PDA

View Full Version : A troubling trend



forbiddendonut
03-12-2012, 10:23 AM
Based on kenpom numbers, Duke is an incredibly weak #2 seed. In the kenpom era (2003-), these are the three #2 seeds whose kenpom numbers were in the same ballpark as the 2012 Duke Blue Devils: 2003 Wake Forest, 2006 Tennessee and 2010 Villanova. Let's look at how they did in the tournament (or just look away now if you are a Duke fan and have a weak stomach) -

2003 Wake: Beat #15 East Tenn. St. by 3 and lost to #10 Auburn by 6
2006 Tennessee: Beat #15 Winthrop by 2 and lost to #7 Wichita St by 7
2010 Villanova: Beat #15 Robert Morris by 3 and lost to #10 St Mary's by 7

I swear I did not pick these three teams knowing that they performed similarly in the tournament. I just looked back to find teams that were similar in ranking to Duke and also placed on the #2 line and then found their results which are strikingly consistent (and deeply troubling if you are a Duke fan).

Only 2003 Wake (0.8928) and 2006 Tennessee (0.8897) are behind Duke (0.8971) in kenpom ranking but their numbers include their putrid performance in the NCAA tournament [2010 Nova ended up at 0.9011]. I would guess that at the time of seeding, Duke is the lowest rated #2 seed of the kenpom era.

moonpie23
03-12-2012, 10:31 AM
man, all this wailing and bemoaning and gnashing of teeth.....sheesh.....we're gonna play who we play and see what happens.....I think all the negative AND positive aspects of this year's team have be gone over and over and over, but, it is what it is....

matchups may or may NOT favor us, we're gonna find out friday night just what our "game face" looks like for this big dance...

i'm not saying just put your blue glasses on and let it slide, i'm just saying this is where we are.....it's almost like we're just setting up all the excuses for if or when we lose.....

we're gonna find out who's gonna walk the walk this weekend...

jv001
03-12-2012, 10:39 AM
Based on kenpom numbers, Duke is an incredibly weak #2 seed. In the kenpom era (2003-), these are the three #2 seeds whose kenpom numbers were in the same ballpark as the 2012 Duke Blue Devils: 2003 Wake Forest, 2006 Tennessee and 2010 Villanova. Let's look at how they did in the tournament (or just look away now if you are a Duke fan and have a weak stomach) -

2003 Wake: Beat #15 East Tenn. St. by 3 and lost to #10 Auburn by 6
2006 Tennessee: Beat #15 Winthrop by 2 and lost to #7 Wichita St by 7
2010 Villanova: Beat #15 Robert Morris by 3 and lost to #10 St Mary's by 7

I swear I did not pick these three teams knowing that they performed similarly in the tournament. I just looked back to find teams that were similar in ranking to Duke and also placed on the #2 line and then found their results which are strikingly consistent (and deeply troubling if you are a Duke fan).

Only 2003 Wake (0.8928) and 2006 Tennessee (0.8897) are behind Duke (0.8971) in kenpom ranking but their numbers include their putrid performance in the NCAA tournament [2010 Nova ended up at 0.9011]. I would guess that at the time of seeding, Duke is the lowest rated #2 seed of the kenpom era.

I guess we can dispell this by beating Lehigh by 10+ points. GoDuke!

Bluedog
03-12-2012, 10:41 AM
I guess we can dispell this by beating Lehigh by 10+ points. GoDuke!

Well, even Michigan State, a #1 seed, didn't beat Lehigh by 10+ points this season. Lehigh is a decent team and shouldn't be taken lightly.

CDu
03-12-2012, 10:42 AM
We'll probably play both games closer than we should. But I think we'll win both games, because we're the far better team in each case. Lehigh creates matchup problems, but we're just far better talentwise overall. Notre Dame is a great matchup for us, and Xavier creates some matchup problems but not enough to beat us without help from us.

We're the type of team that you don't want to face if you're a top seed because we can defend inside and shoot the 3. But we're not a team that I'd feel very comfortable picking to win it all, or even make the Elite 8. But I am pretty confident that we'll get out of the first weekend.

COYS
03-12-2012, 10:58 AM
Duke's an interesting team, however, in that we've managed to defy the KenPom odds for a large part of the season. I think that to achieve something big in the NCAAT this season, we've got to get our offense clicking. When it's on, it's been ranked as high as number 1 in Pomeroy. The the two ACCT games really ate into our offensive ranking. However, I still think the potential is there. I wonder if it will be an advantage for us to play opponents that are unfamiliar with us. If Ryan is back and healthy, opposing teams will have to remember to guard as many as 4 players tight on the perimeter, even in transition. VaTech and FSU did a good job staying at home on our shooters, but they also had the benefit of playing us multiple times this season. However, our NCAAT opponents will have to fight their instincts to help on drives and guard the paint in transition, which will hopefully leave our shooters open on the perimeter.

Scorp4me
03-12-2012, 11:01 AM
Well then I guess it's good news we didn't get a #1 seed. Would hate to be the first #1 seed to lose in the first round. And second...haven't we already beat quite a few #1 seeds this year?

MCFinARL
03-12-2012, 11:14 AM
Well then I guess it's good news we didn't get a #1 seed. Would hate to be the first #1 seed to lose in the first round. And second...haven't we already beat quite a few #1 seeds this year?

yes, half of them. To be fair, though, we caught Michigan State early, and they are playing much better now than they were then.

OldPhiKap
03-12-2012, 11:16 AM
I think a possible difference between this Duke team and the ones you mention is tournament experience (both players and coaches). The juniors and seniors have been on a team that went all the way and know how to deal with the hubbub. K certainly does as well, obviously. Not sure how that compares to the three examples. Also, not sure if they played about an hour from their campus or not.


While I think we have a very challenging game Friday, with two good teams battling to play the winner on Sunday, I think we generally do a good job of getting into tournament mode. Maui may have been a long time ago, but this team has experience winning a tournament under its belt.

Should be fun, Lehigh is good. Game on.

Des Esseintes
03-12-2012, 11:22 AM
Based on kenpom numbers, Duke is an incredibly weak #2 seed. In the kenpom era (2003-), these are the three #2 seeds whose kenpom numbers were in the same ballpark as the 2012 Duke Blue Devils: 2003 Wake Forest, 2006 Tennessee and 2010 Villanova. Let's look at how they did in the tournament (or just look away now if you are a Duke fan and have a weak stomach) -

2003 Wake: Beat #15 East Tenn. St. by 3 and lost to #10 Auburn by 6
2006 Tennessee: Beat #15 Winthrop by 2 and lost to #7 Wichita St by 7
2010 Villanova: Beat #15 Robert Morris by 3 and lost to #10 St Mary's by 7

I swear I did not pick these three teams knowing that they performed similarly in the tournament. I just looked back to find teams that were similar in ranking to Duke and also placed on the #2 line and then found their results which are strikingly consistent (and deeply troubling if you are a Duke fan).

Only 2003 Wake (0.8928) and 2006 Tennessee (0.8897) are behind Duke (0.8971) in kenpom ranking but their numbers include their putrid performance in the NCAA tournament [2010 Nova ended up at 0.9011]. I would guess that at the time of seeding, Duke is the lowest rated #2 seed of the kenpom era.

It's worth recalling here that you are comparing apples to oranges somewhat. The rankings you cite are year-end numbers. The thing is, the tournament itself causes a great deal of movement at the top of the rankings, where the teams that advance also tend to rise. If those three #2 seeds went out early, they were going to fall compared to the teams that survived. Similarly, if Duke makes the Final Four, I guarantee its kenpom ranking will be top ten. As the numbers people would say, those kenpom rankings are descriptive, not predictive.

On the other hand, looking at pre-tourney kenpom numbers would be interesting, but I don't think we have general access to those. Although Kedsy has made an annual habit of saving those pages, I think.

MChambers
03-12-2012, 11:39 AM
I wouldn't worry about it. It's a very small sample size. In the pre-KenPom era, lots of weak #2s did pretty well. The 1990 Duke team was really struggling going into the NCAA tournament and reached the final. (Don't ask what happened then.)

OldPhiKap
03-12-2012, 11:43 AM
I wouldn't worry about it. It's a very small sample size. In the pre-KenPom era, lots of weak #2s did pretty well. The 1990 Duke team was really struggling going into the NCAA tournament and reached the final. (Don't ask what happened then.)

True. And we were in a worse position when we played Temple in '88 with Mark Macon. No way Duke can win that game. . . .

Matches
03-12-2012, 12:41 PM
I wouldn't worry about it. It's a very small sample size. In the pre-KenPom era, lots of weak #2s did pretty well. The 1990 Duke team was really struggling going into the NCAA tournament and reached the final. (Don't ask what happened then.)

I'm pretty sure that game was never played. They held the whole tournament up to the final and then just... stopped.

That's how I remember it, at least.

Jderf
03-12-2012, 12:57 PM
I wouldn't worry about it. It's a very small sample size. In the pre-KenPom era, lots of weak #2s did pretty well. The 1990 Duke team was really struggling going into the NCAA tournament and reached the final. (Don't ask what happened then.)

Yeah, it's hard to really call this a "trend." More like a few scattered, possibly related events -- hardly enough to guarantee doom and gloom. I actually think it is somewhat promising, seeing as, when it comes to the NCAA tournament, past events seem to be a perfect predictor of what isn't going to happen the second time around.

forbiddendonut
03-12-2012, 01:12 PM
The similarity in score margins was most striking to me. I should not have used the word trend. I thought it was interesting.

I don't have the pre-NCAAT numbers (as referenced above) but other #2 seeds in the 0.88-0.91 range pre-NCAAT may include: 2006 Ohio St, 2009 Oklahoma, 2009 Michigan St, 2011 UNC and 2011 Florida.

If anyone has the pre-NCAAT rankings saved, that would eliminate the guessing.

pfrduke
03-12-2012, 01:19 PM
2006 Ohio St, 2009 Oklahoma, 2009 Michigan St, 2011 UNC and 2011 Florida

Results in order:
R32 - lost to #7 Georgetown
E8 - lost to #1 UNC
Runner-up - lost to #1 UNC
E8 - lost to #4 Kentucky
E8 - lost to #8 Butler

Indoor66
03-12-2012, 01:23 PM
Results in order:
R32 - lost to #7 Georgetown
E8 - lost to #1 UNC
Runner-up - lost to #1 UNC
E8 - lost to #4 Kentucky
E8 - lost to #8 Butler

Past performance does not predict future results.

pfrduke
03-12-2012, 01:24 PM
Past performance does not predict future results.

Agree 100% (particularly when the results are so disparate). Just providing information.

dukeballboy88
03-12-2012, 01:29 PM
We should beat Lehigh and Xavier or Notre Dame in Greensboro regardless of what Ken Pom says. Now Baylor may pose a problem if they get passed South Dakota St. South Dakota St is my upset special and if that happens, Duke should play Kentucky in the elite 8. Take that KenPom.

MCFinARL
03-12-2012, 03:33 PM
Past performance does not predict future results.

Who are you, my broker?

NSDukeFan
03-12-2012, 03:48 PM
I am a big fan of KenPom and his ratings. Things have seemed a bit odd this year with his ratings though. The big Ten teams seemed to be ranked way too high, starting with Wisconsin, who was ranked #1 for much of the preseason, I believe. They are still ranked #6, though I doubt there are many who believe they are the #6 team in the country. I also have my doubts that Memphis should be #9, Wichita State #10, and St. Louis #15 all ahead of Duke, who has a much better resume. Maybe KenPom's rankings are a more accurate look at the stronger teams and Duke is not as strong as their record and resume, but I believe (perhaps through dark blue glasses) Duke is a legitimate #2 seed and a few points from a potential #1 seed.

Troublemaker
03-12-2012, 04:27 PM
To be concerned or not just depends on whether Duke's "Luck" rating in Pomeroy is just that -- luck -- or whether it is a reflection of being some parts lucky and some parts clutch.

As others may have noted, Duke's Luck rating has been very high all season long. Meaning theoretically, Duke's record is better than Duke's true worth as a team.

However, imo, Duke does in fact have some considerable clutchness in them. That doesn't mean that Duke hits big shots all the time and never, ever fails. But, certain players on this team have always struck me as "big shot" makers -- namely Dawkins, Curry, Thornton, Kelly, and somewhat Rivers (obviously he hit the biggest clutch shot of them all this season, though). It's a guttural thing but I'm very comfortable with those guys taking a shot in high-leverage situations.

If I'm right and Duke is legitimately clutch and can be expected to win more than our fair share of close games, then Duke's Pomeroy rating will undervalue Duke.

Kedsy
03-12-2012, 04:37 PM
The similarity in score margins was most striking to me. I should not have used the word trend. I thought it was interesting.

I don't have the pre-NCAAT numbers (as referenced above) but other #2 seeds in the 0.88-0.91 range pre-NCAAT may include: 2006 Ohio St, 2009 Oklahoma, 2009 Michigan St, 2011 UNC and 2011 Florida.

If anyone has the pre-NCAAT rankings saved, that would eliminate the guessing.


It's worth noting that this year's Pomeroy pythagorean numbers are much lower in general. I don't know why. I mean, 2009 Michigan State would be #6 in this year's Pomeroy, and 2009 Oklahoma would be #9. Could he have changed his formula, or is there more parity, or what?

Anyway, here are the pre-tournament numbers for teams mentioned here in the 2009-2011 date range:

2009 Oklahoma, #17 (.9340)
2009 Michigan State, #13 (.9425)
2010 Villanova, #15 (.9369)
2011 UNC, #14 (.9379)
2011 Florida, #19 (.9221)

2012 Duke, #17 (.8971)


When my computer died I lost all my Pomeroy data earlier than 2009, so I only have the past three years at this point.

crimsondevil
03-12-2012, 04:39 PM
To be concerned or not just depends on whether Duke's "Luck" rating in Pomeroy is just that -- luck -- or whether it is a reflection of being some parts lucky and some parts clutch.

Not that I necessarily disagree (although I am always suspicious of "clutch-ness" as a skill), but the luck rating could also reflect any quality/skill that is not included in Pomeroy's rating.

Possibly worth noting again that Pomeroy himself calls Wisconsin's rating an anomaly - may apply to others as well (SLU?).

loran16
03-12-2012, 04:45 PM
Based on kenpom numbers, Duke is an incredibly weak #2 seed. In the kenpom era (2003-), these are the three #2 seeds whose kenpom numbers were in the same ballpark as the 2012 Duke Blue Devils: 2003 Wake Forest, 2006 Tennessee and 2010 Villanova. Let's look at how they did in the tournament (or just look away now if you are a Duke fan and have a weak stomach) -

2003 Wake: Beat #15 East Tenn. St. by 3 and lost to #10 Auburn by 6
2006 Tennessee: Beat #15 Winthrop by 2 and lost to #7 Wichita St by 7
2010 Villanova: Beat #15 Robert Morris by 3 and lost to #10 St Mary's by 7

I swear I did not pick these three teams knowing that they performed similarly in the tournament. I just looked back to find teams that were similar in ranking to Duke and also placed on the #2 line and then found their results which are strikingly consistent (and deeply troubling if you are a Duke fan).

Only 2003 Wake (0.8928) and 2006 Tennessee (0.8897) are behind Duke (0.8971) in kenpom ranking but their numbers include their putrid performance in the NCAA tournament [2010 Nova ended up at 0.9011]. I would guess that at the time of seeding, Duke is the lowest rated #2 seed of the kenpom era.

But you're missing the remaining #s. Let's look at the opponents these teams lost to:
2010 Nova - St. Mary's - .8591 Pythag Rating (Admittedly including win over Nova)
2006 Tenn - Wichita State - .8491 Pythag (Including win over Tennessee).

Both of these teams faced opponents who were within 20 kenpom ranks of them. Or less in Tenn's case.

Now let's look at Duke's potential 2nd round matchups:
.8184 Notre Dame
.7812 Xavier.

These opponents are a good deal worse than SMC and WSU.

So yeah, it's true we're a pretty weak 2 seed. But we got the easiest draw to the S16.

DukieTiger
03-12-2012, 04:57 PM
It's worth noting that this year's Pomeroy pythagorean numbers are much lower in general. I don't know why. I mean, 2009 Michigan State would be #6 in this year's Pomeroy, and 2009 Oklahoma would be #9. Could he have changed his formula, or is there more parity, or what?

Anyway, here are the pre-tournament numbers for teams mentioned here in the 2009-2011 date range:

2009 Oklahoma, #17 (.9340)
2009 Michigan State, #13 (.9425)
2010 Villanova, #15 (.9369)
2011 UNC, #14 (.9379)
2011 Florida, #19 (.9221)

2012 Duke, #17 (.8971)


When my computer died I lost all my Pomeroy data earlier than 2009, so I only have the past three years at this point.



I know that these are post-tournament numbers, but bear with me here...

2012 Duke is currently #17 and at .8971 as Kedsy said.

2007 Duke, after losing to VCU in the tournament was ranked #11 at .9491. That's way higher than this year's Duke team, who has actually had a better season by far. That's higher than this year's UNC team as well. That's higher than every single final four team from last year. Additionally, there were 6 teams in 2007 that ranked higher than Kentucky is right now. You had to go all the way to the low 40s before you found a team ranked similarly to Duke this year.

That's all to say that something isn't consistent from year to year in KenPom's rankings. I don't know if he's changed his formula or whether this is just the relative difference in the teams from each year. But how can a team that went 22-11 rank better than a team that has beaten 2 of the top 3 in KenPom's rankings? How can that team rank higher (based on Pyth%) than this year's UNC team? How can that team rank higher in its respective year than EVERY SINGLE final four team from last year?

Pomeroy's rankings are simply an indicator of how likely that team is to win a game against an average team in that given year. It's tough to use it to say anything conclusive across seasons. Especially when you have teams like Butler and VCU making the tournament last year.

camion
03-12-2012, 05:07 PM
To be concerned or not just depends on whether Duke's "Luck" rating in Pomeroy is just that -- luck -- or whether it is a reflection of being some parts lucky and some parts clutch.

As others may have noted, Duke's Luck rating has been very high all season long. Meaning theoretically, Duke's record is better than Duke's true worth as a team.

However, imo, Duke does in fact have some considerable clutchness in them. That doesn't mean that Duke hits big shots all the time and never, ever fails. But, certain players on this team have always struck me as "big shot" makers -- namely Dawkins, Curry, Thornton, Kelly, and somewhat Rivers (obviously he hit the biggest clutch shot of them all this season, though). It's a guttural thing but I'm very comfortable with those guys taking a shot in high-leverage situations.

If I'm right and Duke is legitimately clutch and can be expected to win more than our fair share of close games, then Duke's Pomeroy rating will undervalue Duke.


If you are consistently lucky it over an extended time then it probably isn't luck. Maybe it's preparation meeting opportunity.

loran16
03-12-2012, 05:33 PM
If you are consistently lucky it over an extended time then it probably isn't luck. Maybe it's preparation meeting opportunity.

Or it's luck. A single season is not a long period of time statistically (30 games) - there are bound to be outliers in those seasons, whose results are better than they should be simply due to random variation.