Week of 2/22/10 polls

[pfrduke]

Something that could be fun to track...

The Pomeroy ratings this morning-

1. Duke .9818
2. Kansas .9817
3. W Va .9643
4. Syrac .9638
5. Purdue .9620

I list these because it will be interesting to see how a couple games tonight affect things.

Kansas is playing Oklahoma, who is not highly regarded by KenPom being ranked #98. Kansas is at home and a prohibitive favorite in the game. How big a win will they need to get points and move ahead of Duke again?

WVa obviously just lost to UConn (the #43 team at .8912). How much will they lose from that loss?

With not a lot of other games on the national slate tonight, it will be interesting to see how these limited results affect things.

--Jason "tune in tomorrow for an update" Evans

Well, Kansas is killing OU. Even with OU being ranked low, a 22-point victory ought to be enough to get a nudge back over Duke. WVA will very likely, I believe, fall below Syracuse, and possibly Purdue as well, but not by much.

[Hermy-own]

More important than the human poll, Duke is now the #1 rated team in Ken Pomeroy's rankings. This is a very big deal because this season, for the first time (I think), the selection committee will bring Ken's rankings with them into the room.

So, the two sources (aside from actual game results) that the NCAA will us to seed the tournament will be the RPI and the KenPom ratings. Duke is #1 in the Pomeroy and #2 in the RPI. That would suggest that we have an excellent chance to get a #1 seed.

--Jason "worth noting, us and Kansas are a lot ahead of the other teams in the Pom rankings" Evans

I don't want to distract this thread from it's discussion which is the pursuit of a #1 seed - there are other threads for discussion of whether we deserve one. I just want to say that this is exactly the sort of statistic that makes me distrust Pomeroy. I think most people watching last night's game can agree that it was far from a good game, we shot way too many 3's, and that we showed lots of weaknesses. And suddenly we jump over Kansas to #1 in the Pomeroy rankings? Is there anyone who thinks we are actually the #1 team in the country? That doesn't make me have a lot of faith in the computer generate rankings.

[M B Walker]

Aside from bragging rights, I'm not sure there's much of a difference between being the 4th #1 seed and the 1st #2 seed. By the S-curve we'd end up in the same bracket anyway.

And that would probably be out West. The four regional hosts are East (Syracuse), Midwest (St Louis), South (Houston), and West (Salt Lake). My first guess is Duke and Syracuse being slotted out West, with Duke playing a first/second round game in Buffalo. Or perhaps in the East, in a bracket with someone like Kentucky (which the NCAA would love), playing the first weekend in Jacksonville. Of those two, I think I'd prefer being matched up with Kentucky.

With no games anywhere in ACC country this year, the bracket will be at least as interesting as our seed.

[pfrduke]

I don't want to distract this thread from it's discussion which is the pursuit of a #1 seed - there are other threads for discussion of whether we deserve one. I just want to say that this is exactly the sort of statistic that makes me distrust Pomeroy. I think most people watching last night's game can agree that it was far from a good game, we shot way too many 3's, and that we showed lots of weaknesses. And suddenly we jump over Kansas to #1 in the Pomeroy rankings? Is there anyone who thinks we are actually the #1 team in the country? That doesn't make me have a lot of faith in the computer generate rankings.

I'm not sure we're the best team in the country; I don't think we are, but I'm also not sure we're not. But I have two essentially unrelated comments in response.

First, don't put too much stock in one game. Did Kansas look like the number 1 team in the country when it barely beat Cornell at home, or when Colorado took it to overtime on the road?

Second, while we may not be the best team, I agree with Pomeroy's rankings in the following respect - we have played, on the aggregate, as well relative to our competition as any other team in the country, Kansas included. Duke has played 13 top-50 teams, gone 10-3 against them, and outscored them by a total of 155 points - an average of almost 12 points per game.

I know people keep waiting for the other shoe to drop with this team, and maybe it will. But I think may folks have greatly undersold the performance that this team has put in to date. We are, and have played like, a very good team.

[DukeUsul]

Aside from bragging rights, I'm not sure there's much of a difference between being the 4th #1 seed and the 1st #2 seed. By the S-curve we'd end up in the same bracket anyway.

Oh but it would make a huge difference. It would determine jersey color!

Edit: talking specifically about a hypothetical Final 8 game.

[pfrduke]

Aside from bragging rights, I'm not sure there's much of a difference between being the 4th #1 seed and the 1st #2 seed. By the S-curve we'd end up in the same bracket anyway.

And that would probably be out West. The four regional hosts are East (Syracuse), Midwest (St Louis), South (Houston), and West (Salt Lake). My first guess is Duke and Syracuse being slotted out West, with Duke playing a first/second round game in Buffalo. Or perhaps in the East, in a bracket with someone like Kentucky (which the NCAA would love), playing the first weekend in Jacksonville. Of those two, I think I'd prefer being matched up with Kentucky.

With no games anywhere in ACC country this year, the bracket will be at least as interesting as our seed.

I'd much rather play the 4/5 winner than the 3/6 winner, especially with teams like Georgetown, WVA, Texas (if they finish strong), and MSU showing up as potential three seeds.

[Duvall]

I just want to say that this is exactly the sort of statistic that makes me distrust Pomeroy. I think most people watching last night's game can agree that it was far from a good game, we shot way too many 3's, and that we showed lots of weaknesses. And suddenly we jump over Kansas to #1 in the Pomeroy rankings?

Computer looks at results, not appearances. It may not have been a pretty win, but in the end Duke's offensive rebounding offset the poor shooting for a decent offensive performance, and Duke was credited with a strong defensive effort due to the fact that Virginia Tech didn't score much.

[Kedsy]

Aside from bragging rights, I'm not sure there's much of a difference between being the 4th #1 seed and the 1st #2 seed. By the S-curve we'd end up in the same bracket anyway.

This has been discussed extensively in another thread, but #1 seeds make the Elite Eight at a rate 50% higher than #2 seeds. They make the Final Four twice as much as #2 seeds, and they win the championship almost 4x as much as #2 seeds. I don't know how much of that has to do with #1 seeds actually being better than #2s, and how much has to do with the easier road, but I don't need to know to say there's a lot more advantage to being a #1 seed than "bragging rights."

[mike88]

I don't want to distract this thread from it's discussion which is the pursuit of a #1 seed - there are other threads for discussion of whether we deserve one. I just want to say that this is exactly the sort of statistic that makes me distrust Pomeroy. I think most people watching last night's game can agree that it was far from a good game, we shot way too many 3's, and that we showed lots of weaknesses. And suddenly we jump over Kansas to #1 in the Pomeroy rankings? Is there anyone who thinks we are actually the #1 team in the country? That doesn't make me have a lot of faith in the computer generate rankings.

IIRC, Va tech had the 6th best defense coming in to last night's game per Pomeroy- what you saw last night was the number 1 offense finding ways to overcome a very good defense- when we shoot poorly, it is not always because we have weaknesses, but also because the other teams have strengths. We have faced a lot of good defenses this year, and have had several games where the result looked a lot like last night (ugly % shooting, 12 point win)

[AIRFORCEDUKIE]

I think if we either win out the regular season and make it to the finals of the acc we get a one seed. If we drop one more game in the regular season and then win the Acc tourney we will also get a one seed. I think a one seed will be the key to getting back to the elite eight and then having a shot at the final four. It would do wonders for our confidence to have one easy game, against a 16 seed and then a potentially easy game against an 8/9 seed. If you look at the last few years we struggled in the early rounds and I think it hurt our confidence in the bigger games. If we can get our young guys some confidence in the early rounds it would help a lot. Easiest way to do that is to be a one seed. I see Syracuse and Purdue dropping two more games before seeding time, maybe even Kentucky.

Airforcedukie.

[JasonEvans]

Something that could be fun to track...

The Pomeroy ratings this morning-

1. Duke .9818
2. Kansas .9817
3. W Va .9643
4. Syrac .9638
5. Purdue .9620

I list these because it will be interesting to see how a couple games tonight affect things.

Kansas is playing Oklahoma, who is not highly regarded by KenPom being ranked #98. Kansas is at home and a prohibitive favorite in the game. How big a win will they need to get points and move ahead of Duke again?

WVa obviously just lost to UConn (the #43 team at .8912). How much will they lose from that loss?

With not a lot of other games on the national slate tonight, it will be interesting to see how these limited results affect things.

--Jason "tune in tomorrow for an update" Evans

So, here are today's numbers right next to yesterday's.

Today:

1. Duke .9818
2. Kansas .9799
3. Syracuse .9638
4. BYU .9618
5. Purdue .9617
6. W Va .9593

Yesterday:

1. Duke .9818
2. Kansas .9817
3. W Va .9643
4. Syrac .9638
5. Purdue .9620

So, Kansas' 13-point home victory over Oklahoma hurt them in the numbers to the tune of .0018 points. It maybe gives Duke a little bit of breathing room at the top. WVa's road loss to UConn did not hurt them all that much, reducing their Pom score by .0050 and knocking them from #3 to #6.

-Jason "well, that was an intersting little exercise" Evans

[KyDevilinIL]

I'd much rather play the 4/5 winner than the 3/6 winner, especially with teams like Georgetown, WVA, Texas (if they finish strong), and MSU showing up as potential three seeds.

I'd say all of the teams you mentioned are going to land anywhere from a No. 3 to No. 5 seed, so Duke being a No. 1 or No. 2 seed, in my mind, still doesn't make much difference.

This tournament has the potential to be an absolute mess from a seeding and bracket-advancement standpoint. There's an enormous blob of very-good-to-decent teams out there, and the probable Nos. 1 and 2 seeds simply aren't so far ahead that they couldn't start losing games immediately.

It could be a wacky tourney. Which probably means it'll be straight chalk.

[moonpie23]

i am going out on a limb here and predicting that we will have to play whatever team we have to play....I can't seem to get into the mass speculation of WHO that team might be...

aside from UK, KU, Tx, Orange, Nova, Kstate, Purdue, g'town, OSU, and a few others, i'd say it's safe to predict we will have our hands full at some point.


i love watching you guys break it down tho..

[CrazieDUMB]

I don't want to distract this thread from it's discussion which is the pursuit of a #1 seed - there are other threads for discussion of whether we deserve one. I just want to say that this is exactly the sort of statistic that makes me distrust Pomeroy. I think most people watching last night's game can agree that it was far from a good game, we shot way too many 3's, and that we showed lots of weaknesses. And suddenly we jump over Kansas to #1 in the Pomeroy rankings? Is there anyone who thinks we are actually the #1 team in the country? That doesn't make me have a lot of faith in the computer generate rankings.

Often times when looking at these things you have to remember that your (well, at least for me) view of Duke is going to be a lot different from your view of other teams. For the casual fan, it usually means you think your team is better than it really is. For the people who watch every game, read every article and post on message boards, it can often be the opposite. Because we know this team so well, we know their limitations. At the same time, we also know our strengths and potential; thus, if someone doesn't live up to the best game they've had this year it looks like a disappointment.

Right now a lot of people are worried that Miles and Mason have disappeared, that while Z is a monster on the boards of late he's still not scoring too much, etc etc. Do you bring the same kind of critical eye when you watch Kansas or Kentucky play? I doubt it, simply because when you don't know a team as well you tend to notice spectacular plays more and subtle ineffieciencies less.

The wonderful thing about these computer polls is that they strip the game of bias. Of course it's not perfect, and to a degree numbers will never really be able to tell the full story; even with a perfect model, there's too much variation game to game. But these are the best rankings science has to offer, which IMO is better than any scout or human, no matter how much basketball they watch.

[Memphis Devil]

IIRC, Va tech had the 6th best defense coming in to last night's game per Pomeroy- what you saw last night was the number 1 offense finding ways to overcome a very good defense- when we shoot poorly, it is not always because we have weaknesses, but also because the other teams have strengths. We have faced a lot of good defenses this year, and have had several games where the result looked a lot like last night (ugly % shooting, 12 point win)

Great point...all to often we (at least I know I do) watch Duke games and categorize it as either a good performance or a poor performance simply based on how the Duke players performed and completely ignoring the 12 guys wearing the other uniforms and the impact that they had on said perfromance.

[Olympic Fan]

The four regional hosts are East (Syracuse), Midwest (St Louis), South (Houston), and West (Salt Lake). My first guess is Duke and Syracuse being slotted out West, with Duke playing a first/second round game in Buffalo.

Not going to get into this (we have other threads discussing this), except to say that with the NCAA's new geographical placement for first-second round games, Duke will almost certainly open in Jacksonville and not Buffalo.

Jason, looking ahead on Pomeroy, it will be interesting to see if Kansas can catch Duke overthe course of the regular season. They finish at Oklahoma State (47), Kansas State (11), and at Missouri (13). Duke closes out with Tulsa (76), at Virginia (108), at Maryland (34) and UNC (92).

I think there's a good chance that if both win out, Kansas will edge ahead going into the conference tournaments. Of course, if one or the other loses (as could happen), the other would be in good shape.

[JasonEvans]

The wonderful thing about these computer polls is that they strip the game of bias. Of course it's not perfect, and to a degree numbers will never really be able to tell the full story; even with a perfect model, there's too much variation game to game. But these are the best rankings science has to offer, which IMO is better than any scout or human, no matter how much basketball they watch.

Well, let me correct you about one thing. I do not think the computers do a better job of evaluating teams better than humans. There are things beyond stats that go into evaluating a performance.

If a team has a big lead and goes into a coasting mode which allows the lead to dwindle, a stat-driven computer might see that as a close victory when the human observer saw a blowout that was not as close as the final score. A computer cannot differentiate from a basket made early in the game and one made under a pressure situation. There are a million ways our human ability to judge things like situations and momentum far outweigh a stat-only computer evaluation.

That said, computers have one huge, huge, huge advantage over humans... a computer can "watch" every single game in the country. No human can do that. When Ken Pomeroy or Jeff Sagarin plug in their data, the computers are weighing every single game played across the country. We are talking about more than ten thousand games (350 teams in Div 1 playing about 30 games each). The most passionate and dedicated college hoops fan watches maybe 100 complete games total. Even someone like Andy Katz, who gets paid handsomely to watch and analyze college hoops, probably watches no more than 300 games a year.

So, I would say that most of us who have watched every Duke game this season (and for many seasons) know Duke better than the computers do. Heck, many of us have probably seen enough games by some other ACC teams that we know those teams better than the computers. But, how many Kansas games have we watched? How many Syracuse games? Nova? Kentucky? Purdue? I'd be very surprised if any of us here have seen all of these teams play more than 3 or 4 times (I am talking about complete games, not catching the final 5 minutes of an interesting game). And, to take it a step further, how many of us really know much about the opponents Kansasy/Kentucky/Syracuse/Purdue were playing that day?

That Kentucky-Vandy game was great the other day, but it was the first time I've watched more than 3 minutes of Vandy basketball all year. Were they playing great or poorly or average? I can make a guess based on what I have heard and read about them but my objective analysis of the game is fundamentally lacking a ton of data.

Look, I am not saying the computers are to be 100% trusted over human observation. I agree that Duke probably is not the best team in the land, as KenPom suggests right now. But, when I am weighing how good teams are, I certainly include computer analysis along with my own observations and give it plenty of weight. I am betting that the selection committee will do the same thing.

--Jason "I am thrilled that the committee will have KenPom numbers with them in the room... I think it is very good for Duke!!" Evans

[JasonEvans]

Jason, looking ahead on Pomeroy, it will be interesting to see if Kansas can catch Duke overthe course of the regular season. They finish at Oklahoma State (47), Kansas State (11), and at Missouri (13). Duke closes out with Tulsa (76), at Virginia (108), at Maryland (34) and UNC (92).

I think there's a good chance that if both win out, Kansas will edge ahead going into the conference tournaments. Of course, if one or the other loses (as could happen), the other would be in good shape.

Of course, there is who you play but there is also how well you play against that team. KenPom is not like the RPI where you get massive credit just for playing a tough opponent. Still, you are right that Kansas will have more chances to "impress" the computer than Duke will down the stretch.

I really want Duke to finish #1 in the KenPom. I think this will carry weight in the selection committee room and can be the thing that pushes us to a #1 seed over Purdue.

-Jason "just win baby-- win out and I am confident we will get that top seed" Evans

[Chitowndevil]

I really want Duke to finish #1 in the KenPom. I think this will carry weight in the selection committee room and can be the thing that pushes us to a #1 seed over Purdue.

I dunno...

Ken Pomeroy's invaluable efficiency ratings are explicitly listed in the "resources" section of our committee materials, alongside Sagarain ratings and Jerry Palm's CollegeRPI.com. But they're listed last...

In other words, while the burgeoning tempo-free statistics movement has made major inroads in college basketball media, but it doesn't have much of a place in the committee room. Sad smiley face.

Link

[JasonEvans]

I dunno...

Link

How can the mock selection committee know what the real selection committee will use?

Frankly, anything in the room that lists Duke at #1 or can be used as ammunition to make Duke a #1 seed is fine with me!!

--Jason "I have a feeling that one Duke or Purdue will make the choice clear over the next couple weeks" Evans