team in the group to even get to the Sweet 16 was Wisconsin, which did it twice.

Teams that didn't quite make the cut:

2012 Georgetown (300 tempo; 152 NCSOS), 3 seed lost to 11 seed in round of 32
2012 St. Mary's (261 tempo; 233 NCSOS), 7 seed lost to 10 seed in round of 64
2011 Notre Dame (273 tempo; 263 NCSOS), 2 seed lost to 10 seed in round of 32
2010 Wisconsin (340 tempo; 131 NCSOS), 4 seed lost to 12 seed in round of 32

I used the new Pomeroy formula and end-of-tournament rankings, which might skew things a bit, but probably not too much since teams don't change tempo or non-conference schedule all that much in the tournament. Also, full disclosure, if you go back the 7 seasons from 2003 to 2009, there are only 7 such teams and only 1 of them lost to team seeded 3 spots or more below them (and none of the 7 teams went past the Sweet 16, but 5 of the 7 reached the Sweet 16). Still, Wisconsin doesn't appear to have been a problem before 2010, and there were so few teams that met the criteria in those years (1.0 per year as opposed to 3.5 per year in the later sample), so I'm running with it.

Now just because they all lost in the NCAAT doesn't mean they're overrated in Pomeroy, but it might. And either way it's pretty amazing, right? I don't know about you, but when I'm filling out my bracket this year I'll be looking for teams with slow tempos and bad non-conference schedules as potential upset victims. For what it's worth, there are three teams currently in danger of meeting my criteria in the upcoming NCAA tournament: Syracuse (345 tempo; 218 NCSOS), Cincinnati (318 tempo; 315 NCSOS), and Pittsburgh (290 tempo; 285 NCSOS). You heard it here first.

Great work Kedsy. That took a lot of time to conduct that analysis. Not that you probably care, but I would've sporked you if I could've.

[Kedsy]

Great work Kedsy. That took a lot of time to conduct that analysis. Not that you probably care, but I would've sporked you if I could've.

Thanks, Tom.

[MChambers]

Now just because they all lost in the NCAAT doesn't mean they're overrated in Pomeroy, but it might. And either way it's pretty amazing, right? I don't know about you, but when I'm filling out my bracket this year I'll be looking for teams with slow tempos and bad non-conference schedules as potential upset victims. For what it's worth, there are three teams currently in danger of meeting my criteria in the upcoming NCAA tournament: Syracuse (345 tempo; 218 NCSOS), Cincinnati (318 tempo; 315 NCSOS), and Pittsburgh (290 tempo; 285 NCSOS). You heard it here first.
This is great. I've been wondering about this very thing (slow tempo plus weak NCSOS), but was too busy or lazy to take a look to see whether my suspicion was correct. Ohio State might belong in this group, too.

But I can't really explain why these characteristics might correlate with a lack of NCAA tournament success. Could it be a signal that their defenses can't cope with quick, talented teams?

[vick]

This is great. I've been wondering about this very thing (slow tempo plus weak NCSOS), but was too busy or lazy to take a look to see whether my suspicion was correct. Ohio State might belong in this group, too.

But I can't really explain why these characteristics might correlate with a lack of NCAA tournament success. Could it be a signal that their defenses can't cope with quick, talented teams?

To the degree that the correlation with pace is meaningful, I would guess it's just the fact that with fewer possessions, upsets are more likely, for the same reason an upset is less likely with best-of-7 than single-elimination.

[MChambers]

To the degree that the correlation with pace is meaningful, I would guess it's just the fact that with fewer possessions, upsets are more likely, for the same reason an upset is less likely with best-of-7 than single-elimination.
But why is it also correlated with a weak NCSOS? Is it that the weak NCSOS means that the team's rating is overstated and that combined with fewer possessions leads to upsets?

[Kedsy]

But why is it also correlated with a weak NCSOS? Is it that the weak NCSOS means that the team's rating is overstated and that combined with fewer possessions leads to upsets?

My theory is that teams that play an extremely slow pace need to control the tempo to succeed, and against poor competition it's easy to do just that. In conference play, you're used to your opposition so you can potentially hide your flaws because you're so familiar with whatever's coming, so it doesn't really help in this regard. When these teams get to the Big Dance, they haven't had enough experience controlling the tempo against a quality opponent. If they fail to do so, or if the other team is comfortable playing at their pace, it leads to a close game, and in the NCAAT, if you're in a close game upsets get pretty likely, especially if your rating is overstated in the first place.

It's a theory, anyway.

[ice-9]

Duke now #2 on KenPom on the strength of its out-worldly 128.5 adjusted offense. Wow! The differential between AdjO and AdjD is 28.4, compared to Arizona's at 26.9. How good would we be if we can only improve that defense.

[pfrduke]

My theory is that teams that play an extremely slow pace need to control the tempo to succeed, and against poor competition it's easy to do just that. In conference play, you're used to your opposition so you can potentially hide your flaws because you're so familiar with whatever's coming, so it doesn't really help in this regard. When these teams get to the Big Dance, they haven't had enough experience controlling the tempo against a quality opponent. If they fail to do so, or if the other team is comfortable playing at their pace, it leads to a close game, and in the NCAAT, if you're in a close game upsets get pretty likely, especially if your rating is overstated in the first place.

It's a theory, anyway.

Do the tournament losses come in games where the losing team isn't able to play at the pace it likes, though? I don't recall, for example, Cornell running Wisconsin out of the building.

[bob blue devil]

Since Pomeroy altered his ...

Wonderful work. Thanks for sharing.

I don't know that pomeroy ever adequately addressed the why for the Wisconsin effect. Let me toss out an idea. His system is based on per possession efficiency. Lets assume that teams let off the gas when they are up 'big' and that all teams view 'big' in absolute (not tempo adjusted) terms - so duke and Wisconsin let off the gas a bit when they are up 20. If you have 2 equally efficient teams playing overmatched opponents, the faster tempo team will hit a 'big' lead earlier and take its foot off the gas earlier. Thus it will play a greater number of possessions at a reduced efficiency, which pomeroy's system doesn't handicap. Hence advantage to slower tempo teams playing weak opposition. What do you think - is this reasonable?

[ChillinDuke]

Do the tournament losses come in games where the losing team isn't able to play at the pace it likes, though? I don't recall, for example, Cornell running Wisconsin out of the building.

Well, depending on your definition of run, it looks like they did...

http://scores.espn.go.com/ncb/recap?gameId=300800275

- Chillin

[pfrduke]

Well, depending on your definition of run, it looks like they did...

http://scores.espn.go.com/ncb/recap?gameId=300800275

- Chillin

I guess I meant in terms of pace. That was a 61 possession game in a season where Wisconsin played 60 possessions/game. Wisconsin got to play at its pace, Cornell was just ruthlessly efficient on offense.

[Kedsy]

Wonderful work. Thanks for sharing.

I don't know that pomeroy ever adequately addressed the why for the Wisconsin effect. Let me toss out an idea. His system is based on per possession efficiency. Lets assume that teams let off the gas when they are up 'big' and that all teams view 'big' in absolute (not tempo adjusted) terms - so duke and Wisconsin let off the gas a bit when they are up 20. If you have 2 equally efficient teams playing overmatched opponents, the faster tempo team will hit a 'big' lead earlier and take its foot off the gas earlier. Thus it will play a greater number of possessions at a reduced efficiency, which pomeroy's system doesn't handicap. Hence advantage to slower tempo teams playing weak opposition. What do you think - is this reasonable?

That does make sense. Still left to be explained is why such teams who also play lousy schedules tend to fare poorly in the NCAA tournament.

[Kedsy]

I guess I meant in terms of pace. That was a 61 possession game in a season where Wisconsin played 60 possessions/game. Wisconsin got to play at its pace, Cornell was just ruthlessly efficient on offense.

Yeah, Wisconsin got out-Wisconsined by a Cornell team that Pomeroy ranked as the 4th best adjusted offense in the country and was used to playing at a slow pace (#245).

Wisconsin's NCSOS wasn't terrible that year (#131) but if they were a bit overrated (at .9029) and Cornell a bit underrated (at .8322) because they had to play in the Ivy League (and thus their overall schedule strength was poor even though their NCSOS was good (#70)). That game may simply have been a game between two fairly evenly matched teams, both playing within their comfort zone pace-wise, with one of them playing at the top of its game and the other caught a little flatfooted.

[Duvall]

That does make sense. Still left to be explained is why such teams who also play lousy schedules tend to fare poorly in the NCAA tournament.

Isn't the simple (and probably wrong) explanation that Pomeroy hasn't adjusted enough for strength of schedule, so teams are overly rewarded for beating up on weak competition?

[Kedsy]

Isn't the simple (and probably wrong) explanation that Pomeroy hasn't adjusted enough for strength of schedule, so teams are overly rewarded for beating up on weak competition?

If the ratings were perfect (obviously impossible in practice, but possible in theory), then he wouldn't have to adjust at all. If Duke played Elon and were 30 points better and won by 38, then Duke performed better than expected and its rating should adjust accordingly. The problem with that, of course, is a 30 point win and a 38 point win are pretty much the same. Once you get over, say, 25 points teams either "let off the gas," as Bob Blue Devil suggested, or the coach starts taking out his regulars, and so on. After some indeterminate point, how well you perform depends on how "ruthless" your players are or how much better your scrubs are than their scrubs. Adjusting for strength of schedule only indirectly addresses this problem -- you can win 82-50 even against supposedly strong competition. So you could argue Pomeroy shouldn't have adjusted for strength of schedule at all (and I'm not sure if that's what he's done, anyway), that a better route would be discounting points over a certain amount. Although that would open a big can of worms concerning exactly where to draw the line and how much to discount, etc., and would begin to make the system less objective. Especially since ruthlessness and a more skilled bench are qualities that help define how good a team is. It's a difficult conundrum.

Also, none of this would t explain why slower paced teams tend to benefit more from beating up weak competition than faster paced teams. I think Bob Blue Devil's thought may make the most sense on that front.

[bob blue devil]

That does make sense. Still left to be explained is why such teams who also play lousy schedules tend to fare poorly in the NCAA tournament.

right. ideas i don't find terribly compelling, but could contribute:
- these teams are over-ranked/seeded as a result of their inflated win-loss and pomeroy rankings
- they don't improve as much during the season because of the weaker competition doesn't challenge them (i do think this is a big problem for mid majors, but not majors)
- they tend to be more dependent on long-range shooting, which takes a hit in the big arenas/domes
- tournament play is more half court-oriented and other teams develop more comfort around playing at that tempo via their conference tournaments

really i'm just throwing darts, if you couldn't tell.

[TexHawk]

Although that would open a big can of worms concerning exactly where to draw the line and how much to discount, etc., and would begin to make the system less objective. Especially since ruthlessness and a more skilled bench are qualities that help define how good a team is. It's a difficult conundrum.


But even that has its limits. Every school puts their walk-ons in under 2 minutes left in blowouts, and when you are playing walk-ons v. walk-ons, that doesn't tell you anything meaningful. (Last year Kansas had two walk-ons with famous dads, Evan Manning and Tyler Self, who were/are woefully bad. They would consistently get outscored 8-0, 12-0, 14-0, which would throw off somebody's metrics somewhere.)

I imagine you could mitigate that somewhat by de-valuing offense/defense when the majority of players are on the court average less than 2-3 mpg. That's a lot of box scores to crunch, across 300+ teams and all of their games, but that's what computers are for. :) But then what do we do with Deano and Roy, who throw their walk-ons in to teach the starters a lesson, or act as stand-ins/bait/fodder when the opposing fans rush the court?

[Kedsy]

But even that has its limits. Every school puts their walk-ons in under 2 minutes left in blowouts, and when you are playing walk-ons v. walk-ons, that doesn't tell you anything meaningful. (Last year Kansas had two walk-ons with famous dads, Evan Manning and Tyler Self, who were/are woefully bad. They would consistently get outscored 8-0, 12-0, 14-0, which would throw off somebody's metrics somewhere.)

I imagine you could mitigate that somewhat by de-valuing offense/defense when the majority of players are on the court average less than 2-3 mpg. That's a lot of box scores to crunch, across 300+ teams and all of their games, but that's what computers are for. :) But then what do we do with Deano and Roy, who throw their walk-ons in to teach the starters a lesson, or act as stand-ins/bait/fodder when the opposing fans rush the court?

I totally agree, and the problem isn't limited to walk-ons. If a team generally plays a 7-man rotation, and in the last 10 minutes of blowouts they generally play the 8 through 12 guys, how do those 10 minutes say anything valuable about the team's rating? Maybe it tells you how much depth they have in case of an injury, but that's a very minor component of how good a team is. Also, different coaches empty their benches at different times. If one team brings in the scrubs with 10 minutes to play and the other coach waits until 2 minutes to play, the 8 minutes in between are a huge problem for any rating system.

On the other hand, there's no practical way for Pomeroy to measure any of that.

Finally, props for the Roy dig. Nice.

[Wander]

And even without actually substituting guys, you have the stallball problem. If we're up by a medium-sized amount and go into stallball mode, it probably hurts our offensive efficiency, even when it's the best strategy for winning the game and even though we're still playing our best players.

[Kedsy]

And even without actually substituting guys, you have the stallball problem. If we're up by a medium-sized amount and go into stallball mode, it probably hurts our offensive efficiency, even when it's the best strategy for winning the game and even though we're still playing our best players.

That's a good point. Coaching strategies can affect things a lot, on both sides of the ball. For instance, if a team's behind, fouling every possession is a viable strategy, but sometimes it leads to a somewhat consistent 2 points per possession, which could uncharacteristically hurt your defensive efficiency.

[pfrduke]

And even without actually substituting guys, you have the stallball problem. If we're up by a medium-sized amount and go into stallball mode, it probably hurts our offensive efficiency, even when it's the best strategy for winning the game and even though we're still playing our best players.

Maybe - we've been pretty efficient in stallball. It hurts our overall points total, maybe, but not necessarily the efficiency.

[vick]

Maybe - we've been pretty efficient in stallball. It hurts our overall points total, maybe, but not necessarily the efficiency.

Maybe this year, but I think Wander's general point about when a team enters stalling, or fouling, has to be right, otherwise we'd be fools not to play the 'stallball' offense from the opening tip (I realize this argument has the feel of an extreme 'efficient markets' viewpoint but I think the logic is right).

[Wander]

Maybe this year, but I think Wander's general point about when a team enters stalling, or fouling, has to be right, otherwise we'd be fools not to play the 'stallball' offense from the opening tip (I realize this argument has the feel of an extreme 'efficient markets' viewpoint but I think the logic is right).

Yeah, that's what I was thinking. You're basically giving yourself a 10 or 15 second shot block instead of a 35 second shot clock. Given a large enough sample size, I think that has to hurt your offensive efficiency on average. Teams like Duke that are good at it minimize that effect, but I don't see how it would be eliminated entirely.

[davekay1971]

Yeah, that's what I was thinking. You're basically giving yourself a 10 or 15 second shot block instead of a 35 second shot clock. Given a large enough sample size, I think that has to hurt your offensive efficiency on average. Teams like Duke that are good at it minimize that effect, but I don't see how it would be eliminated entirely.

I'm sure it isn't. Of course, the point is whether the sacrifice in points per possession is enough to outweigh the benefit of limiting the other team's possessions when you have a lead. In the pre-shot clock era, a team could theoretically have an offensive efficiency of 0 in stall ball, go into stall ball with minutes left on the clock, and still guarantee a win. With a shorter shot clock, you are going to have to score some in stall ball, but should still win if you're just efficient enough.

[Duvall]

I'm sure it isn't. Of course, the point is whether the sacrifice in points per possession is enough to outweigh the benefit of limiting the other team's possessions when you have a lead. In the pre-shot clock era, a team could theoretically have an offensive efficiency of 0 in stall ball, go into stall ball with minutes left on the clock, and still guarantee a win.

Offensive efficiency of zero?


http://www.youtube.com/watch?v=YRVmc6dO0g8

[tommy]

Kedsy's work here is very interesting. Piggybacking on his analysis, I thought I'd take a look at those 14 teams in the last 4 tournaments that met his criteria -- playing really slow and having played weak non conference schedules -- 12 of which were upset in the tournament, and seeing who they lost to, how fast or slow those opponents were used to playing, and how fast or slow the tournament game actually was played, and to whose advantage. Here are the results:




Team
Seed
Avg. Poss/Game
Opponent
Seed
Avg. Poss/Game
Poss this game
Advantage
Result













2013 Gtown
2
63.4
FGCU
15
70.2
76
FGCU
FGCU by 10


2013 Wisc
5
62.3
Ole Miss
12
71.3
62
Wisc
Miss by 11


2013 K St.
4
63.6
LaSalle
13
66.9
57
K State
LaSalle by 2


2013 N Dame
7
63.8
Iowa St
10
71
65
N Dame
ISU by 18


2012 Wisc
4
59.1
Syracuse
1
65.4
52
Wisc
Syr by 1


2012 Mich
4
61.9
Ohio U.
13
67.2
57
Michigan
Ohio by 5


2012 N Dame
7
62.4
Xavier
10
68.1
57
N Dame
Xavier by 4


2011 Pitt
1
63.6
Butler
8
65.3
59
Pitt
Butler by 1


2011 Wisc
4
57
Butler
8
65.3
62
None
Butler by 7


2011 Cincy
6
63.9
Uconn
3
66.2
59
Cincy
Uconn by 11


2011 Tex AM
7
64.1
Fla State
10
69.5
59
Tex A&M
FSU by 7


2010 Pitt
3
63.9
Xavier
6
69.9
64
Pitt
Xavier by 3


2010 Marq
6
65.3
Washington
11
73
66
Marquette
UW by 2


2010 N Dame
6
64.7
O Dominion
11
63.1
53
None
ODU by 1





Reading the columns left to right, this shows the upset victim (actually, two weren't upsets -- 2012 Wisconsin and 2011 Cincinnati -- as they were the lower seeded teams), then their seed, their season average possessions per game, the opponent, their seed, their average possessions per game, then importantly the number of possession in the tournament game between the two teams, then a column where I assess to whose advantage the actual tempo was, meaning to whose preferred tempo was the game played. If there was no real discernible advantage, like if it was close to right between the two teams' averages, or the two teams' averages themselves were very close to each other, than I said "none." And then the last column is the result of the game.

What this looks like to me is that these slow paced teams for the most part were able to keep these NCAA Tournament games at the pace they liked, or close to it, but they lost anyway. With the exception of Georgetown last year against Florida Gulf Coast, these teams didn't get sped up and get forced to play an uncomfortable game. Rather, the other, lower seeded team was usually able to play closer to the higher seeded team's pace, and beat them anyway.

Folks smarter than I am may differ, but I'm a little surprised at this data. I thought that if these slow-paced higher-seeded (for the most part) teams could keep the tournament games at their pace, they'd be successful in doing that. Heck, they were successful with the strategy all year. That's how they got the high seeds to begin with. But they weren't able to carry it over and take advantage of the strategy in the tournament. I'm also aware that tournament games in general slow down, so the pace numbers may naturally tend to be closer to these already-slow paced favorites' preferred tempo, but that only should've made the games more to their liking. Which makes it even more surprising that with the games largely played at their pace, they lost almost every single one.

[ice-9]

Folks smarter than I am may differ, but I'm a little surprised at this data. I thought that if these slow-paced higher-seeded (for the most part) teams could keep the tournament games at their pace, they'd be successful in doing that. Heck, they were successful with the strategy all year. That's how they got the high seeds to begin with. But they weren't able to carry it over and take advantage of the strategy in the tournament. I'm also aware that tournament games in general slow down, so the pace numbers may naturally tend to be closer to these already-slow paced favorites' preferred tempo, but that only should've made the games more to their liking. Which makes it even more surprising that with the games largely played at their pace, they lost almost every single one.

Isn't the key variable that those 14 teams played a weak schedule? So they were able to beat up bad opponents with their slow pace. But, pretty much every team in the NCAA tournament -- even the 16 seeds in this age of parity -- are quite good, and that may not be something the 14 teams are used to. So when the game gets close, 12 of those teams play a little tighter and then lose?

Interesting stuff! Great work Kedsy and tommy, would spork you guys if I could.

Now I know who to mark down for upsets in the NCAA tournament. ;)

[Kedsy]

I'm a little surprised at this data. I thought that if these slow-paced higher-seeded (for the most part) teams could keep the tournament games at their pace, they'd be successful in doing that. Heck, they were successful with the strategy all year. That's how they got the high seeds to begin with. But they weren't able to carry it over and take advantage of the strategy in the tournament. I'm also aware that tournament games in general slow down, so the pace numbers may naturally tend to be closer to these already-slow paced favorites' preferred tempo, but that only should've made the games more to their liking. Which makes it even more surprising that with the games largely played at their pace, they lost almost every single one.

Thanks, Tom, for following up. It does seem a bit odd that these teams lost despite the games being played at their tempo.

Ice-9 may be on the right track. Consider the combination of: (a) these teams might be somewhat overrated because they performed really well against a poor schedule; (b) these teams are not used to playing good teams with which they are unfamiliar (assuming they are very familiar with conference foes); (c) a game with fewer possessions, if played between two relatively equal teams, has more of a chance to be close; and (d) because of the high pressure in the tournament, close games seem to go either way and don't necessarily favor the better team.

Based on the above, I still wouldn't predict 12 upsets in 14 games, but I wouldn't be surprised at a lot of upsets, either.

Finally, I have no idea which if any of the factors I list above are relevant, but if (b) above matters at all, it might affect Syracuse and Pitt, who both play at slow paces and have poor non-conference schedules, because in their first year in the ACC they are not particularly familiar with their conference foes, and thus it's almost like the whole season is non-conference for them (and including their ACC games, their overall schedules are rated pretty high). So, in an odd way, the fact that they're new to the league might save them from the hex. No such luck for Cincinnati (#316 in tempo; #319 in NCSOS), so their tournament game(s) might be an interesting test.

[Olympic Fan]

I was listening to Bilas and company repeat again that their assertion that the two best conferences in America are the Big Ten and the Big 12.

While I think that's true if you are talking about the depth of talent and the strength of the bottom teams, it's not true if you're looking at the strengh at the top. As far as dork polls go, check out Pomeroy.

As of this morning, the top three teams in the ACC were No. 2 (Duke), No. 4 (Syracuse) and No. 6 (Virginia). That's three teams better than anybody in the Big Ten: No. 7 (Iowa), No. 12 (Michigan State), No. 14 (Ohio State) or Big 12: No. 9 (Kansas), No. 20 (Iowa State), No. 25 (Oklahoma State). The SEC doesn't have anybody in the top 10, while the Pac Ten drops from No. 1 Arizona to No. 19 (UCLA) to No. 32 (Oregon). The second-best conference -- at the top -- is the new Big East with No. 3 (Creighton), No. 5 (Villanova) ... but then a big drop to No. 41 (Xavier).

The ACC is soft in the middle and weak at the bottom ... but it's the toughest in the country at the top.

[Skitzle]

How does KenPoms AdjD get calculated? Is there anyway for someone to calculate it for all games after the ND loss? I'm curious to see where Duke is since the beginning of January...

Is anyone else curious?

The highest points allowed defensive efficiency rating by a national champion since 2002 was 92.9 by UNC in 2009. We're currently at 100.4.

Something tells me that if we were to look only since the ND game our number would be a lot better..

[Wander]

I was listening to Bilas and company repeat again that their assertion that the two best conferences in America are the Big Ten and the Big 12.

While I think that's true if you are talking about the depth of talent and the strength of the bottom teams, it's not true if you're looking at the strengh at the top. As far as dork polls go, check out Pomeroy.

As of this morning, the top three teams in the ACC were No. 2 (Duke), No. 4 (Syracuse) and No. 6 (Virginia). That's three teams better than anybody in the Big Ten: No. 7 (Iowa), No. 12 (Michigan State), No. 14 (Ohio State) or Big 12: No. 9 (Kansas), No. 20 (Iowa State), No. 25 (Oklahoma State). The SEC doesn't have anybody in the top 10, while the Pac Ten drops from No. 1 Arizona to No. 19 (UCLA) to No. 32 (Oregon). The second-best conference -- at the top -- is the new Big East with No. 3 (Creighton), No. 5 (Villanova) ... but then a big drop to No. 41 (Xavier).

The ACC is soft in the middle and weak at the bottom ... but it's the toughest in the country at the top.

Why would you judge a conference by just its top few teams? If you're going to do that, you might as well, you know, just talk about individual teams. Duke doesn't have to play Virginia and Syracuse 18 times.

Bilas' assertion is well-justified. I mean, the Big 12 has 10 teams (insert joke) and its 9th place team has a better non-conference resume than, what, 12 of 15 ACC teams?

[sporthenry]

How does KenPoms AdjD get calculated? Is there anyway for someone to calculate it for all games after the ND loss? I'm curious to see where Duke is since the beginning of January...

Is anyone else curious?

The highest points allowed defensive efficiency rating by a national champion since 2002 was 92.9 by UNC in 2009. We're currently at 100.4.

Something tells me that if we were to look only since the ND game our number would be a lot better..

He does conference only. Duke is at 103.7 for all conference games. Sure, that includes the ND game but kicking that out wouldn't make it better than our current 100.4.

[pfrduke]

He does conference only. Duke is at 103.7 for all conference games. Sure, that includes the ND game but kicking that out wouldn't make it better than our current 100.4.

I believe the conference only is raw, not adjusted, so it's not quite equivalent to compare between the two.

[Olympic Fan]

Why would you judge a conference by just its top few teams? If you're going to do that, you might as well, you know, just talk about individual teams. Duke doesn't have to play Virginia and Syracuse 18 times.

Bilas' assertion is well-justified. I mean, the Big 12 has 10 teams (insert joke) and its 9th place team has a better non-conference resume than, what, 12 of 15 ACC teams?

I never asserted that the ACC was the best conference or that the Big 12/Ten were not. I specifically mentioned that the middle of the ACC is soft and the bottom is weak.

My point was that the ACC is the toughest conference at the top ... and Pomeroy's data supports that allegation.

[brevity]

The ACC is soft in the middle and weak at the bottom ... but it's the toughest in the country at the top.


Why would you judge a conference by just its top few teams?

In February, I'm not sure why you would. March is another story.

You can gauge a conference's overall (top-to-bottom) strength when you build a bracket. Amateur bracketologists -- a redundant term, if you think about it -- are doing this right now. Maybe their results have more Big 12 teams than any other, but none are higher than a 3 seed.

Come March, once the bracket is in place, we turn to conferences' top-heavy strength when we project results, and again when we see the actual results. We notice, hey, only 5 ACC teams made the tournament, but 4 of them made the Sweet 16.

A great illustration of this is 2011, when the Big East placed 11 teams into the tournament. Were they the best conference? I'm pretty sure DBR debated it at length before and after Selection Sunday. As it turned out, Connecticut won the title, but Marquette was the only other Big East team to make the Sweet 16, as many of their conference brethren were upset.

We have this debate over conference strength every year because there is no satisfying answer. On Selection Sunday, you can argue that the best conference is already decided: the conference that places the most teams in the tournament shows the most overall strength. The tournament itself is random; it comes down to matchups. Teams from the same conference are not interchangeable; if 1-seed Syracuse loses to 4-seed Iowa, that doesn't mean that Duke would do the same. In the alternative, you can argue that the tournament is all that matters, and that the conference with the most survivors after each weekend is the best for that whole season.

[Kedsy]

The highest points allowed defensive efficiency rating by a national champion since 2002 was 92.9 by UNC in 2009. We're currently at 100.4.

Offense is WAY up this year. I'm not sure looking at the efficiency numbers from previous years tells you very much. For example, this year 94.4 ranks 20th in defense. Last year, the same number ranked 49th.

[El_Diablo]

Offense is WAY up this year. I'm not sure looking at the efficiency numbers from previous years tells you very much. For example, this year 94.4 ranks 20th in defense. Last year, the same number ranked 49th.

Yeah, the defensive efficiency numbers are up, but only by a point or two. And it's not like adjusting our efficiency down a point or two would get us anywhere near any past champions. Our defense is 93rd overall right now (and if adjusted down a couple points would have been in the 90s-100s last year).

The worst defensive team to win a championship since Pomeroy started his rankings was ranked #21 for that season.* That's Syracuse/Villanova/Kansas territory right now.

*Along with the #1 offense.

[COYS]

Offense is WAY up this year. I'm not sure looking at the efficiency numbers from previous years tells you very much. For example, this year 94.4 ranks 20th in defense. Last year, the same number ranked 49th.

This is a good point. I think comparing the adjusted O and D efficiency stats across years can be misleading. The emphasis on hand checking this year seems like a reasonable reason offensive efficiency has gone up (more fouls = more free throws = more points . . .or at least easier drives to the hoop as defenses adjust). I think efficiency margin is perhaps a better way to compare different years.

Duke's current adjusted efficiency margin is 128.8Oeff - 100.4Deff = +28.4. According to KenPom, Duke is playing .284 points per possession better than the average team this year. At the end of the tournament last season, Louisville was +31. UK was +31.4 in 2012. UCONN was only +22.5 by the end of the tourney. Duke was +30.5 in 2010 and UNC was +29.5 in 09.

Considering that the numbers for these past champions are post tournament and so include their run to the title, I would imagine Duke is in a pretty good place, historically, in terms of how much better per possession the team has played. This is not to say I don't hope this team manages to improve on defense a little but more (man would we be dominant if we could improve defense even two more points per 100 possessions). However, based on how Duke has performed relative to their opponents on a possession by possession basis, Duke does not seem to be an outlier in terms of its ability to win a title.

As a side note, I know it's a small sample size and I only calculated the champions since 2009, but that UCONN team really looks every bit the surprise we all thought they were at the time. If I have more time at a computer (and not on my phone which is my usual posting device) I'll run the differential for all teams that have made it to the final four in the KenPom era. I wonder if UCONN would still stand out as much.

[Kedsy]

The worst defensive team to win a championship since Pomeroy started his rankings was ranked #21 for that season.* That's Syracuse/Villanova/Kansas territory right now.

*Along with the #1 offense.

This is not quite true. Going into the tournament, UNC's defense in 2009 was ranked #35. UConn's defense in 2011 was ranked #31.

Now, I realize #93 is not even close to #35 but:

(a) It's a long way from Selection Sunday;

(b) We look at these numbers like they're gospel, but even at this point in the season, one bad game has a huge influence on the numbers. For example, our unadjusted defensive efficiency for the season is 100.68. For all games other than the Vermont game, our unadjusted defensive efficiency is 99.03. Since Vermont is not a particularly good offensive team (Pomeroy offense rank #110), I would assume the adjusted numbers would be even further apart than the unadjusted numbers, but even assuming the difference is the same, that would drop our defensive efficiency to 98.7 or 98.8, which would drop our rank from #93 to #71. Now, I'm not saying we should completely ignore the Vermont game, but it also shouldn't completely define how we view the team now;

(c) Defense is only half the formula. I haven't run the numbers, but it doesn't appear that having a good defense is correlated to winning in the NCAAT any more than having a good offense;

(d) As illustrated by UConn moving from #31 to #14 in the six games of the 2011 tournament, all that really matters is how well you play D in the tourney;

Anyway, we have between 8 and 10 games before the NCAAT. If UConn could drop its defensive efficiency rating 17 spots in 6 games, it's possible for us to drop 25 or so in our remaining games. Then factor out Vermont, and we drop another 20+. That puts us in the 40s, which actually isn't too far from the 35 and 31 that UNC and UConn had going into the 2009 and 2011 tournaments.

Which of course is just semantics. My point is past numbers don't exclude future success. We have a really good team with defensive and concentration issues. If we overcome those issue during the tournament and we're the beneficiaries of a little luck, we certainly have a chance to win it.

[freshmanjs]

This is not quite true. Going into the tournament, UNC's defense in 2009 was ranked #35. UConn's defense in 2011 was ranked #31.

Now, I realize #93 is not even close to #35 but:

(a) It's a long way from Selection Sunday;

(b) We look at these numbers like they're gospel, but even at this point in the season, one bad game has a huge influence on the numbers. For example, our unadjusted defensive efficiency for the season is 100.68. For all games other than the Vermont game, our unadjusted defensive efficiency is 99.03. Since Vermont is not a particularly good offensive team (Pomeroy offense rank #110), I would assume the adjusted numbers would be even further apart than the unadjusted numbers, but even assuming the difference is the same, that would drop our defensive efficiency to 98.7 or 98.8, which would drop our rank from #93 to #71. Now, I'm not saying we should completely ignore the Vermont game, but it also shouldn't completely define how we view the team now;

(c) Defense is only half the formula. I haven't run the numbers, but it doesn't appear that having a good defense is correlated to winning in the NCAAT any more than having a good offense;

(d) As illustrated by UConn moving from #31 to #14 in the six games of the 2011 tournament, all that really matters is how well you play D in the tourney;

Anyway, we have between 8 and 10 games before the NCAAT. If UConn could drop its defensive efficiency rating 17 spots in 6 games, it's possible for us to drop 25 or so in our remaining games. Then factor out Vermont, and we drop another 20+. That puts us in the 40s, which actually isn't too far from the 35 and 31 that UNC and UConn had going into the 2009 and 2011 tournaments.

Which of course is just semantics. My point is past numbers don't exclude future success. We have a really good team with defensive and concentration issues. If we overcome those issue during the tournament and we're the beneficiaries of a little luck, we certainly have a chance to win it.

of course, it's not really fair to drop Vermont and look at rankings unless you drop every team's worst defensive game.

[Kedsy]

Duke's current adjusted efficiency margin is 128.8Oeff - 100.4Deff = +28.4. According to KenPom, Duke is playing .284 points per possession better than the average team this year. At the end of the tournament last season, Louisville was +31. UK was +31.4 in 2012. UCONN was only +22.5 by the end of the tourney. Duke was +30.5 in 2010 and UNC was +29.5 in 09.

Here are pre-tournament spreads for the past 5 champions:

2013 Louisville: 33.6
2012 Kentucky: 34.1
2011 UConn: 22.9
2010 Duke: 35.6
2009 UNC: 31.6

For what it's worth, VCU made the Final Four in 2011 after having pre-tournament numbers of 109.6 offense and 100.8 defense, for a spread of only +8.8.

If you decide to run the numbers and want pre-tournament data, I can send you text files of Selection Sunday Pomeroy for the past five seasons. Just PM me if you want the data.

[Kedsy]

of course, it's not really fair to drop Vermont and look at rankings unless you drop every team's worst defensive game.

Of course. The point is we're nowhere close to the same team we were against Vermont, but that game alone means a 20+ difference in our defensive rank. The full season numbers don't entirely define who can do well in the tournament and who can't.

[Skitzle]

Of course. The point is we're nowhere close to the same team we were against Vermont, but that game alone means a 20+ difference in our defensive rank. The full season numbers don't entirely define who can do well in the tournament and who can't.

Exactly,

That's why I've been curious about our Defense Specifically since the Notre Dame & Clemson Games. Those two game we lost because of bad defense.

The Syracuse game, the BC game, and the Virginia game had stretches of bad D (Though I think the Syracuse D wasn't terrible. CJ Fair couldn't miss in the second half. He made a TON of tough shots.)

Guess its not possible to get to these numbers.

That said if Duke were to win the NCAA Championship, they would be doing it with an historically bad Defense (as compared to other champions :D)

[Kedsy]

That said if Duke were to win the NCAA Championship, they would be doing it with an historically bad Defense (as compared to other champions :D)

Slight quibble -- if we win the championship, our defense will probably be really good. Our season-long defensive numbers will be historically bad.

[Wander]

Of course. The point is we're nowhere close to the same team we were against Vermont, but that game alone means a 20+ difference in our defensive rank. The full season numbers don't entirely define who can do well in the tournament and who can't.

Sure. I used to look at all these trends and think that a team couldn't win the national championship with an over-reliance on full court offense, or that a team couldn't win the national championship with only 3 scorers, but along came 2009 and 2010. I've learned that there's no magic formula and plenty of ways to win. That said, we can talk all we want about CJ Fair hitting tough shots or whatever, but we've played 18 full games since Vermont, and our defense is still ranked in the 90's, so I'm not sure "we're nowhere close to the same team we were against Vermont," at least on defense.

[Des Esseintes]

Sure. I used to look at all these trends and think that a team couldn't win the national championship with an over-reliance on full court offense, or that a team couldn't win the national championship with only 3 scorers, but along came 2009 and 2010. I've learned that there's no magic formula and plenty of ways to win. That said, we can talk all we want about CJ Fair hitting tough shots or whatever, but we've played 18 full games since Vermont, and our defense is still ranked in the 90's, so I'm not sure "we're nowhere close to the same team we were against Vermont," at least on defense.

Agreed. We all want to believe there's a way to predict something as insanely random as the NCAA tournament, and most of the time that just leads to silly overfitting of past data to draw conclusions. I would love for Duke to be great at defense, but mostly I just want Duke to be great at basketball. The particular route it gets there is of far less importance. I'm willing to buy the research that says defensively elite teams have a slight advantage in later rounds compared to offensively elite teams, but regardless the margins of this advantage are awfully small. Being elite, full-stop, is the main thing.

[MChambers]

Agreed. We all want to believe there's a way to predict something as insanely random as the NCAA tournament, and most of the time that just leads to silly overfitting of past data to draw conclusions. I would love for Duke to be great at defense, but mostly I just want Duke to be great at basketball. The particular route it gets there is of far less importance. I'm willing to buy the research that says defensively elite teams have a slight advantage in later rounds compared to offensively elite teams, but regardless the margins of this advantage are awfully small. Being elite, full-stop, is the main thing.
I remember thinking that UNC wouldn't win the championship in 2009, because they weren't that good at defense, but as the season went on their defense improved and was affirmatively good by tournament time.

I'm hoping our defense continues to improve

[toooskies]

Slight quibble -- if we win the championship, our defense will probably be really good. Our season-long defensive numbers will be historically bad.

But, our season-long offensive numbers will be historically good to compensate! We're 5 points better per 100 on offense than the nearest team, which would be the greatest gap in the Kenpom era.

[crote]

In the blurb about Duke in the power rankings he he put out today (http://sportsillustrated.cnn.com/college-basketball/news/20140213/power-rankings-syracuse-arizona-wichita-state-florida-villanova/?eref=sihp), Luke Winn mentioned that Duke has a shot at having not only the most efficient offense this year, but the most efficient offense since Pomeroy began tracking efficiency stats in 2003. So I looked back in the archives, and Duke's current Offensive Rating of 128.5 is the best on record by a healthy margin. The next closest is Chris Paul's 04-05 Wake team, which finished the season at 124.0. The best Duke team on record was the 09-10, which ended the season with a 120.0 rating (tops in the nation that year). Who knows if this will hold out til the end of the tourney, but it does illustrate just how good this offense is.

And one slightly encouraging note from the Winn piece: he looked for teams from the past that are similar to Duke's efficiency profile this year. Two of the closer matches were the 02-03 teams from Marquette and Texas, both of which made the Final Four.

[Troublemaker]

In the blurb about Duke in the power rankings he he put out today (http://sportsillustrated.cnn.com/college-basketball/news/20140213/power-rankings-syracuse-arizona-wichita-state-florida-villanova/?eref=sihp), Luke Winn mentioned that Duke has a shot at having not only the most efficient offense this year, but the most efficient offense since Pomeroy began tracking efficiency stats in 2003. So I looked back in the archives, and Duke's current Offensive Rating of 128.5 is the best on record by a healthy margin. The next closest is Chris Paul's 04-05 Wake team, which finished the season at 124.0.

This is partly explained by the new rules emphases this season to help offenses.

From 2003 to 2013, the average D-1 Offensive Efficiency (OE) across all seasons was 101.1, and no individual season had an average OE higher than 101.9. This season, however, the D-1 average is 104.3, which is a significant change.

Still, it seems to me the difference between this season and previous seasons is: 104.3 - 101.1 = 3.2

Subtract 3.2 from Duke's 128.5 this season, and we'd still have the highest OE of the KenPom era at 125.3

(Maybe that method of translation would be considered crude / improper by statisticians, though. Not sure.)

[Des Esseintes]

After beating Tech, Duke's kenpom defensive efficiency now stands at 99.0, good for #68. Over the past couple of weeks, our number has improved something like 1.5-2.0 points and about 35 spots in the rankings. If we can effect similar improvement by the start of the tournament, I think we will be able to say our D has proven itself steady enough for a run at the title.

[Olympic Fan]

Interesting that Duke climbed back to No. 3 in the Ken Pom ratings. Duke and Creighton have been flipflopping for the last week or so. First Duke was third, then Creighton jumped after their lopsided win over Villanova ... now Duke is back in third.

Ken Pom is projecting Duke as a 79-78 winner at UNC. In fact, Duke is projected as the favorite in all its remaining games ... although the cumulative projection is for s 25-6 regular season finish. That would mean one more loss.

I think if it comes, it comes in the next two games -- after that, there's a very good chance that Duke is home free -- Va Tech at home, at Wake (after an eight-day layoff) and UNC at home.

I'd take 25-6 ... but I'd be thrilled at 26-5 (and 15-3 ACC)

[sporthenry]

Earlier in the year, Kenpom argued that the NCAA should do away with top 50/top 100 wins and go with quality wins. His thought process was that the #30 ranked team beating a top 50 opponent on a neutral floor is equivalent to beating a top 20 team on the road or a top 73 team away. An equivalent top 100 team moves it to 57 and 177 respectively.

Right now, that means beating Cincy at home is equivalent to beating MD on neutral court or Indiana away. Likewise, beating Vandy on neutral court is equivalent to beating Dayton at home or Western Kentucky away.

Before the game, Syracuse had a 96% chance of winning. That means losing to BC at home is the equivalent to losing to a team like Sacred Heart on the road. Now imagine if they had lost to Sacred Heart.

Now Kenpom also released an article saying that for some reason, when we think about home court and wins/losses, we seem to separate the two such that Wichita State's win at SLU is not seen as better than Cuse's win vs Duke.

I bring this up only because I hope KenPom goes Jay Bilas on this topic to change people's minds about what constitutes a good win/bad loss. It would be cool to see him develop some type of equivalency feature so that people can understand what that win/loss means in relation to home court.

On another note, I'd also like to see him argue about the unimportance of actually winning, meaning that 'Cuse's win over NC State, albeit a win, would actually hurt them. This topic would be quite a bit more controversial b/c our sports are predicated on winning but it always amazes me that Syracuse needs 2 buzzer beater wins to squeak by, yet at the end of the day, we ignore who they were against and just look at the W. I suppose this is what Kenpom's luck measures and I imagine most teams would have bad games (Duke/Vermont or Duke/Maryland come to mind) so it might even itself out. I just never understood rewarding or not punishing Syracuse for a game like NC State.

[JasonEvans]

On another note, I'd also like to see him argue about the unimportance of actually winning, meaning that 'Cuse's win over NC State, albeit a win, would actually hurt them. This topic would be quite a bit more controversial b/c our sports are predicated on winning but it always amazes me that Syracuse needs 2 buzzer beater wins to squeak by, yet at the end of the day, we ignore who they were against and just look at the W. I suppose this is what Kenpom's luck measures and I imagine most teams would have bad games (Duke/Vermont or Duke/Maryland come to mind) so it might even itself out. I just never understood rewarding or not punishing Syracuse for a game like NC State.

I agree with you to some extent, though it is worth noting that the ultimate prize (NCAA title) is predicated solely on winning. If you play great and dominate but things fall apart and you get really, really unlucky... your fate is exactly the same as the team that got their door blown off by 30 points.

-Jason "worth noting that Duke has had its share of luck this season-- Virginia and Maryland were both games we controlled, that we almost gave away late" Evans

[TexHawk]

-Jason "worth noting that Duke has had its share of luck this season-- Virginia and Maryland were both games we controlled, that we almost gave away late" Evans

According to Kenpom, Duke has been mostly unlucky this season. Ranking 262 out of 351 on his luck scale. Of the contenders... Duke, Louisville, UK, KU are all 250+. Zona is pretty lucky (#58). Nova is crazy lucky (#10).

Granted, that's probably the least proven of Kenpom's rankings formula, or at least it feels that way.

[Bluedog]

I just never understood rewarding or not punishing Syracuse for a game like NC State.

I agree, but by the same token, we definitely should have been punished for our slim victory over Maryland instead of moving up to #5 in the polls...(And dropped after the Vermont game, of course). In sports, a win by an inch often gets your the same amount of credit as a win by a mile whether right or wrong. That's how it's always been basically.

[vick]

I agree, but by the same token, we definitely should have been punished for our slim victory over Maryland instead of moving up to #5 in the polls...(And dropped after the Vermont game, of course). In sports, a win by an inch often gets your the same amount of credit as a win by a mile whether right or wrong. That's how it's always been basically.

As a bit of a side note, it's funny how large the Vermont game still looms, considering they are now 64th in Pomeroy, ahead of NC State, Notre Dame, Miami, Wake, Georgia Tech, BC, and Virginia Tech. Don't get me wrong, it was a bad defensive performance, but the evidence now suggests that Vermont wasn't nearly as terrible as it looked at the time.

ETA: Plugging into a spreadsheet, I believe Pomeroy would have Maryland favored over Vermont by 1 point on a neutral court, suggesting the Vermont and Maryland performances were basically equivalent.

[Bluedog]

As a bit of a side note, it's funny how large the Vermont game still looms, considering they are now 64th in Pomeroy, ahead of NC State, Notre Dame, Miami, Wake, Georgia Tech, BC, and Virginia Tech. Don't get me wrong, it was a bad defensive performance, but the evidence now suggests that Vermont wasn't nearly as terrible as it looked at the time.

ETA: Plugging into a spreadsheet, I believe Pomeroy would have Maryland favored over Vermont by 1 point on a neutral court, suggesting the Vermont and Maryland performances were basically equivalent.

Wow, I didn't realize that - thanks for pointing it out. They've won 14 out of their last 15 games - admittedly, their best win in that stretch is vs. 142 Yale and their loss is vs. 152 Stony Brook. Still, they've been destroying the 300+ teams, so that counts for something. Still surprised they shot up that much playing teams basically all outside the top 200 in the country. They're leading the America East as was predicted...

[tbyers11]

I agree with you to some extent, though it is worth noting that the ultimate prize (NCAA title) is predicated solely on winning. If you play great and dominate but things fall apart and you get really, really unlucky... your fate is exactly the same as the team that got their door blown off by 30 points.

-Jason "worth noting that Duke has had its share of luck this season-- Virginia and Maryland were both games we controlled, that we almost gave away late" Evans

Duke has had some luck in the games that you mention. If you try to quantify luck in really close games (I'll define this as a game that was a one possession difference (3 pts) in the final minute) where one strange play or ref whistle can make the outcome pretty much a coin flip. I have Syracuse at 5-1 (vs Miami, vs Pitt, @ Pitt, vs NCST, vs Duke, vs BC) and Duke at 3-2 (vs Vermont, @ND, vs UVa, @Syr, vs MD).

Good teams should win the majority of close games if they are going to be good teams but Syracuse had escaped more than its probabilistic share of close games this season. BC was their third such game in a row and 4th in 6 games dating back to Duke. They have been tempting fate for a long time now.

With respect to the original question of Syracuse falling in rankings. Winning is the ultimately the only thing that matters for championships and polls, but Pomeroy's ratings (which are win/loss blind) have been lowering Syracuse. They were a strong second at the time of the Duke game, #5 before the BC game and are currently #9 after losing to BC.

[Monmouth77]

I bring this up only because I hope KenPom goes Jay Bilas on this topic to change people's minds about what constitutes a good win/bad loss. It would be cool to see him develop some type of equivalency feature so that people can understand what that win/loss means in relation to home court.

On another note, I'd also like to see him argue about the unimportance of actually winning, meaning that 'Cuse's win over NC State, albeit a win, would actually hurt them. This topic would be quite a bit more controversial b/c our sports are predicated on winning but it always amazes me that Syracuse needs 2 buzzer beater wins to squeak by, yet at the end of the day, we ignore who they were against and just look at the W. I suppose this is what Kenpom's luck measures and I imagine most teams would have bad games (Duke/Vermont or Duke/Maryland come to mind) so it might even itself out. I just never understood rewarding or not punishing Syracuse for a game like NC State.

I'll bite on this. I love the idea of an equivalency feature that would bring some texture and focus to the KenPom ratings. It would help the human pollsters and NCAA Tournament selection committee (eventually) make better evaluations of wins and losses and make better seeding decisions.

But I am really bothered by the idea that winning is "unimportant." Maybe this just dovetails with my overall instincts that we must balance the ever creeping advance of technology (and statistical insight) with a bit of humanity, but I mean, winning is hugely important. It may not be "everything" or "the only thing" to look at in assessing a team's relative strengths and weaknesses, but it has instrinsic value of its own. You play to win the game. And if you play well but fail to win, that might mean something positive, but it also means you failed in the objective of the game, which is to have the most points at the end of 40 minutes -- not to have played the most efficiently in 40 minutes.

Yes, it is perhaps the case that "luck" instead of "moxie" or "clutch play" accounts for the difference in close games, but that ought to even out over time, and even if it doesn't, history rightly rewards close game winners (not more efficient close game losers). Who cares who played more efficiently in "The Greatest Game Ever Played"? I don't. I care that Duke executed the greatest buzzer beating play of all time to win the game.

[Wander]

I agree with you to some extent, though it is worth noting that the ultimate prize (NCAA title) is predicated solely on winning. If you play great and dominate but things fall apart and you get really, really unlucky... your fate is exactly the same as the team that got their door blown off by 30 points.




But I am really bothered by the idea that winning is "unimportant." Maybe this just dovetails with my overall instincts that we must balance the ever creeping advance of technology (and statistical insight) with a bit of humanity, but I mean, winning is hugely important. It may not be "everything" or "the only thing" to look at in assessing a team's relative strengths and weaknesses, but it has instrinsic value of its own. You play to win the game. And if you play well but fail to win, that might mean something positive, but it also means you failed in the objective of the game, which is to have the most points at the end of 40 minutes -- not to have played the most efficiently in 40 minutes.


Predictive rankings like kenpom should care solely about efficiency because that is the most accurate and objective way to build data to predict games. Selection and seeding for the NCAA tournament should care solely about winning because that's the actual objective of the game. This is why one should never use "team X is ranked Y in kenpom so they should be in the tournament."

[sporthenry]

But I am really bothered by the idea that winning is "unimportant." Maybe this just dovetails with my overall instincts that we must balance the ever creeping advance of technology (and statistical insight) with a bit of humanity, but I mean, winning is hugely important. It may not be "everything" or "the only thing" to look at in assessing a team's relative strengths and weaknesses, but it has instrinsic value of its own. You play to win the game. And if you play well but fail to win, that might mean something positive, but it also means you failed in the objective of the game, which is to have the most points at the end of 40 minutes -- not to have played the most efficiently in 40 minutes.

I guess saying winning is unimportant was the wrong thing to say. Perhaps, we should stop our over reliance on wins and losses. Wins and losses still matter but there is something beyond that (sort of like how pitcher's wins and losses has evolved). I guess it depends what you think the selection committee's job is but if it is to put the best teams in the tourney/give the best teams the highest seeds, then I think efficiency matters. If its to reward W/L, fair enough but then we should acknowledge that.

Now granted, the NCAAT will always be random and I'm not suggesting that changes. It is my favorite time of the year, partially b/c of the upsets, but again, the best team doesn't always win. If we wanted that, they'd go the NBA route but that does get boring at times.

[RepoMan]

I was looking at the Pomeroy ratings, and it looked like there were an unusual number of teams that, like Duke, sport an imbalanced Offense / Defense rating. I don't have the patience to figure out if that is true, but it reminded me that I thought it was the case that, in most seasons, the National Champion has been rated in the top 20 in both. I did take the time to look at that, and it appears that that is true in every year except 2009, when UNC missed the rule by a hair with a 21 Defense rating.

At the moment, only Louisville, Florida, Wichita St, and Villanova pass that test. Maybe this is a good year to break the streak.

(Note: Syracuse and Kansas aren't too far off.)

[ice-9]

I was looking at the Pomeroy ratings, and it looked like there were an unusual number of teams that, like Duke, sport an imbalanced Offense / Defense rating. I don't have the patience to figure out if that is true, but it reminded me that I thought it was the case that, in most seasons, the National Champion has been rated in the top 20 in both. I did take the time to look at that, and it appears that that is true in every year except 2009, when UNC missed the rule by a hair with a 21 Defense rating.

At the moment, only Louisville, Florida, Wichita St, and Villanova pass that test. Maybe this is a good year to break the streak.

(Note: Syracuse and Kansas aren't too far off.)

Those rankings were post-tournament, but if you look at pre-tournament data (think Kedsy has them) some of the national champs didn't have a top 20 rating in both offence and defence. The 2009 UNC team for instance had a defence that was much lower than 21 pre-tournament and then kicked it up a notch to improve all the way there.

[tbyers11]

Those rankings were post-tournament, but if you look at pre-tournament data (think Kedsy has them) some of the national champs didn't have a top 20 rating in both offence and defence. The 2009 UNC team for instance had a defence that was much lower than 21 pre-tournament and then kicked it up a notch to improve all the way there.

I don't have the Selection Sunday KenPom database that Kedsy does but I have seen the data enough times to memorize the fact that 2009 UNC's D Efficiency rank was 35 and that improved to 21 over the course of the tournament.

[Kedsy]

Those rankings were post-tournament, but if you look at pre-tournament data (think Kedsy has them) some of the national champs didn't have a top 20 rating in both offence and defence. The 2009 UNC team for instance had a defence that was much lower than 21 pre-tournament and then kicked it up a notch to improve all the way there.


I don't have the Selection Sunday KenPom database that Kedsy does but I have seen the data enough times to memorize the fact that 2009 UNC's D Efficiency rank was 35 and that improved to 21 over the course of the tournament.

You are both correct. Here are pre-tournament ranking for last five national champs:

2013 Louisville: 15 offense, 1 defense
2012 Kentucky: 2 off., 6 def.
2011 UConn: 21 off., 31 def.
2010 Duke: 1 off., 4 def.
2009 UNC: 1 off., 35 def.

If you look at Final Four teams, you have such gems as 2011 VCU (59 off., 143 def.), 2011 Butler (39 off., 77 def), 2010 Michigan State (38 off., 27 def.), 2009 Villanova (25 off., 25 def.), 2013 Wichita State (50 off., 30 def.), and so on. In fact, in the past five years, 14 of the 20 Final Four teams (70%) have been outside the top 20 in one or the other (or both) of the Pomeroy efficiency rankings.

There is no formula. You just have to play well at the right time.

[Kedsy]

On February 10, our Pomeroy AdjD rating was 100.4, ranked #92 in the country.

Now, five games later, our Pomeroy AdjD is 97.9, ranked #58 in the country.

If we get to the ACC final and make another 2.5 point jump over our next five games, we'll go into the NCAA tournament with a 95.4 AdjD rating, and would be ranked approximately #26 in the country.

I think we're pointing in the right direction.

[COYS]

On February 10, our Pomeroy AdjD rating was 100.4, ranked #92 in the country.

Now, five games later, our Pomeroy AdjD is 97.9, ranked #58 in the country.

If we get to the ACC final and make another 2.5 point jump over our next five games, we'll go into the NCAA tournament with a 95.4 AdjD rating, and would be ranked approximately #26 in the country.

I think we're pointing in the right direction.

Bingo. And while our offense has slumped just a bit (slump is a relative term, though), I would be willing to bet that the rest over the next week might be just what the doctored ordered to improve our three point shooting and reestablish our offense as a dominant force. Personally, I'm happy with where our team is headed, especially considering that we've been improving on defense and still have some obvious areas in which we can improve, offensively (get Andre back on track, get Quinn healthy and effective, find Jabari's outside shot so that he can hit the open three from time to time).

[sagegrouse]

2013 Louisville: 15 offense, 1 defense
2012 Kentucky: 2 off., 6 def.
2011 UConn: 21 off., 31 def.
2010 Duke: 1 off., 4 def.
2009 UNC: 1 off., 35 def.



If you do the statistically questionnable -- average ordinal rankings -- then the NCAA champs average #8 on offense and #19 on defense. Medians are #2 on offense and #6 on defense.

So, what wins the NCAA championship -- offense or defense?

[Troublemaker]

One data point that seems relevant is last year's Michigan team that entered the NCAAT with the #2 offense and #51 defense and almost won the NCAA title. I mean, late in the 1st half of the championship game against Louisville, the Wolverines had the lead and a ~75% expected win probability for the game before Louisville's Luke Hancock went wild.

What does this mean? Perhaps nothing.

[Kedsy]

So, what wins the NCAA championship -- offense or defense?

I think doing one or the other exceptionally well, and then performing better than projected with the other. (I suppose if you do both offense and defense exceptionally well (e.g., 2010 Duke, 2012 UK), then you don't have to outperform the projections but merely meet them.)

Although 2011 UConn wouldn't fit that profile, so maybe there's no formula whatsoever.

[toooskies]

It seems very odd to see declarations about who can or can't win a championship based on stats whose very nature are designed to predict probabililties of future success. One of the assumptions of the KenPom model is that any outcome is possible, and its likelihood is dependent upon previous performance. And at the very least, 11 championship winners isn't enough to come to any valid conclusion about who can or can't win.

[davekay1971]

I think doing one or the other exceptionally well, and then performing better than projected with the other. (I suppose if you do both offense and defense exceptionally well (e.g., 2010 Duke, 2012 UK), then you don't have to outperform the projections but merely meet them.)

Although 2011 UConn wouldn't fit that profile, so maybe there's no formula whatsoever.

KenPom does a nice job of helping to identify teams that are consistently better at phases of the game, that's all. A team that, over the course of a full season and conference tournament, has been exceptionally good at offense or defense SHOULD be exceptionally good at that aspect of the game in the tournament. However, this is single elimination, and young players who may play well outside of their norm for a given tournament game. All it takes if for a team to play well above or below their efficiency for one game for them to advance beyond their expectation, or fall far below it.

The only formula I see is that teams which rank very high in offense and defense, or exceptionally high in one and good in the other, have a consistently better chance of performing well in the NCAAT.

[pfrduke]

It's fun watching the collateral effect of game-by-game performance on the teams at the top. For example, Wisconsin beats American by a lot? Florida (which played the Badgers) sees its rating improve slightly. Pitt wallops Colorado? Arizona (which played the Buffs twice) sees its rating dip. Arizona, Louisville, and Florida haven't touched the court since Sunday and the performance of other teams has taken them from 1-2-3 in that order when the field was announced to now being Louisville-Florida-Arizona. The 'ville is likely to drop slightly after Harvard's defeat of Cincinnati, too.

ACC performance in the NIT (and the aforementioned Pitt win) have also leapfrogged Duke above Villanova into the #6 spot (but we're a healthy shouting distance from the Wheat Shockers).

[Kedsy]

...the two things Wisconsin is famous for are playing a very slow tempo and playing a pretty weak non-conference schedule. So I looked through Pomeroy's historical rankings for teams with a tempo in the bottom 50 of the country and a NCSOS in the bottom 150. Since the "problem" seems to date back to 2010, I looked at the last four seasons:



Year Team Pom Rk Tempo Rk NCSOS rk Seed Result
2013 Georgetown 12 340 298 2 Lost to 15 seed in round of 64
2013 Wisconsin 13 318 241 5 Lost to 12 seed in round of 64
2013 Kansas St 21 299 232 4 Lost to 13 seed in round of 64
2013 Notre Dame 37 319 322 7 Lost to 10 seed in round of 64
2012 Wisconsin 7 345 271 4 Lost to 1 seed in round of 16
2012 Michigan 23 329 204 4 Lost to 13 seed in round of 64
2012 Notre Dame 37 328 239 7 Lost to 10 seed in round of 64
2011 Pitt 3 318 269 1 Lost to 8 seed in round of 32
2011 Wisconsin 8 344 225 4 Lost to 8 seed in round of 16
2011 Cincinnati 21 299 335 6 Lost to 3 seed in round of 32
2011 Texas A&M 49 326 327 7 Lost to 10 seed in round of 64
2010 Pitt 25 322 229 3 Lost to 6 seed in round of 32
2010 Marquette 30 304 308 6 Lost to 11 seed in round of 64
2010 Notre Dame 31 312 244 6 Lost to 11 seed in round of 64


So, in the last four tournaments, there were 14 teams seeded 7 or better with tempo in the lowest 50 and NCSOS in the bottom 150. Of those 14 teams, twelve of them lost to teams seeded 3 or more spots below them, including 9 first round upsets and a #8 beating a #1. Ironically, the only team in the group to even get to the Sweet 16 was Wisconsin, which did it twice.

Teams that didn't quite make the cut:

2012 Georgetown (300 tempo; 152 NCSOS), 3 seed lost to 11 seed in round of 32
2012 St. Mary's (261 tempo; 233 NCSOS), 7 seed lost to 10 seed in round of 64
2011 Notre Dame (273 tempo; 263 NCSOS), 2 seed lost to 10 seed in round of 32
2010 Wisconsin (340 tempo; 131 NCSOS), 4 seed lost to 12 seed in round of 32

I used the new Pomeroy formula and end-of-tournament rankings, which might skew things a bit, but probably not too much since teams don't change tempo or non-conference schedule all that much in the tournament. Also, full disclosure, if you go back the 7 seasons from 2003 to 2009, there are only 7 such teams and only 1 of them lost to team seeded 3 spots or more below them (and none of the 7 teams went past the Sweet 16, but 5 of the 7 reached the Sweet 16). Still, Wisconsin doesn't appear to have been a problem before 2010, and there were so few teams that met the criteria in those years (1.0 per year as opposed to 3.5 per year in the later sample), so I'm running with it.

Now just because they all lost in the NCAAT doesn't mean they're overrated in Pomeroy, but it might. And either way it's pretty amazing, right? I don't know about you, but when I'm filling out my bracket this year I'll be looking for teams with slow tempos and bad non-conference schedules as potential upset victims. For what it's worth, there are three teams currently in danger of meeting my criteria in the upcoming NCAA tournament: Syracuse (345 tempo; 218 NCSOS), Cincinnati (318 tempo; 315 NCSOS), and Pittsburgh (290 tempo; 285 NCSOS). You heard it here first.

I posted the above in the beginning of Feburary. It turned out that the only teams meeting my criteria this year were Cincinnati and Syracuse, and guess what? Both of them lost early to teams seeded 3 or more spots below them, making the count for the last five years 14 losers out of 16 teams meeting the criteria.

I put my money where my mouth was and picked Harvard and Dayton to win. Of course, I chose Duke as my champion so my bracket's busted anyway, but I still think this particular stat is pretty amazing.

[Troublemaker]

I posted the above in the beginning of Feburary. It turned out that the only teams meeting my criteria this year were Cincinnati and Syracuse, and guess what? Both of them lost early to teams seeded 3 or more spots below them, making the count for the last five years 14 losers out of 16 teams meeting the criteria.

I put my money where my mouth was and picked Harvard and Dayton to win. Of course, I chose Duke as my champion so my bracket's busted anyway, but I still think this particular stat is pretty amazing.

Great research here. I wish I had remembered all this when picking my various brackets. I had Syracuse to at least the Elite 8 in all of them.

[uh_no]

Great research here. I wish I had remembered all this when picking my various brackets. I had Syracuse to at least the Elite 8 in all of them.

no sympathy! they won a lot of tight games, and then their karma ran out and they finally lost some games down the stretch....without a strong non-conference, there was little recommending them, except for perhaps their play against duke, if you had held duke in very high regard this year