PDA

View Full Version : Let's Face It - We are the Kings of November and December



Udaman
01-24-2013, 10:47 AM
I know losing Kelly hurt. He's a huge part of our team. We also lost him last year down the stretch, and we lost Irving the year before that. All of those injuries hurt us, and made us worse over the second half of the season. But we now have a 10 year trend of playing our best basketball in the first two months of the season. Here are the stats:

Nov and Dec since '03-04:

Overall Record: 108-6. That's 94.7%
Record Against Teams in Top 25: 21-3, or 87.5%
Record Against Unranked Teams: 87-3 or 96.7%

Those numbers are incredible. We dominate the tournaments and typically the Challenges or special games. And yeah we play some creampuffs, but we also play good teams (included in our record against non top 25 teams are this years wins against Minnesota and VCU, who weren't ranked at the time, but clearly are now).

So here's our record from Jan on (including the year where we won the championship)

Overall: 172-52, or 76.8%
Against Top 25: 36-24, or 60%
Unranked Teams: 136-28 or 82.9%

These are still good numbers....but not great, and nothing at all like we do in November and December. I'm not sure why this disparity exists. Maybe our Nov/Dec games all tend to be at home or on a neutral site. Maybe we are better coached at the beginning, then other teams start to catch up to their talent, and we stay the same. But the fact is we start extremely well, climb in the rankings, and then become a good but not elite team down the stretch (for the most part).

It's incredible to think that this year we pretty much dominated Louisville, Kentucky, Minnesota, Minnesota and VCU, all on neutral courts. And yes, Kelly makes us not as good as we were....but if we played any of those teams tomorrow on a neutral court, I think we lose - and if we played them on the road we would lose by double digits.

House G
01-24-2013, 11:05 AM
Losing one of the best players on the planet (and I submit the best player on our team in 2010-11) to injury may explain the appearance of "cheap early speed" that year.

roywhite
01-24-2013, 11:11 AM
The main thing I draw from this info is that Duke plays good basketball in November and December. Which is not a bad thing at all.

Just look at the more recent years:

2012-13 -- we saw what happened last night and know Kelly is out; who knows what the rest of the season will bring
2011-12 -- not a great team overall, and losing Kelly had a huge impact late season
2010--11 -- how can you make judgments when you lose Kyrie Irving, the NBA ROY of the year, for goodness sakes, early on?
2009--10 --- the team stayed healthy and as I recall, the season turned out pretty well

Mike Corey
01-24-2013, 11:14 AM
I don't think the past few seasons are fairly included without an asterisk, given the precarious injury situations of Kyrie and Kelly.

However, there's something to this.

During the stretch early in Coach K's Duke tenure when we advanced to seven FFs in nine years, we peaked late despite not always having the most sterling of regular season records.

That does seem to have flipped a bit post-2004, whatever the reasons. Except for 2010, of course.

OldPhiKap
01-24-2013, 11:20 AM
Hard to believe how many times we have been a #1 or #2 seed, given that we are only a "good but not elite" team after Christmas.

Surely, our ACC record must bear that out as well. Have we even broke .500 in conference play in the last ten years? Surely, we have not won the regular season race during any of that period, or pulled out an ACC Tournament Championship.

This is why the other schools keep piling up the NC's while we wallow in faded glory of the Turkey-to-Holly seasonality of our program.

When do pitchers and catchers report?

CDu
01-24-2013, 11:23 AM
I don't think the past few seasons are fairly included without an asterisk, given the precarious injury situations of Kyrie and Kelly.

However, there's something to this.

During the stretch early in Coach K's Duke tenure when we advanced to seven FFs in nine years, we peaked late despite not always having the most sterling of regular season records.

That does seem to have flipped a bit post-2004, whatever the reasons. Except for 2010, of course.

I'd say it even applied in 2010. We peaked once in Nov/Dec 2009, then "slumped" in Jan 2010 and into even February 2010, then peaked again in March and April.

I think both of Udaman's explanations are in play. We are better prepared for games in November and December than our opponents, as Coach K figures out his plan for the team earlier in the season than most coaches and is better at motivating his teams for early-season games. As the season wears on, our opponents start to take shape and get more motivated (especially in March/April). Additionally, we do play a disproportionate amount of our games at home early (like most good teams do), which inflates our record a bit relative to January and beyond.

In some years, injuries have been a factor as well. We've generally been healthy in November/December and run into injuries in December/January or later. But even given that, we certainly seem to consistently be a better team (relative to the field) in November/December than we are in January-March.

CDu
01-24-2013, 11:25 AM
Hard to believe how many times we have been a #1 or #2 seed, given that we are only a "good but not elite" team after Christmas.

Surely, our ACC record must bear that out as well. Have we even broke .500 in conference play in the last ten years? Surely, we have not won the regular season race during any of that period, or pulled out an ACC Tournament Championship.

This is why the other schools keep piling up the NC's while we wallow in faded glory of the Turkey-to-Holly seasonality of our program.

When do pitchers and catchers report?

The original post is far from a "sky is falling" post, and didn't merit the sarcastic response. We have generally been world beaters in November/December, and then just very good (and sometimes great) in the new year. It's not something at which to just thumb your nose.

roywhite
01-24-2013, 11:30 AM
I'd say it even applied in 2010. We peaked once in Nov/Dec 2009, then "slumped" in Jan 2010 and into even February 2010, then peaked again in March and April.
I think both of Udaman's explanations are in play. We are better prepared for games in November and December than our opponents, as Coach K figures out his plan for the team earlier in the season than most coaches and is better at motivating his teams for early-season games. As the season wears on, our opponents start to take shape and get more motivated (especially in March/April). Additionally, we do play a disproportionate amount of our games at home early (like most good teams do), which inflates our record a bit relative to January and beyond.

In some years, injuries have been a factor as well. We've generally been healthy in November/December and run into injuries in December/January or later. But even given that, we certainly seem to consistently be a better team (relative to the field) in November/December than we are in January-March.

Well, the 2010 part of this is simply false.

After losing to Georgetown in late January, the team went 18-1, including 8-0 in February.

rsvman
01-24-2013, 11:30 AM
Our teams tend to be really excellent right out of the gate. A testament to good coaching and preparedness? Mental toughness? Focus? I think all of the above.

I don't necessarily think that what happens is that we get worse. Rather, I think that other teams begin to gel as the season wears on, and they get better and better. Our team, very good right out of the gate, plateaus a bit. Thus, other good/elite teams have a better chance of beating us later than they do earlier in the season.


Oh, and I really think conference play, particularly on the road, is more difficult in some ways than even those talent-laden pre-season tournaments.

CDu
01-24-2013, 11:34 AM
Well, the 2010 part of this is simply false.

After losing to Georgetown in late January, the team went 18-1, including 8-0 in February.

So we slumped only in January (losing 3 of our 9 games), but not February. So I was only sort of wrong. I had thought our loss to Maryland was in early February. I had forgotten it was actually our next-to-last regular season game.

Duvall
01-24-2013, 11:40 AM
So we slumped only in January (losing 3 of our 9 games), but not February. So I was only sort of wrong. I had thought our loss to Maryland was in early February. I had forgotten it was actually our next-to-last regular season game.

Not sure that would be evidence of a slump either way. As I recall, Duke didn't play badly in that game, it was just a close road loss to a top-15ish team.

ChillinDuke
01-24-2013, 11:43 AM
The original post is far from a "sky is falling" post, and didn't merit the sarcastic response. We have generally been world beaters in November/December, and then just very good (and sometimes great) in the new year. It's not something at which to just thumb your nose.

I dunno. I'm sorta with OPK on this one.

I think the OP's view is much to do about nothing. At least in the last 4 years (as referenced by the OP), did we ever really enter the season thinking we were a world-beating (my word) team? With the exception of the Kyrie year, I think not.

So, couldn't one just as reasonably have the view that we outperform expectations in Nov/Dec and then perform in Jan-Mar? This is an important difference as viewing it as the OP does lends itself to changing the preseason expectation to that of perform in Nov/Dec and then underperform in Jan-Mar. Sort of not fair to do that.

To be clear, the preseason consensus on this board was we were probably a Top-10 or Top-15 team this year. In that ballpark. By that metric we clearly outperformed in Nov/Dec to the tune of 4 weeks or so at #1 to this point in the year. Now, we lose two (albeit, one of them badly) without RK, and suddenly we are underperforming.

We can agree to disagree - but I don't think it's fair to disregard the preseason consensus (I trust many on this board more than ESPN) in all of this. So yeah - if you think Nov/Dec was the "new normal", then maybe we have a "peaking" issue. I don't know if that's fair, though.

- Chillin

OldPhiKap
01-24-2013, 11:45 AM
The original post is far from a "sky is falling" post, and didn't merit the sarcastic response. We have generally been world beaters in November/December, and then just very good (and sometimes great) in the new year. It's not something at which to just thumb your nose.

We typically play a more difficult schedule after the new year, with real road games and conference foes that know you well. I thus don't think the overall winning percentages mean a whole lot. If we really slumped as suggested, we would not win the conference title or do as well in the NCAA's as we do. It is not that we play worse -- it is that we are playing more even competition and more true road games.

We win preseason tournament games. We win ACC tournament games. We win NCAA tournament games. The biggest difference is that the NCAA requires six wins, the others less.

Did we play better in November than right now? Sure. Does it have more to do with some systematic problem, or losing Ryan Kelly? I think that latter.

Sorry if I am surly this morning, I usually do better after lunch. ;>)

CDu
01-24-2013, 11:45 AM
Not sure that would be evidence of a slump either way. As I recall, Duke didn't play badly in that game, it was just a close road loss to a top-15ish team.

Sure. My point still stands. We peaked once in Nov/Dec, slumped in (only) January (going 6-3), then peaked again in Feb-April.

My main point about 2010 was that we still had a lull period in January. One of the big differences between that season and others is that we were able to find a new spark with Zoubek figuring out how to stay out of foul trouble and be a constant presence on the floor.

Wander
01-24-2013, 11:49 AM
I think both of Udaman's explanations are in play. We are better prepared for games in November and December than our opponents, as Coach K figures out his plan for the team earlier in the season than most coaches and is better at motivating his teams for early-season games. As the season wears on, our opponents start to take shape and get more motivated (especially in March/April). Additionally, we do play a disproportionate amount of our games at home early (like most good teams do), which inflates our record a bit relative to January and beyond.


Agreed with all of this. I'll just propose one other possible factor: I think we tend to have more senior star players than other programs of similar caliber, which might lend itself to fast starts but lower room for improvement in the late season.

CDu
01-24-2013, 11:50 AM
We typically play a more difficult schedule after the new year, with real road games and conference foes that know you well. I thus don't think the overall winning percentages mean a whole lot. If we really slumped as suggested, we would not win the conference title or do as well in the NCAA's as we do. It is not that we play worse -- it is that we are playing more even competition and more true road games.

We win preseason tournament games. We win ACC tournament games. We win NCAA tournament games. The biggest difference is that the NCAA requires six wins, the others less.

Did we play better in November than right now? Sure. Does it have more to do with some systematic problem, or losing Ryan Kelly? I think that latter.

Sorry if I am surly this morning, I usually do better after lunch. ;>)

Except that we also have a worse record against ranked teams. Yes, some of it is simply playing more road games. Part of it is playing against more difficult opponents. But that alone shouldn't account for a nearly 30% difference in winning percentage against top-25 teams.

Do I think it's a "systemic problem?" No. As you note, we've won multiple ACC championships and even won a national championship and made another Final Four in the time period (last 10 years). But I do think it's a trend worthy of discussion, moreso than you seem to be willing to admit.

CDu
01-24-2013, 11:54 AM
We can agree to disagree - but I don't think it's fair to disregard the preseason consensus (I trust many on this board more than ESPN) in all of this. So yeah - if you think Nov/Dec was the "new normal", then maybe we have a "peaking" issue. I don't know if that's fair, though.

- Chillin

I'm not disregarding the pre-season consensus, nor am I saying there is a problem. We often exceed expectations in the early part of the season, and we often return to the norm as the season progresses. Often, that norm is still good enough to win the ACC. Sometimes, we even exceed the norm (like in 2010). Sometimes, that norm is a good-but-not-great team that ultimately gets outplayed by a hotter team in March.

I don't think anyone here is suggesting there is a problem. We're just noting a trend in performance.

FerryFor50
01-24-2013, 11:55 AM
It's the sign of a well-coached, under-manned team.

Naturally, due to coaching and preparation, Duke comes out of the gate more ready to play and more cohesive. With no game film on the current crop of players, there isn't as many spots to break down a team. Plus, you never know how a team plans on playing year to year.

As the season wears on, teams get more chemistry and the talent starts to take over. In-game coaching isn't as important as game film and breaking down a team's weaknesses. Injuries pile up.

Coach K has always gotten more out of less than any coach in the country. That's why you see the trends you see.

nocilla
01-24-2013, 11:56 AM
I think a lot of it is simply competition. I realize that recently some ACC teams have been down right putrid but that is not usually the case. Normally we play a lot of creampuffs in November and December, mostly at home, with a few big games sprinkled in. Then in January we start ACC play and all the games are fairly competitive, with half of them being on the road. The stats used point out that we played only 24 top 25 teams compared to 60 after January. The win percentage goes down from 87.5% to 60% mostly because the early top 25 games are mostly on neutral courts whereas a lot of the later games are road games. Plus a lot of the early big games are based on preseason rankings which are sometimes way off.

Udaman
01-24-2013, 12:00 PM
I'm not disregarding the pre-season consensus, nor am I saying there is a problem. We often exceed expectations in the early part of the season, and we often return to the norm as the season progresses. Often, that norm is still good enough to win the ACC. Sometimes, we even exceed the norm (like in 2010). Sometimes, that norm is a good-but-not-great team that ultimately gets outplayed by a hotter team in March.

I don't think anyone here is suggesting there is a problem. We're just noting a trend in performance.

Agree with CDu here. I'm not saying that we haven't been good down the stretch. We've clearly been a top 20 team most years the second half of the season. But for the first half of the season we have been CLEARLY the best team in the country over that 10 year spread. It's not even close. We simply dominate. And I think it is because we have a great coach and we come out prepared. But, it's also frustrating. Like I said - it's crazy that 6 weeks ago we controlled 3 top 10 teams (and really 5 top 10 teams if you include Minnesota and VCU), and I don't think we could beat any of them on a neutral court today. Could we with Kelly? Yes, I think we could. We miss him. That's for sure. But we have become (for the most part) a peak early team the last decade.

devil84
01-24-2013, 12:02 PM
The games played through December are usually teams and players we don't usually face on a regular basis. Once January hits, we play almost exclusively conference games, where we face opponents we may have played as many as three times last year (and perhaps even once prior in the current season). Seniors could may have played the opponent's seniors 6-8 times or more in their career. They've been to the opponent's venue multiple times, know it, and have a history there. Scouting videos are likely to show Duke against a common, usually conference opponent with which they've had plenty of experience. Even summer pick up games are likely to have more ACC opponents on them than players from other teams, leading to even more familiarity.

ACC teams circle Duke (and UNC) on their calendars every year. When Duke's away, those are the games that sell out their stadiums and bring out the fans. Non-conference games do, too, but the toughest opponents generally are on neutral courts in sold-out venues with multiple fan bases in attendance (still anti-Duke, but not like playing on a home court in front of your own fans). But to have your own home venue transformed from a ho-hum atmosphere to THE GAME makes a huge difference in a team's confidence level.

My point is, that there is a lot of familiarity in conference play that can make it easier to beat (or lose to) an opponent that isn't there with the freshness of non-conference play.

It's probably why, if you look at the records for most other elite teams, that you might just see a similar pattern of winning percentages.

ChillinDuke
01-24-2013, 12:12 PM
Agree with CDu here. I'm not saying that we haven't been good down the stretch. We've clearly been a top 20 team most years the second half of the season. But for the first half of the season we have been CLEARLY the best team in the country over that 10 year spread. It's not even close. We simply dominate. And I think it is because we have a great coach and we come out prepared. But, it's also frustrating. Like I said - it's crazy that 6 weeks ago we controlled 3 top 10 teams (and really 5 top 10 teams if you include Minnesota and VCU), and I don't think we could beat any of them on a neutral court today. Could we with Kelly? Yes, I think we could. We miss him. That's for sure. But we have become (for the most part) a peak early team the last decade.

So we're just noting a trend. Fair enough. But I still think the logic is off - at least for this year, as I haven't gone back and considered the previous 10 years.

Kentucky is most clearly not a top team. VCU is good - I dunno about Top 10. The rest are inarguably very good teams (I'll avoid calling them "Top 10" so as to avoid nitpicking).

But Miami and NC State are also very talented teams. They aren't inarguably very good teams - it's quite arguable. But they are very talented - just with a lot of variance.

Take away Ryan from us and put either of Miami or NC State at home and playing up to their talent level and it's completely reasonable to see us losing. Perhaps not as bad as we lost to Miami, but it's completely reasonable. Not enough for me to consider this anything other than normal goings-on in ACC play. (And I fully admit the loss to Miami was pretty bad looking.)

- Chillin

Chicago 1995
01-24-2013, 12:13 PM
It's the sign of a well-coached, under-manned team.

Naturally, due to coaching and preparation, Duke comes out of the gate more ready to play and more cohesive. With no game film on the current crop of players, there isn't as many spots to break down a team. Plus, you never know how a team plans on playing year to year.

As the season wears on, teams get more chemistry and the talent starts to take over. In-game coaching isn't as important as game film and breaking down a team's weaknesses. Injuries pile up.

Coach K has always gotten more out of less than any coach in the country. That's why you see the trends you see.

How do you fit Duke's very different growth pattern between 86-94 in this theory? Michigan State under Tom Izzo?

FerryFor50
01-24-2013, 12:19 PM
How do you fit Duke's very different growth pattern between 86-94 in this theory? Michigan State under Tom Izzo?

Those 86-94 teams had more talent, IMO.

You didn't have the whole "one and done" culture, either.

Tom Izzo? He's consistent, just like Coach K. But he hasn't won a title since 2000.

nmduke2001
01-24-2013, 12:30 PM
I think one factor may be that Coach K is usually earlier than most in setting his rotation. So he is playing with his core 8 or so from the beginning of the season while some other coaches are getting ALL of their kids some playing time early. Heck, Alex hardly got any minutes even in blow-outs and now we want he and Amile to be big contributors because of injury. Having the rotation set early in the season probably sets up winnning early. Just my opinion.

Chicago 1995
01-24-2013, 12:51 PM
Those 86-94 teams had more talent, IMO.

You didn't have the whole "one and done" culture, either.

Tom Izzo? He's consistent, just like Coach K. But he hasn't won a title since 2000.

Whether the 86-84 teams had more talent or not, I don't know that they were more talented compared to the competition at the same time, which is what would matter, and I don't know how the one-and-done problem would effect the growth of a single team over the course of a season. We're integrating as many new pieces -- maybe more because we lose guys through one and done -- so in theory, we should have more room for growth over the course of a season, not less. Highly rated frosh who get time like Rivers should give us more ceiling, not less.

That's moving the goalposts a bit as to Izzo. The metric we're talking about here isn't winning titles, but about improving from Nov/Dec to March, which Izzo has done, much as I dislike his style of basketball, very consistently and I think, without much question.

This is summoning a firestorm, but I think there's a real argument that K is less willing to sacrifice wins for growth than he used to be, and that instead of developing youth and depth -- which is where a lot of growth comes from -- K is maximizing the odds of winning each individual game.

rsvman
01-24-2013, 01:00 PM
....This is summoning a firestorm, but I think there's a real argument that K is less willing to sacrifice wins for growth than he used to be, and that instead of developing youth and depth -- which is where a lot of growth comes from -- K is maximizing the odds of winning each individual game.

This is an interesting thought. It seemed to me in years past that Coach K cared a lot less about any individual game than maybe he does now. It almost seemed like sometimes he left the players on their own to lose the game in order to make an important teaching point that would help them later on (although it's not at all clear that he actually did that; this is only an impression I got watching some games in the past). That used to bug me a lot, because I kind of thought "you play to win the game." But now maybe I can see that there may be times when it is wiser to look at the long goals rather than the short ones.

And I agree completely that Izzo's teams are always better in February and March than they are in November and December.

rthomas
01-24-2013, 01:04 PM
The only real trend that I see so far for Duke:

Neutral court: no losses
Home court: no losses

True away games: no wins

Kedsy
01-24-2013, 01:26 PM
Sure. My point still stands. We peaked once in Nov/Dec, slumped in (only) January (going 6-3), then peaked again in Feb-April.

My main point about 2010 was that we still had a lull period in January. One of the big differences between that season and others is that we were able to find a new spark with Zoubek figuring out how to stay out of foul trouble and be a constant presence on the floor.

Every team has a "lull period." It's a long season. By your definition even the 1992 team "lulled" in February, going 6-2, while "peaking" in every other month.

davekay1971
01-24-2013, 01:30 PM
I had no idea all those ACC and NCAA tournaments were played in November and December. Wierd. And I thought I knew something about college basketball!

I was wondering, just crazy spitballing here, if maybe our recent 1-2 streak had more to do with losing a critical senior All-ACC caliber player right as we play two top 25 teams, both of whom played their absolute best games of the season against us, on their home courts.

mkirsh
01-24-2013, 01:32 PM
It's obvious to everyone here why we play better pre December vs post December every year. It's a combination of:

1) Coach K refusing to use early season game time to develop players - this means that our best players play more minutes early than other teams who spread minutes to freshman, and by the end of the year their freshman have made them better while ours don't add anything to the table since they haven't developed
2) Similarly, Coach K's insistence on playing his best players 35+ minutes per game catches up to them and they are exhausted and worn out by Feb
3) We don't play any true road games in the early season, and we don't know how to handle it when we go to someone else's gym
4) We wear the blue uniforms more in conference season, and everyone knows the players prefer black and play better in them
5) We always have someone break their foot in Jan because we wear Nike shoes

This is all in jest, of course - did I miss any other common grievances?

I'm in the camp of overperform early and then just perform (not underperform) late, which can be maddening if you allow your expectations to get pulled up by early season wins. Also, conference road wins are always hard to come by, games are more frequent which increases the chance of an off night (favoring the underdogs), and fewer breaks in the schedule.

Also, I'm too lazy to do the analysis, but I wonder if we are hyper sensitive to his (like with broken feet) because we watch Duke carefully - does this pattern hold true for other schools as well? Anecdotally Mich State seems to be the opposite - lose most of their early games but are formidable come tourney time, but do Kansas, Syracuse, UConn, other traditional powers follow similar patterns?

Wander
01-24-2013, 01:36 PM
I had no idea all those ACC and NCAA tournaments were played in November and December. Wierd. And I thought I knew something about college basketball!

Our NCAA tournament performance has underperformed the expectations based on seeding in 2005, 2006, 2007, 2008, maybe 2009 if you include margin of victory, 2011, and 2012. I don't understand how you can possibly think bringing that up is a point in your favor.

Unlike some, I don't think there's anything wrong with our program and am quite happy with where we are, including this season after last night's game, but I don't see how anyone can deny we've been better in November/December (relative to other teams) than in March the majority of the past decade.

Thank god for Scheyer, Smith, Singler, Thomas, and Zoubek. Can you imagine what the perception of us would be without 2010?

CDu
01-24-2013, 01:40 PM
I had no idea all those ACC and NCAA tournaments were played in November and December. Wierd. And I thought I knew something about college basketball!

I was wondering, just crazy spitballing here, if maybe our recent 1-2 streak had more to do with losing a critical senior All-ACC caliber player right as we play two top 25 teams, both of whom played their absolute best games of the season against us, on their home courts.

Way to seemingly not read the original post and make a response based on only the title. Nothing in your post is inconsistent with the content of the original post.

Not that it matters (see my previous two sentences), but I wouldn't use our NCAA tournament results of the last decade as evidence that we did really well in those games. We only came close to our seed's expectations in 3 of those 10 tournaments after all.

CDu
01-24-2013, 01:48 PM
Every team has a "lull period." It's a long season. By your definition even the 1992 team "lulled" in February, going 6-2, while "peaking" in every other month.

Those are hardly comparable. We lost two games to inferior opponents (one in somewhat of a blowout) and then got blown out by another really good team. In 1992, we lost a 2-point game at UNC (a top-10 team) and a 4-point game at Wake Forest (definitely an upset). A close loss against a REALLY good team and a close loss against a solid-but-not-great team is better than a close loss against a sort of good team, a blowout loss against a pretty bad team, and a blowout loss against a really good team.

Matches
01-24-2013, 01:52 PM
Eh. Every team is different. I think you can, if you are so inclined, look at the stats and see a trend of a mid-to-late season decline, but I'm not sure a particular cause can be attributed to it.

2008 team wore down physically at season's end.
2005 & 2011-12 teams had key injuries.
2012 team was ahead of everyone out of the gate because of the China trip.
2013's not over - too soon to draw conclusions.
2007 team just wasn't very good, regressed defensively.

2009-2010 teams played IMO their best ball in February and March.

We've had a spectacular record in November (December means less to me as we usually play fewer games and have more cupcakes during exams). I'm not really willing to turn that into a negative by expecting the team to play at an unconscious clip all season long.

azzefkram
01-24-2013, 01:57 PM
It's really tough to draw any conclusions without seeing how other programs perform. Maybe this is common for many programs. I remember feeling shocked at how many players were transferring out of Duke until I looked around and saw that it happens to almost all the big programs. If I had to guess, this happens just about everywhere. Well maybe not the utter dominance of November and December.

Monmouth77
01-24-2013, 02:01 PM
Whatever the reason for our early season success, I am less alarmed by the team's current struggles (which I think have a lot to do with Miami and NC State's talent, their hostile arenas, and our adjustment to playing without Kelly, which includes substituting freshmen for Kelly) than I am heartened by the contribution of those early wins to our bulletproof RPI.

Even with a few more stumbles (@ UNC? @MD? Miami/State rematches), Duke is going to have an on paper resume worthy of a #1 or #2 seed. We could end up like WVU from 2010 with 6 losses and still be in the conversation for a #1 seed.*

So I look at it as equity.

Obviously, I want to see the team come together and improve. But we have some breathing room to take our lumps, play through some trial and error, and still have a commanding position come Tournament time.

*I acknowledge that the Tournament Selection folks weighs player injuries, so if Kelly never comes back we could lose a seed line or two. I am explicitly assuming that the spirits will not be so unkind as to exclude such a worthy guy from finishing his senior year.

OldPhiKap
01-24-2013, 02:02 PM
Except that we also have a worse record against ranked teams. Yes, some of it is simply playing more road games. Part of it is playing against more difficult opponents. But that alone shouldn't account for a nearly 30% difference in winning percentage against top-25 teams.

Do I think it's a "systemic problem?" No. As you note, we've won multiple ACC championships and even won a national championship and made another Final Four in the time period (last 10 years). But I do think it's a trend worthy of discussion, moreso than you seem to be willing to admit.

I am happy to discuss it. If I take a thread entitled "King of November/December" as antagonistic/provoking, and you do not, I hope we will just agree to disagree on it.

To be clear, I am not picking at you. I respect your posts and often agree with you. This just obviously strike us differently.

To the main point: different teams peak at different times. Some, like 2010, peak at just the right time. Others peak earlier. Some never quite get there at all. I would simply say that, historically, our season gets more difficult once conference play starts. Part of the reason our % falls is because we have to play @ Carolina, @ Md, etc. I would also suggest that a conference team ranked 10-25 is tougher to beat than one you do not have to play every year (especially when it is in a neutral site, as the pre-Christmas games are).

I would suggest the more relevant inquiry is our % at home, at neutral courts, and in true away games. The last category is significantly below the other two. And we don't play those until after Christmas.

Bluealum
01-24-2013, 02:04 PM
I think there's a real argument that K is less willing to sacrifice wins for growth than he used to be, and that instead of developing youth and depth -- which is where a lot of growth comes from -- K is maximizing the odds of winning each individual game.

We approach this questions from so many different angles so many times.... is it possible that there is some truth to this? Playing Josh and Tyler as our primary backups early in the season because they know the system better and likely offer better outcomes in early games, actually just might have some downstream effects. While I am not in the camp that players that don't play in games don't improve, I do think that any game involves both physical and psychological components, of which 'confidence' is the most oft used descriptor for the latter. Confidence, I submit, is disproportionately earned in actual games, while physical skills and system understanding can develop exclusively in practice.

If we played folks like Amile and Alex more and Josh and Tyler less early, perhaps our record would be a little less overwhelming in Nov/Dec, but our improvement over the course of the season with a combination of lesser expectations early, and potentially greater performance late would 'feel' remarkably greater.

In K's early years he made a repeated point of emphasizing that the only real metric was how his teams performed in March, everything else was a leadup to that. He does that noticeably less now, methinks...

STILL, we have great overall outcomes, and yes I am sooooo glad we have 2010. Without that....the cross talk would be too much to handle.

The good news is that a close loss would not have warranted a shakeup or a major rethink. A loss like this just might, and when K really looks for alternate solutions he usually finds one.

Kedsy
01-24-2013, 02:11 PM
Nov and Dec since '03-04:

Overall Record: 108-6. That's 94.7%
Record Against Teams in Top 25: 21-3, or 87.5%
Record Against Unranked Teams: 87-3 or 96.7%

So here's our record from Jan on (including the year where we won the championship)

Overall: 172-52, or 76.8%
Against Top 25: 36-24, or 60%
Unranked Teams: 136-28 or 82.9%



OK, the thing is, we play very few true road games in November and December. So here is our record in home and neutral games after January 1 for the period you've used:

Overall Record: 116-24. That's 82.8%.

However, of the 24 losses, 8 came in the NCAA tournament, 4 came in the ACC tournament, and 5 came to UNC. I don't think any of those losses could fairly be counted in part of a early season/late season trend.

Moreover, 11 of the 24 losses came in two seasons (2007 and 2012). Not counting those two seasons, our overall record at home and neutral courts during the period is 103-13, or 88.8%. Considering we do play a decent number of patsies in November and December, I'd say there's little to no difference there, meaning the "trend" is really just an acknowledgment of two less-than-stellar seasons.

Against top 25 teams: 27-13. That's 67.5%.

Again, 10 of the 13 losses either came in the NCAAT or to our arch-rival UNC and another came to FSU last year in the ACCT, meaning we only lost two (2) games at home/neutral during the period in question that weren't either to UNC or in a post-season tournament.

Against unranked teams: 89-11. That's 89.0%.

In this case, 6 of the 11 losses came in either the NCAA tournament or the ACC tournament. And once again, 2007 and 2012 contained the lion's share of the losses (7 of the 11). Not counting those two seasons, our record against unranked teams at home or neutral courts was 78-4 (95.1%), with only one (1) such loss coming before the post-season.

My conclusion is the "trend" you've discovered really boils down to the fact that we play true road games, games against our arch-rival, and post-season games after January 1 and that we had two below par seasons during the period you've chosen. I don't think it's much of a trend at all.

CDu
01-24-2013, 02:11 PM
I am happy to discuss it. If I take a thread entitled "King of November/December" as antagonistic, and you do not, I hope we will just agree to disagree on it.

To be clear, I am not picking at you. I respect your posts and often agree with you. This just obviously strike us differently.

It was a poorly-worded title by the original poster, I agree. But I don't think the content of the post itself was worthy of the snarky response.


To the main point: different teams peak at different times. Some, like 2010, peak at just the right time. Others peak earlier. Some never quite get there at all. I would simply say that, historically, our season gets more difficult once conference play starts. Part of the reason our % falls is because we have to play @ Carolina, @ Md, etc. I would also suggest that a conference team ranked 10-25 is tougher to beat than one you do not have to play every year (especially when it is in a neutral site, as the pre-Christmas games are).

That's a perfectly fair opinion/theory. Familiarity does factor in here.


I would suggest the more relevant inquiry is our % at home, at neutral courts, and in true away games. The last category is significantly below the other two. And we don't play those until after Christmas.

I also agree that a good bit of the difference is accounted for in the true road games. I'd say that there are multiple things playing in here to varying degree:

- Coach K gets his teams ready sooner than most coaches, so we're better focused/ready than others in November
- We typically (though not always) have had less attrition to the NBA than most, so there is more continuity of team than our opponents early in the season
- We play no true road games and no conference foes in November/December, and familiarity tends to lead to tougher contests
- We have had significant in-season injuries in 2 of the 10 years that have greatly affected our ACC schedule results

vick
01-24-2013, 02:15 PM
It's really tough to draw any conclusions without seeing how other programs perform. Maybe this is common for many programs. I remember feeling shocked at how many players were transferring out of Duke until I looked around and saw that it happens to almost all the big programs. If I had to guess, this happens just about everywhere. Well maybe not the utter dominance of November and December.

There is something to this. The point above about how Duke has "underperformed the expectations based on seeding in 2005, 2006, 2007, 2008, maybe 2009 if you include margin of victory, 2011, and 2012" is an objectively horrible metric, because all top seeds have "underperformed based on seeding"--this literally has to be the case unless there are never any upsets. What you want to do is compare Duke's performance as X seed vs. the NCAA average. Here is a table (http://basketball.realgm.com/article/212166) I found of wins per seed for the first 26 years of the 64-team bracket (so missing the last couple of years but close enough):



1
3.42


2
2.43


3
1.84


4
1.46


5
1.18


6
1.22


7
0.83


8
0.65


9
0.59


10
0.65


11
0.49


12
0.52


13
0.25


14
0.17


15
0.04


16
0.00



For the 1995-present period, Duke has "expected" wins of 2.8 on average, and has actually won 2.4, if my quick calculations are right. So a little bit below seeding but not catastrophic. For 2004-present, the same expectation of 2.8 wins, but only 2.1 in actuality. So, there is an underperformance, but not as much as often made out to be. You can't mechanically look at a long string of 1 and 2 seeds and think we should be in the Final Four, or at least the Elite 8, every year.

Wander
01-24-2013, 02:18 PM
There is something to this. The point above about how Duke has "underperformed the expectations based on seeding in 2005, 2006, 2007, 2008, maybe 2009 if you include margin of victory, 2011, and 2012" is an objectively horrible metric, because all top seeds have "underperformed based on seeding"--this literally has to be the case unless there are never any upsets. What you want to do is compare Duke's performance as X seed vs. the NCAA average.

Actually, I already took that factor into account in my quote that you cite here. It's still the case that Duke has underperformed expectations based on seeding in 2005, 2006, 2007, 2008, maybe 2009, 2011, and 2012.

vick
01-24-2013, 02:22 PM
Actually, I already took that factor into account in my quote that you cite here. It's still the case that Duke has underperformed expectations based on seeding in 2005, 2006, 2007, 2008, maybe 2009, 2011, and 2012.

You know you're right--I focused on the part about scoring margin and misunderstood. So while I don't think it's as bad as people say, I sold short what you were saying, my apologies.

CDu
01-24-2013, 02:23 PM
For the 1995-present period, Duke has "expected" wins of 2.8 on average, and has actually won 2.4, if my quick calculations are right. So a little bit below seeding but not catastrophic. For 2004-present, the same expectation of 2.8 wins, but only 2.1 in actuality. So, there is an underperformance, but not as much as often made out to be.

I'd argue that a 0.7 win underperformance over a 10-year span is pretty substantial. That's a 25% underperformance.


You can't mechanically look at a long string of 1 and 2 seeds and think we should be in the Final Four, or at least the Elite 8, every year.

Actually, the results suggest you SHOULD expect a #1 seed to make the Elite-8. An average win total of 3.42 wins would put them in the elite-8, and winning ~42% of those elite-8 games (assuming no Final Four wins, of course). For a #2 seed, the expectation is one game less than that, or Sweet-16, with a ~43% win probability in the Sweet-16 (again, assuming no further wins). Duke's 2.8-win expectation suggests that we should be consistently making the Sweet-16 and more often than not making the Elite-8. To be 0.7 wins short of that average suggests that, in 2-3 of every 3-4 years, we're coming up a game shorter than expectations. And when you consider that the 2004 and 2010 teams exceeded expectations, that means that we've fallen short in 8 of the remaining 10 seasons over that span.

Yes, it's harder for a #1 or #2 seed to meet expectations than it is for a #15 seed or even a #8 seed. But that doesn't mean that we haven't played below expectations in the tournament over the last decade.

Wander
01-24-2013, 02:29 PM
You know you're right--I focused on the part about scoring margin and misunderstood. So while I don't think it's as bad as people say, I sold short what you were saying, my apologies.

No worries. Thanks for the table - I had figured out the expected wins for a 1 seed a while back, but never the numbers for anything below that, so it's cool to see.

devil84
01-24-2013, 02:34 PM
I decided to take a look at the last several years of stats for various teams. I took the last 5 years of NCAA champions: Kentucky, Connecticut, Duke, Carolina, and Kansas, and then added Michigan State and Butler, as they have been in the Final Four multiple times in that time frame. That looked like a fairly decent selection of recently elite teams.

I looked at this year (with incomplete data), and the previous 4 years. Why 4? Because I only have so much time to spend on this project. :)

I compiled the win percentage for November and December, for January through the end of the season (including the conference tournament, since that wasn't broken out separately), and for January through the end of the tournament. I've shown them in order by January through the end of the season (the second column).

Here are the results:
Nov-Dec Jan-Mar Jan-Apr
Kansas 88.5 87.2 84.4
Butler 72.3 79.8 79.0
Duke 95.0 78.5 78.3
Kentucky 83.9 75.3 76.7
Michigan State 80.3 71.4 71.7
Carolina 81.5 69.5 72.3
Connecticut 86.7 56.3 58.8
Our Nov-December stat is wonderfully gaudy. From January through the end of the year, we're not chopped liver, though. We're still doing quite well. It just SEEMS bad, because there's such a huge drop-off from 95% to 78%, a drop of over 16 percentage points! Butler is the only one that does better during conference play. Kansas is pretty consistent from non-con to conference play. Kentucky and Michigan State drop about 8 percentage points, Carolina drops 12, and Connecticut over 30!

Winning "only" 78.5% of our games from January on doesn't appear to be horrible. I'm pleasantly surprised with these stats.

MIKESJ73
01-24-2013, 02:36 PM
Every year when conference play starts almost every team has great records in major conferences. Duke usually plays a couple of good teams and has a tourney with top competition, but most of the teams in Nov. and Dec or gimme's. Once conference play starts there are no gimme's.

Heck, Clemson looks like world beaters if you go by their Pre-Jan record over the same time period as the first post: Since 04 Clemson has won 80% of their games (101-26) before January. After January they won 46% (83-96).

I wouldn't over analyze the program. We have more championships then anybody else over the last 25 years with the last one only a couple of years back...

Chicago 1995
01-24-2013, 02:44 PM
I am happy to discuss it. If I take a thread entitled "King of November/December" as antagonistic/provoking, and you do not, I hope we will just agree to disagree on it.

To be clear, I am not picking at you. I respect your posts and often agree with you. This just obviously strike us differently.

To the main point: different teams peak at different times. Some, like 2010, peak at just the right time. Others peak earlier. Some never quite get there at all. I would simply say that, historically, our season gets more difficult once conference play starts. Part of the reason our % falls is because we have to play @ Carolina, @ Md, etc. I would also suggest that a conference team ranked 10-25 is tougher to beat than one you do not have to play every year (especially when it is in a neutral site, as the pre-Christmas games are).

I would suggest the more relevant inquiry is our % at home, at neutral courts, and in true away games. The last category is significantly below the other two. And we don't play those until after Christmas.

I think that's right, but I think it's undercut by how weak the ACC has been over the last 5-7 years. Our schedule is tougher in that in conference play we actually play true road games, something save for the ACC-Big 10 challenge, we don't do much of in Nov and Dec -- and I'd note that we lost our last two true road games in the challenge at UW and OSU. But the ACC hasn't been the murderer's row it was at times in the past over the last decade. The ACC certainly hasn't been as tough as the Big East has been for the last several years or as tough as the Big 10 is this year.

I'd also note that the idea of regression -- or at least lack of progression -- has sources other than simple results. Some of it is subjective, in that we just don't look as good as we did early on. A basic example would be last season's wins in Maui to start the season vs. how we looked in February and March before Kelly was hurt. In just looking to winning percentages, a win is a win is a win, but if you're considering team growth, beating Michigan and Kansas on back to back days was more impressive than limping through an OT win over Va Tech.

Some of it has been that we've had a number of seasons too that have ended with a whimper, and that's two factors that make the difference between November and March stark. The first is the number of losses. Last year, we lost three of four to end the season (and yes, Kelly's injury is important to consider). The year before, we had a terrible performance at UNC to end the regular season, looked bad against Michigan and got blown out by Arizona. In 08-09, we lost to UNC to close the regular season struggled in the second round and got blitzed in the Sweet 16. That pattern was similar to 07-08, with a loss to UNC to close the regular season, struggles in the ACC tourney and two bad NCAA performances. The results over those stretches aren't great. Now that's where we're usually playing high level competition -- UNC at least once, the ACC tourney and NCAA tourney. At the same time, the results aren't great.

The other factor is that our NCAA tourney performance, save 2010 has been marked by underachievement since 2004. That's not all our fault. We were overseeded in at least a couple of those years -- 2005 and 2007 -- and we had injuries that got in the way with Kyrie and Ryan the last couple of years. That being said, in 2005, we were a 1 seed that lost to a 4, and a 4 that we'd beaten earlier in the year. In 2006, we were a #1 seed, had two first team AAs and lost to a 4. In 2007, we lost in the first round to an 11 seed. In 2008, we lost in the second round to a 7 seed. In 2009, we lost to a 3 seed BADLY. A 2/3 game is a toss up, but we were totally outclassed by Nova. 2010 was a wonderful, glorious exception that was spurred by growth and improvement arising from a senior being healthy for the first time in his career. 2011 was weird with Kyrie coming back, but we again, got outclassed by an Arizona team that we should have beaten. Last year we lost to a 15 seed.

No win in the NCAA tourney is easy, and everyone, save one lucky team, ends with an L. But when you compare the number of high quality wins we pile up in November and December over teams as good or better than those we're getting beaten by in the NCAA tourney, it's a fair impression that improvement isn't there.

CDu
01-24-2013, 02:48 PM
I decided to take a look at the last several years of stats for various teams. I took the last 5 years of NCAA champions: Kentucky, Connecticut, Duke, Carolina, and Kansas, and then added Michigan State and Butler, as they have been in the Final Four multiple times in that time frame. That looked like a fairly decent selection of recently elite teams.

I looked at this year (with incomplete data), and the previous 4 years. Why 4? Because I only have so much time to spend on this project. :)

I compiled the win percentage for November and December, for January through the end of the season (including the conference tournament, since that wasn't broken out separately), and for January through the end of the tournament. I've shown them in order by January through the end of the season (the second column).

Here are the results:
Nov-Dec Jan-Mar Jan-Apr
Kansas 88.5 87.2 84.4
Butler 72.3 79.8 79.0
Duke 95.0 78.5 78.3
Kentucky 83.9 75.3 76.7
Michigan State 80.3 71.4 71.7
Carolina 81.5 69.5 72.3
Connecticut 86.7 56.3 58.8
Our Nov-December stat is wonderfully gaudy. From January through the end of the year, we're not chopped liver, though. We're still doing quite well. It just SEEMS bad, because there's such a huge drop-off from 95% to 78%, a drop of over 16 percentage points! Butler is the only one that does better during conference play. Kansas is pretty consistent from non-con to conference play. Kentucky and Michigan State drop about 8 percentage points, Carolina drops 12, and Connecticut over 30!

Winning "only" 78.5% of our games from January on doesn't appear to be horrible. I'm pleasantly surprised with these stats.

Thanks for compiling this. It does appear, from this small sample of elite teams, that we are outperforming our Jan-March numbers in Nov/Dec moreso than other really good teams by virtue of playing that much better than the others in Nov/Dec.

bedeviled
01-24-2013, 03:09 PM
Yes, we are the Kings of Nov/Dec!! We are also quite good the rest of the year. Other posters have presented reasonable explanations, so I'll just add more data (source) (http://basketball.realgm.com/article/224651/Mike_Krzyzewski_Owns_November). It is based per-coach rather than per-team. It uses the Pythagorean winning percentage for up to the last 10 years if available. I've separated out the tables to rank for Nov/Dec, Jan/Feb, and Mar/Apr respectively.




Avg. Pyth. Win%
Current Team
Nov/Dec


1
Mike Krzyzewski
Duke :cool:
0.969


2
Bill Self
Kansas
0.952


3
Bo Ryan
Wisconsin
0.94


4
Roy Williams
North Carolina
0.938


5
Rick Barnes
Texas
0.931


6
Billy Donovan
Florida
0.917


7
Rick Pitino
Louisville
0.916


8
Jamie Dixon
Pittsburgh
0.915


9
Thad Matta
Ohio St.
0.914


10
Jim Boeheim
Syracuse
0.904


11
Bob Huggins
West Virginia
0.903


12
John Calipari
Kentucky
0.901


13
Tom Izzo
Michigan St.
0.896


14
Mark Few
Gonzaga
0.895


15
Bruce Weber
Kansas St.
0.891


16
Frank Martin
South Carolina
0.885


17
Dave Rose
BYU
0.883


18
Tubby Smith
Minnesota
0.882


19
Jay Wright
Villanova
0.875


20
Mike Montgomery
California
0.872


21
Sean Miller
Arizona
0.851


22
John Thompson III
Georgetown
0.85


23
Steve Alford
New Mexico
0.846


24
Kevin Stallings
Vanderbilt
0.843


25
Leonard Hamilton
Florida St.
0.843


26
Matt Painter
Purdue
0.841


27
Lorenzo Romar
Washington
0.84


28
Tony Bennett
Virginia
0.837


29
Josh Pastner
Memphis
0.837


30
Brad Stevens
Butler
0.829


31
Ben Howland
UCLA
0.822


32
Mike Brey
Notre Dame
0.821


33
Mark Gottfried
NC State
0.818


34
Frank Haith
Missouri
0.814


35
Chris Mack
Xavier
0.799


36
Mick Cronin
Cincinnati
0.786


37
John Beilein
Michigan
0.781


38
Buzz Williams
Marquette
0.768


39
Scott Drew
Baylor
0.662






Avg. Pyth. Win%
Current Team
Jan/Feb


1
Bill Self
Kansas
0.953


2
Mike Krzyzewski
Duke :cool:
0.95


3
Roy Williams
North Carolina
0.936


4
Bo Ryan
Wisconsin
0.929


5
John Calipari
Kentucky
0.927


6
Rick Barnes
Texas
0.923


7
Thad Matta
Ohio St.
0.915


8
Rick Pitino
Louisville
0.907


9
Bob Huggins
West Virginia
0.903


10
Tom Izzo
Michigan St.
0.903


11
Jamie Dixon
Pittsburgh
0.902


12
Billy Donovan
Florida
0.9


13
Jim Boeheim
Syracuse
0.899


14
Frank Martin
South Carolina
0.899


15
Mike Brey
Notre Dame
0.888


16
Tubby Smith
Minnesota
0.883


17
Mike Montgomery
California
0.872


18
Bruce Weber
Kansas St.
0.871


19
Jay Wright
Villanova
0.871


20
Matt Painter
Purdue
0.866


21
Chris Mack
Xavier
0.866


22
Kevin Stallings
Vanderbilt
0.862


23
Mark Few
Gonzaga
0.859


24
John Thompson III
Georgetown
0.857


25
Dave Rose
BYU
0.854


26
Ben Howland
UCLA
0.854


27
Steve Alford
New Mexico
0.846


28
Sean Miller
Arizona
0.837


29
Tony Bennett
Virginia
0.834


30
Leonard Hamilton
Florida St.
0.832


31
Frank Haith
Missouri
0.829


32
Lorenzo Romar
Washington
0.828


33
Mark Gottfried
NC State
0.812


34
John Beilein
Michigan
0.794


35
Josh Pastner
Memphis
0.786


36
Brad Stevens
Butler
0.78


37
Buzz Williams
Marquette
0.778


38
Scott Drew
Baylor
0.735


39
Mick Cronin
Cincinnati
0.723






Avg. Pyth. Win%
Current Team
Mar/Apr


1
Bill Self
Kansas
0.948


2
Roy Williams
UNC
0.942


3
John Calipari
Kentucky
0.942


4
Tom Izzo
Michigan St.
0.924


5
Rick Pitino
Louisville
0.921


6
Thad Matta
Ohio St.
0.919


7
Billy Donovan
Florida
0.917


8
Jamie Dixon
Pittsburgh
0.915


9
Jim Boeheim
Syracuse
0.915


10
Mike Krzyzewski
Duke :cool:
0.893


11
Mark Few
Gonzaga
0.882


12
Rick Barnes
Texas
0.881


13
Bo Ryan
Wisconsin
0.877


14
Josh Pastner
Memphis
0.869


15
Brad Stevens
Butler
0.869


16
Tubby Smith
Minnesota
0.868


17
Leonard Hamilton
Florida St.
0.862


18
Chris Mack
Xavier
0.856


19
Bob Huggins
West Virginia
0.855


20
Frank Martin
South Carolina
0.853


21
Lorenzo Romar
Washington
0.852


22
Bruce Weber
Kansas St.
0.85


23
Mike Brey
Notre Dame
0.848


24
Mark Gottfried
NC State
0.844


25
Ben Howland
UCLA
0.84


26
Jay Wright
Villanova
0.835


27
John Beilein
Michigan
0.834


28
Frank Haith
Missouri
0.832


29
Tony Bennett
Virginia
0.827


30
Steve Alford
New Mexico
0.826


31
Sean Miller
Arizona
0.809


32
Matt Painter
Purdue
0.797


33
Mike Montgomery
California
0.793


34
Kevin Stallings
Vanderbilt
0.793


35
Dave Rose
BYU
0.771


36
Scott Drew
Baylor
0.751


37
Buzz Williams
Marquette
0.735


38
John Thompson III
Georgetown
0.725


39
Mick Cronin
Cincinnati
0.639

Rich
01-24-2013, 04:03 PM
This is an interesting thought. It seemed to me in years past that Coach K cared a lot less about any individual game than maybe he does now. It almost seemed like sometimes he left the players on their own to lose the game in order to make an important teaching point that would help them later on (although it's not at all clear that he actually did that; this is only an impression I got watching some games in the past). That used to bug me a lot, because I kind of thought "you play to win the game." But now maybe I can see that there may be times when it is wiser to look at the long goals rather than the short ones.

I tend to agree. I remember Coach K would occasionally sub out the entire starting lineup for a few minutes to make a point. I know, in last night's game he subbed 3 at once, but still not the same. Here's another example somewhat on point - When Billy King played in the '80's, he was a defenisve stalwart, but a lousy free throw shooter. Nevertheless, Coach K left him in in close end-game situations under the premise "I'm going with the guys that got me here." He purposefully did not do the offense/defense substitutions we often see because he didn't want King to feel like he was going to get pulled from the game. Granted, Coach K didn't necessarily run a play to get the ball in King's hands either, but he didn't pull him from the game. At least last year with Plumlee, we would see Coach K make the substitution so that Plumlee wouldn't end up on the line. It's a different way to manage a game and players and I think Coach K's philosophy has changed somewhat in this regard.

mr. synellinden
01-24-2013, 05:27 PM
Thank god for Scheyer, Smith, Singler, Thomas, and Zoubek. Can you imagine what the perception of us would be without 2010?

Truthfully - I think this is a huge point. Since 2001, and other than 2004 and 2010, Duke has been abysmal in the NCAA tournament. Before anyone says anything - yes 99% of the programs would kill to have our abysmal record. But we're talking about Duke and its performance relative to its seeding. Obviously this is separate and apart from incredible success in the ACC tournament, which makes the following all the more perplexing.

In 2002, as defending national champions, with four NBA starters on the roster, we lost to #5 seeded Indiana in a Sweet Sixteen game we were favored to win by 13 points.

In 2003, as a #3 seed, we lost to #2 seeded Kansas in a Sweet Sixteen game in which we were a 3.5 point underdog.

In 2004, as a #1 seed, we lost to #2 seeded UCONN in a Final Four game in which we were a 2.0 point underdog. **I still think the referees cost us this game and we outplayed them.

In 2005, as a #1 seed, we lost (by 10) to #5 seeded Michigan State in a Sweet Sixteen game in which we were a 4.5 point favorite.

In 2006, as a #1 seed, we lost (by 8) to #4 seeded LSU in a Sweet Sixteen game in which we were a 6.0 point favorite.

AT THIS POINT, WE HAD LOST FOUR OUT OF FIVE YEARS IN THE SWEET SIXTEEN, THREE OF WHICH WE WERE THE #1 SEED AND AN AVERAGE 8.0 POINT FAVORITE

In 2007, as a #8 seed, we lost to #9 seeded VCU in a first round game in which we were a 6.5 point favorite. (BY THE WAY, WE LOST 8 OF OUR LAST 12 GAMES THAT YEAR)

In 2008, as a #2 seed, we lost to # 7 seeded West Virginia in a second round game in which we were a 4.0 point favorite.

(**Also worth noting we beat #15 seeded Belmont by 1 in a first round game in which we were a 20.0 point favorite.)

In 2009, as a #2 seed, we lost to #3 seeded Villanova (by 23 points) in a Sweet Sixteen game in which we were a 2.0 favorite

AT THIS POINT WE HAD LOST 5 STRAIGHT TOURNAMENT GAMES IN WHICH WE WERE A FAVORITE

2010 - ONE SHINING MOMENT

2011 - as a #1 seed, we lost to #5 seeded Arizona (by 16 points) in a Sweet Sixteen game in which we were an 8.5 point favorite.

2012 - as a #2 seed, we lost to #15 seeded Lehigh (by 5 points) in a first round game in which we were a 12.0 point favorite.


Thus, without 2010:

Duke has been eliminated in the NCAA tournament seven straight times as a favorite.

Duke has not made the elite 8 or Final Four in seven straight years despite being a #1 or #2 seed in six out of seven of those years

Duke has underperformed, relative to the spread, by an average of 16 points in its elimination games during the past seven years.

Duke has not exceeded expectations in the NCAA tournament, relative to seeding, once since 1991 and the UNLV game.


Objectively, Duke has been terrible in the NCAA tournament since 2001 - abysmal since 2004, and 2010 saved us from that narrative being discussed more often.

vick
01-24-2013, 05:50 PM
In 2002, as defending national champions, with four NBA starters on the roster, we lost to #5 seeded Indiana in a Sweet Sixteen game we were favored to win by 13 points.

In 2003, as a #3 seed, we lost to #2 seeded Kansas in a Sweet Sixteen game in which we were a 3.5 point underdog.

In 2004, as a #1 seed, we lost to #2 seeded UCONN in a Final Four game in which we were a 2.0 point underdog. **I still think the referees cost us this game and we outplayed them.

In 2005, as a #1 seed, we lost (by 10) to #5 seeded Michigan State in a Sweet Sixteen game in which we were a 4.5 point favorite.

In 2006, as a #1 seed, we lost (by 8) to #4 seeded LSU in a Sweet Sixteen game in which we were a 6.0 point favorite.

AT THIS POINT, WE HAD LOST FOUR OUT OF FIVE YEARS IN THE SWEET SIXTEEN, THREE OF WHICH WE WERE THE #1 SEED AND AN AVERAGE 8.0 POINT FAVORITE

In 2007, as a #8 seed, we lost to #9 seeded VCU in a first round game in which we were a 6.5 point favorite. (BY THE WAY, WE LOST 8 OF OUR LAST 12 GAMES THAT YEAR)

In 2008, as a #2 seed, we lost to # 7 seeded West Virginia in a second round game in which we were a 4.0 point favorite.

(**Also worth noting we beat #15 seeded Belmont by 1 in a first round game in which we were a 20.0 point favorite.)

In 2009, as a #2 seed, we lost to #3 seeded Villanova (by 23 points) in a Sweet Sixteen game in which we were a 2.0 favorite

AT THIS POINT WE HAD LOST 5 STRAIGHT TOURNAMENT GAMES IN WHICH WE WERE A FAVORITE

2010 - ONE SHINING MOMENT

2011 - as a #1 seed, we lost to #5 seeded Arizona (by 16 points) in a Sweet Sixteen game in which we were an 8.5 point favorite.

2012 - as a #2 seed, we lost to #15 seeded Lehigh (by 5 points) in a first round game in which we were a 12.0 point favorite.

You have a point here, and some of these were clearly real disappointments (2002, 2006, arguably 2011 and 2012 although injuries sort of cloud the issue in those years). But other years, I really don't think a reasonable fan base should think of as disappointing. Take 2005--this is a team whose second leading forward by minutes played was Lee Melchionni. This is a team that started Reggie Love at power forward for four games--a walk-on wide receiver from the football team! We shouldn't be upset they lost in the sweet sixteen--we should be on our hands and knees in gratitude for a massively overachieving performance leading to the one-seed. Part of the "problem" in other words is that teams which simply do not have any business talent-wise being thought of as elite wind up being well-coached and get good seeds, but then eventually run out of steam. That's a relatively good problem to have in the grand scheme of being a fan.

Kedsy
01-24-2013, 06:16 PM
You have a point here, and some of these were clearly real disappointments (2002, 2006, arguably 2011 and 2012 although injuries sort of cloud the issue in those years). But other years, I really don't think a reasonable fan base should think of as disappointing. Take 2005--this is a team whose second leading forward by minutes played was Lee Melchionni. This is a team that started Reggie Love at power forward for four games--a walk-on wide receiver from the football team! We shouldn't be upset they lost in the sweet sixteen--we should be on our hands and knees in gratitude for a massively overachieving performance leading to the one-seed. Part of the "problem" in other words is that teams which simply do not have any business talent-wise being thought of as elite wind up being well-coached and get good seeds, but then eventually run out of steam. That's a relatively good problem to have in the grand scheme of being a fan.

I agree. A lot of this talk is really about our expectations as fans. Personally, I always hope we'll do well in the tournament, but rationally I don't always expect it. I never expect a national championship, for example. Although some years I expect a Final Four. Specifically, this century, I expected to make the Final Four in 2001, 2002, 2004, 2006, 2010, and 2011 (although by the time the tournament came around in 2011, the Kyrie situation tempered my expectations). So, to me, the team underperformed in 2002, 2006, and (sort of) 2011. While I'm always disappointed after a Duke loss, I didn't expect anything past the Sweet 16 in 2000, 2003, 2005, 2007, 2008, 2009, or 2012, so the only times I felt the team underperformed were the two first round losses. Thus, speaking only of my own personal viewpoint, the team underperformed 5 out of 13 years this century. Is that a lot? No idea. My guess is it's no worse than anybody else.

Again in my personal view, if you expect a national championship every year, you're just asking to be disappointed. Nevertheless, it feels like that's the expectation of a lot of folks around here. If people didn't feel that way, perhaps they wouldn't complain about our tournament underperformance so much and instead (as suggested by Vick) might revel in our regular season overperformance leading to a high seed.

FWIW, before Ryan's injury I expected to make the Final Four this year. Depending on when Ryan gets back and how quickly we readjust to his presence, this season feels a little like 2011 to me right now. Hopefully we'll re-integrate Ryan better than we re-integrated Kyrie, and we'll have a happier ending.

Greg_Newton
01-24-2013, 06:22 PM
Thanks for compiling this. It does appear, from this small sample of elite teams, that we are outperforming our Jan-March numbers in Nov/Dec moreso than other really good teams by virtue of playing that much better than the others in Nov/Dec.

From the numbers bedeviled posted, it appears that we're also usually better in Jan/Feb. I'm guessing that if devil84 had separated out Mar/Apr for Jan/Feb in his analysis, it would have shown a more significant late-season drop.

Ben1029
01-24-2013, 06:30 PM
From the numbers bedeviled posted, it appears that we're also usually better in Jan/Feb. I'm guessing that if devil84 had separated out Mar/Apr for Jan/Feb in his analysis, it would have shown a more significant late-season drop.

The reason for this is the starters play too many minutes and are tired at this point.:cool::cool::cool:

bedeviled
01-24-2013, 06:46 PM
The reason for this is the starters play too many minutes and are tired at this point.
I never laugh at these jokes, but this one got me with it's straightforward "honesty"!;)

I would like to point out that some coaches <cough, Roy> got boosts in their Mar/Apr win% from participating in the NIT instead of the winners' bracket.

devil84
01-24-2013, 07:45 PM
From the numbers bedeviled posted, it appears that we're also usually better in Jan/Feb. I'm guessing that if devil84 had separated out Mar/Apr for Jan/Feb in his analysis, it would have shown a more significant late-season drop.

I've got them broken out. They are in order by tournament percentage. Note that this includes a year in the NIT for Carolina, Kentucky, and Connecticut; Butler has 2 years in the NCAA, one in the CIB, and one with no post season tourney. Again, all conference tournament games are in the March column (I had no way of separating them out).

Blue numbers represent the best percentage, red the worst.




Nov
Dec
Jan
Feb
Mar
Tourney


Carolina
78.1%
84.8%
68.6%
71.0%
68.8%
84.2%


Kentucky
83.9%
83.9%
78.4%
69.0%
78.9%
83.3%


Duke
97.1%
92.3%
75.6%
85.3%
72.2%
76.9%


Butler
69.0%

75.0%
78.0%
82.4%
77.8%
75.0%


Connecticut
90.0%
83.3%
60.5%
54.5%
50.0%
73.3%


Michigan State
80.0%
80.6%
78.9%
62.1%
70.6%
73.3%


Kansas
86.2%
90.6%
90.2%
87.1%
78.6%
60.0%



It's interesting that Kansas has some gaudy numbers in January and leads impressively in February, yet comes in last tournament winning percentage. Note that their championship came in 2008, which the year prior to the window for these stats.

Duke comes in 1st in November by a large margin, 1st in December, 4th in January, 2nd in February, 4th in March, and 3rd in tournaments.

Carolina comes in 5th, 3rd, 5th, 4th, and 5th -- pretty mediocre by these standards. Yet they are first in tournament winnings (aided by a deep run in the NIT the year after winning the NCAA Championship).

bedeviled
01-24-2013, 07:55 PM
Duke comes in 1st in November by a large margin, 1st in December, 4th in January, 2nd in February, 4th in March, and 3rd in tournaments.Thanks for the effort! Small correction: Duke is 5th in January. :mad: <shakes fist at January!!>

lotusland
01-24-2013, 08:16 PM
Thanks for the effort! Small correction: Duke is 5th in January. :mad: <shakes fist at January!!>

When are UNC and NCSU Kings? I'd rather be Duke. Not too many folks predicted we would be ranked #1 this early anyway so let's not have a panic attack just because we played a stinker in January. Let's see how we respond and if a couple of our heralded rectuits can make the most of their opportunity to contribute. Hopefully Ryan comes back and we have a solid bench by March.

wsb3
01-25-2013, 10:56 AM
I agree with this post. The period of seven in nine final fours was remarkable. I was fortunate enough to attend senior day in 91 with a friend..(Koubek and Buckley) vs Clemson. Neither of us left that night thinking that was the team that would give us the National Champion we had pulled for since the mid sixties. The following weekend they were crushed in the ACC Championship lending even more doubt.

What makes that streak even more remarkable is how much talent was present across the country. What a masterful job of peaking at the right time. Now it seems we often play our best before January. Except as stated by Mike Corey the 2010 season. I won't pretend to know all the reasons.


I don't think the past few seasons are fairly included without an asterisk, given the precarious injury situations of Kyrie and Kelly.

However, there's something to this.

During the stretch early in Coach K's Duke tenure when we advanced to seven FFs in nine years, we peaked late despite not always having the most sterling of regular season records.

That does seem to have flipped a bit post-2004, whatever the reasons. Except for 2010, of course.

mkirsh
01-25-2013, 12:00 PM
Truthfully - I think this is a huge point. Since 2001, and other than 2004 and 2010, Duke has been abysmal in the NCAA tournament. Before anyone says anything - yes 99% of the programs would kill to have our abysmal record. But we're talking about Duke and its performance relative to its seeding. Obviously this is separate and apart from incredible success in the ACC tournament, which makes the following all the more perplexing.

In 2002, as defending national champions, with four NBA starters on the roster, we lost to #5 seeded Indiana in a Sweet Sixteen game we were favored to win by 13 points.

In 2003, as a #3 seed, we lost to #2 seeded Kansas in a Sweet Sixteen game in which we were a 3.5 point underdog.

In 2004, as a #1 seed, we lost to #2 seeded UCONN in a Final Four game in which we were a 2.0 point underdog. **I still think the referees cost us this game and we outplayed them.

In 2005, as a #1 seed, we lost (by 10) to #5 seeded Michigan State in a Sweet Sixteen game in which we were a 4.5 point favorite.

In 2006, as a #1 seed, we lost (by 8) to #4 seeded LSU in a Sweet Sixteen game in which we were a 6.0 point favorite.

AT THIS POINT, WE HAD LOST FOUR OUT OF FIVE YEARS IN THE SWEET SIXTEEN, THREE OF WHICH WE WERE THE #1 SEED AND AN AVERAGE 8.0 POINT FAVORITE

In 2007, as a #8 seed, we lost to #9 seeded VCU in a first round game in which we were a 6.5 point favorite. (BY THE WAY, WE LOST 8 OF OUR LAST 12 GAMES THAT YEAR)

In 2008, as a #2 seed, we lost to # 7 seeded West Virginia in a second round game in which we were a 4.0 point favorite.

(**Also worth noting we beat #15 seeded Belmont by 1 in a first round game in which we were a 20.0 point favorite.)

In 2009, as a #2 seed, we lost to #3 seeded Villanova (by 23 points) in a Sweet Sixteen game in which we were a 2.0 favorite

AT THIS POINT WE HAD LOST 5 STRAIGHT TOURNAMENT GAMES IN WHICH WE WERE A FAVORITE

2010 - ONE SHINING MOMENT

2011 - as a #1 seed, we lost to #5 seeded Arizona (by 16 points) in a Sweet Sixteen game in which we were an 8.5 point favorite.

2012 - as a #2 seed, we lost to #15 seeded Lehigh (by 5 points) in a first round game in which we were a 12.0 point favorite.


Thus, without 2010:

Duke has been eliminated in the NCAA tournament seven straight times as a favorite.

Duke has not made the elite 8 or Final Four in seven straight years despite being a #1 or #2 seed in six out of seven of those years

Duke has underperformed, relative to the spread, by an average of 16 points in its elimination games during the past seven years.

Duke has not exceeded expectations in the NCAA tournament, relative to seeding, once since 1991 and the UNLV game.


Objectively, Duke has been terrible in the NCAA tournament since 2001 - abysmal since 2004, and 2010 saved us from that narrative being discussed more often.

Not sure if this confirms or disputes the original point, but think it is interesting to look at tournament performance relative to preseason expectations:

2002 - #1 pre-season, losing in sweet 16 pretty big disappointment
2003 - #6, losing in sweet 16 slight disappointment (one round early)
2004 - #2, made final four as expected
2005 - #11, losing in sweet 16 as expected
2006 - #1, losing in sweet 16 big disappointment
2007 - #12, losing in first round big disappointment
2008 - #13, losing in second round slight disappointment
2009 - #8, losing in sweet 16 slight disappointment
2010 - #9, Winner, substantial over-achievement
2011 - #1, losing in sweet 16 big disappointment (but with substantial KI injury)
2012 - #6 - results TBD

obviously a lot of flaws with this (mainly that preseason rankings are highly speculative), but using them as a rough guide to early expectations our tournament performance doesn't look as bad - not particularly good, but not terrible either. I see 4 of what I would call "big disappointments" in losing more than one round early (ie #4 team not making elite 8) in 2002, 2006, 2007, and 2011, 3 "slight disappointments" where we lose one round early (ie #4 team making elite 8 but not final 4) in 2003, 2008, and 2009, 2 years where we performed as expected in 2004 and 2005, and one year where we substantially over-achieved and won it all.

So going back to the main argument of whether we over-perform early and under-perform late vs over-perform early and just perform to expectations late, (don't think anyone is arguing that we don't over-perform early), it is probably true that we don't perform up to expectations in March, but it's also true that we ride the roller coaster during the year and let the hot starts raise expectations so March losses feel that much worse than if you re-set to early expectations.

devilsadvocate85
01-25-2013, 12:09 PM
I agree with this post. The period of seven in nine final fours was remarkable. I was fortunate enough to attend senior day in 91 with a friend..(Koubek and Buckley) vs Clemson. Neither of us left that night thinking that was the team that would give us the National Champion we had pulled for since the mid sixties. The following weekend they were crushed in the ACC Championship lending even more doubt.

What makes that streak even more remarkable is how much talent was present across the country. What a masterful job of peaking at the right time. Now it seems we often play our best before January. Except as stated by Mike Corey the 2010 season. I won't pretend to know all the reasons.



1) Has anyone considered that in some years we may have been "over-seeded" in the NCAA tournament? Regardless of claims to the contrary, I think it's impossible for the seedings not to have some "name bias".

2) Using Las Vegas odds / point spreads to look at a single game is not really valid. Point spreads are set to balance the betting on an individual game. They are not set to indicate what anyone thinks will actually happen.


I personally feel this kind of thing is overblown. The longer you sustain excellence, the harder it is to measure up to your past.

rsvman
01-25-2013, 01:30 PM
...
Thus, without 2010:

Duke has been eliminated in the NCAA tournament seven straight times as a favorite.

Duke has not made the elite 8 or Final Four in seven straight years despite being a #1 or #2 seed in six out of seven of those years

Duke has underperformed, relative to the spread, by an average of 16 points in its elimination games during the past seven years.

Duke has not exceeded expectations in the NCAA tournament, relative to seeding, once since 1991 and the UNLV game.


Objectively, Duke has been terrible in the NCAA tournament since 2001 - abysmal since 2004, and 2010 saved us from that narrative being discussed more often.

2010 didn't save us from a narrative, it brought us a national title. You can't just omit 2010 and then say we were terrible in the tournament. I'm struggling for an apt metaphor. This one isn't exactly what I want, but it makes me laugh, so I'll use it: A guy goes to 7-11 on his way home from work every Friday and buys a lottery ticket. In 2010, he wins $43 million in the lottery. Then his wife says, "Aside from that time you won $43 million, you've been abysmal in the lottery."

The 2010 victory, unexpected as it was, is as much a part of our tournament history as is the loss to Lehigh, unexpected as it was. It doesn't just stave off the narrative, it's an actual counter-example to the narrative.

mo.st.dukie
01-25-2013, 02:00 PM
If you removed 1986, 1988, 1989, 1990, 1991, 1992, 1994, 1999, 2001, 2004, and 2010 Duke and Coach K have been pretty terrible in the tournament. 21 of Coach K's seasons have been pretty awful by Duke's standards and only 11 (34%) have been good enough. He needs to pick it up.

freshmanjs
01-25-2013, 02:13 PM
2010 didn't save us from a narrative, it brought us a national title. You can't just omit 2010 and then say we were terrible in the tournament. I'm struggling for an apt metaphor. This one isn't exactly what I want, but it makes me laugh, so I'll use it: A guy goes to 7-11 on his way home from work every Friday and buys a lottery ticket. In 2010, he wins $43 million in the lottery. Then his wife says, "Aside from that time you won $43 million, you've been abysmal in the lottery."

The 2010 victory, unexpected as it was, is as much a part of our tournament history as is the loss to Lehigh, unexpected as it was. It doesn't just stave off the narrative, it's an actual counter-example to the narrative.

there is no question that Duke's ncaa tournament performance during the period 1986 - 2001 was far superior to the performance from 2002 - today. maybe that is luck or small sample size randomness. or maybe it isn't. but denying that the performance has been worse is just not reflective of the facts. the lottery analogy just doesn't hold up. if your lottery guy had a 1 in 10 chance of winning the lottery and won it once in 20 tries, the analogy would be closer. and the wife would be right.

rsvman
01-25-2013, 02:19 PM
I know the lottery analogy was terrible, but it still has a point. You lose the lottery every time you don't win it. If you win it, you didn't lose it. Looking back and saying, "if it weren't for the fact that you won the lottery, you always lose" is kind of like saying to the CEO of Citibank "if it weren't for the fact that you have a lot of money, you would be poor."

So, yes, we have underperformed relative to expectations several times in the recent past; an unnerving number, for diehard fans like myself. But 2010 is still a counter-example, and it shows that we can also overperform relative to expectations.

Kedsy
01-25-2013, 02:36 PM
there is no question that Duke's ncaa tournament performance during the period 1986 - 2001 was far superior to the performance from 2002 - today. maybe that is luck or small sample size randomness. or maybe it isn't. but denying that the performance has been worse is just not reflective of the facts.

Well, facts plus a very subjective selection of the sample. In the six season sample from 1995 to 2000, we only made one Final Four and won zero championships. In fact, we only made it as far as the Sweet 16 three of the six years. That's a lot worse than the past six seasons.

The real facts are we had an unbelievable run from 1986 to 1994, unmatched by any team, ever, in the modern NCAA tournament era, and since then we've had great teams with an astonishing record in the ACC tournament but with only sporadic NCAAT success. So?

freshmanjs
01-25-2013, 02:44 PM
Well, facts plus a very subjective selection of the sample. In the six season sample from 1995 to 2000, we only made one Final Four and won zero championships. In fact, we only made it as far as the Sweet 16 three of the six years. That's a lot worse than the past six seasons.

The real facts are we had an unbelievable run from 1986 to 1994, unmatched by any team, ever, in the modern NCAA tournament era, and since then we've had great teams with an astonishing record in the ACC tournament but with only sporadic NCAAT success. So?

the "so" is that Duke's performance has been significantly different recently vs early on in K's tenure. maybe that is uninteresting to you. i find it quite interesting and can consider several possible explanations including small sample size / random variation. your "real" facts (not sure what you mean by that) say the same thing. the recent ncaa tournament performance has not been as good as it used to be.

i apologize that my facts were not "real"...

Kedsy
01-25-2013, 03:10 PM
the "so" is that Duke's performance has been significantly different recently vs early on in K's tenure. maybe that is uninteresting to you. i find it quite interesting and can consider several possible explanations including small sample size / random variation. your "real" facts (not sure what you mean by that) say the same thing. the recent ncaa tournament performance has not been as good as it used to be.

i apologize that my facts were not "real"...

What I meant by "real" is we made 7 Final Fours in 9 years. Expanding that range through 2001 as you did, presumably to make the more recent stretch seem worse, is not a "fact," it's a subjective sample selection designed to make a point that isn't necessarily valid.

The 1986 through 1994 stretch was glorious but unlikely. Nobody has done it since, and unless they change the rules there's a good chance nobody will ever do it again. So, yes, comparing any team's or coach's performance to that is "uninteresting" to me. Attempting to show a difference that really isn't there similarly lacks interest.

There have been 18 seasons after our 1994 championship game appearance. Split them evenly almost any way you want and you won't see much of a difference. Two nine season segments? 1995-2003: six times to Sweet 16 or past, two Final Fours, one championship; 2004-2012: six times to Sweet 16 or past, two Final Fours, one championship. Exactly the same. Three six season segments? 1995 to 2000: three times to S16 or past, one Final Four, zero championships; 2001 to 2006: six times to S16 or past, two Final Fours, one championship; 2007 to 2012: three times to S16 or past, one Final Four, one championship. So, not exactly the same but the outlier is actually in the middle of the timeframe and the worst segment is the earliest.

Unless you want to analyze why the first five years and the most recent 18 years of K's Duke career don't compare favorably to an incomparable nine year stretch in the middle, I don't see the point of this discussion.

freshmanjs
01-25-2013, 03:12 PM
Unless you want to analyze why the first five years and the most recent 18 years of K's Duke career don't compare favorably to an incomparable nine year stretch in the middle, I don't see the point of this discussion.

then why do you keep engaging in it?

OldPhiKap
01-25-2013, 03:12 PM
2010 didn't save us from a narrative, it brought us a national title. You can't just omit 2010 and then say we were terrible in the tournament. I'm struggling for an apt metaphor. This one isn't exactly what I want, but it makes me laugh, so I'll use it: A guy goes to 7-11 on his way home from work every Friday and buys a lottery ticket. In 2010, he wins $43 million in the lottery. Then his wife says, "Aside from that time you won $43 million, you've been abysmal in the lottery."

The 2010 victory, unexpected as it was, is as much a part of our tournament history as is the loss to Lehigh, unexpected as it was. It doesn't just stave off the narrative, it's an actual counter-example to the narrative.

"Other than that, Mrs. Lincoln, how did you enjoy the play?"

It is difficult to give a complete narrative by taking out key facts.

Agree 100%.

On a related note, Brad Stevens got to the national championship game 100% of the time from 2010-2011, but hasn't gotten there since. Big O-fer.

freshmanjs
01-25-2013, 03:18 PM
What I meant by "real" is we made 7 Final Fours in 9 years. Expanding that range through 2001 as you did, presumably to make the more recent stretch seem worse, is not a "fact," it's a subjective sample selection designed to make a point that isn't necessarily valid.

The 1986 through 1994 stretch was glorious but unlikely. Nobody has done it since, and unless they change the rules there's a good chance nobody will ever do it again. So, yes, comparing any team's or coach's performance to that is "uninteresting" to me. Attempting to show a difference that really isn't there similarly lacks interest.

There have been 18 seasons after our 1994 championship game appearance. Split them evenly almost any way you want and you won't see much of a difference. Two nine season segments? 1995-2003: six times to Sweet 16 or past, two Final Fours, one championship; 2004-2012: six times to Sweet 16 or past, two Final Fours, one championship. Exactly the same. Three six season segments? 1995 to 2000: three times to S16 or past, one Final Four, zero championships; 2001 to 2006: six times to S16 or past, two Final Fours, one championship; 2007 to 2012: three times to S16 or past, one Final Four, one championship. So, not exactly the same but the outlier is actually in the middle of the timeframe and the worst segment is the earliest.

Unless you want to analyze why the first five years and the most recent 18 years of K's Duke career don't compare favorably to an incomparable nine year stretch in the middle, I don't see the point of this discussion.



that is not how i look at it. i am interested in performance in the ncaa-t RELATIVE to performance in the regular season in the same year. since you are accusing me (rudely) of cherry picking time frames to make a false point (a remarkably annoying accusation), i will choose 1 year chunks.

here is how i'd assess ncaat relative to regular season

1986 - on target
1987 - on target
1988 - exceeded
1989 - exceeded
1990 - exceeded
1991 - exceeded
1992 - on target
1993 - fell short
1994 - exceeded
1995 - n/a
1996 - on target
1997 - fell short
1998 - on target
1999 - on target
2000 - on target (given dunleavy sickness)
2001 - on target or exceeded
2002 - fell short
2003 - on target
2004 - on target / exceeded

2005 - fell short
2006 - fell short
2007 - fell short
2008 - fell short
2009 - fell short
2010 - exceeded
2011 - fell short
2012 - fell short


perhaps this sheds some light on why i drew the time periods the way i did. as i've said several times already, maybe this is explained by randomness or small sample size. but, i really don't see how you could say there is no difference in pre 2004 vs post.

Kedsy
01-25-2013, 03:22 PM
Truthfully - I think this is a huge point. Since 2001, and other than 2004 and 2010, Duke has been abysmal in the NCAA tournament. Before anyone says anything - yes 99% of the programs would kill to have our abysmal record. But we're talking about Duke and its performance relative to its seeding.

Just to be complete, we also underperformed to our seeding in 1984 (3 lost to 6), 1985 (3 lost to 11), 1993 (3 lost to 6), 1996 (8 lost to 9), 1997 (2 lost to 10), 1998 (1 lost to 2) and 2000 (1 lost to 5). So we were also "abysmal" before 1986 and from 1995 to 2000. The only periods we didn't underperform to our seeding the majority of the time were 1986 to 1994 and the very subjectively chosen period of 2001 to 2004.

Wander
01-25-2013, 03:33 PM
2005 - fell short
2006 - fell short
2007 - fell short
2008 - fell short
2009 - fell short
2010 - exceeded
2011 - fell short
2012 - fell short


Yeah, this is sort of the key point. We've won less than the expected number of tournament games relative to seeding in all the "fell short" years listed here. Again, my opinion is that it's really nothing to panic about, and I suspect it doesn't mean a whole lot, but I don't see why it's not an interesting point of discussion.

It might be useful to engage the null hypothesis for those who think this isn't even worth discussing. How many years would we have to "fall short" (again, defined by winning less than the average number of games according to our seed) before it does become a valid point of discussion? We're at 7 of the past 8 years at the moment. If it becomes 11 of the past 12, is it still not worth discussing? What about 15 of the past 16?

freshmanjs
01-25-2013, 03:40 PM
Yeah, this is sort of the key point. We've won less than the expected number of tournament games relative to seeding in all the "fell short" years listed here. Again, my opinion is that it's really nothing to panic about, and I suspect it doesn't mean a whole lot, but I don't see why it's not an interesting point of discussion.

It might be useful to engage the null hypothesis for those who think this isn't even worth discussing. How many years would we have to "fall short" (again, defined by winning less than the average number of games according to our seed) before it does become a valid point of discussion? We're at 7 of the past 8 years at the moment. If it becomes 11 of the past 12, is it still not worth discussing? What about 15 of the past 16?

to put the data another way, in the 19 year period 1986 - 2004, duke never "fell short" two years in a row. not ever in 19 years. amazing. now it has happened in 7 of the last 8. again, i'm not commenting on causality, but it sure is a striking difference in performance.

Kedsy
01-25-2013, 03:58 PM
here is how i'd assess ncaat relative to regular season

1986 - on target
1987 - on target
1988 - exceeded
1989 - exceeded
1990 - exceeded
1991 - exceeded
1992 - on target
1993 - fell short
1994 - exceeded
1995 - n/a
1996 - on target
1997 - fell short
1998 - on target
1999 - on target
2000 - on target (given dunleavy sickness)
2001 - on target or exceeded
2002 - fell short
2003 - on target
2004 - on target / exceeded

2005 - fell short
2006 - fell short
2007 - fell short
2008 - fell short
2009 - fell short
2010 - exceeded
2011 - fell short
2012 - fell short


perhaps this sheds some light on why i drew the time periods the way i did. as i've said several times already, maybe this is explained by randomness or small sample size. but, i really don't see how you could say there is no difference in pre 2004 vs post.

Again, you're being subjective. Personally I'd characterize many of these seasons differently. For example, we were the pre-season #1 team in the country in 1988-89 -- how could reaching the Final Four and losing the semifinal be considered exceeding expectations? And I think a lot of people around here would argue we fell short in 1999, and I doubt many would consider losing to a lower seeded directional school in 1996 to be "on target." In 1998 we fell short just as much as we did in 2009; in 2000 we fell short as much as we did in 2011; and in 2005 and 2008, while we lost to lower seeds, I personally didn't expect much (if at all) better than we did. So, in my subjective opinion, even using your arbitrary cutoff, the years from 1995 to 2004 included four or five "fall shorts" (depending on how you view 1999) and four or five "on targets." The years from 2005 to 2012 included four or five "fall shorts" (depending on how you view 2005) and three or four "on targets" or "exceeded." Not so much of a difference.

Also, you're completely ignoring 1981 to 1985 (or at least 1984 to 1985 if all you care about is the NCAAT), but we "fell short" in those years too, and they're "pre 2004," right?

Your earlier post spoke about pre-2001 vs. post and now you're talking about pre-2004 vs. post, so to me it seems like you're moving the goalposts again. I know you've mentioned small sample size, but the later you go with your cutoff date the smaller the sample becomes. At this point you're comparing 24 years to 8, which doesn't make a lot of sense to me.

throatybeard
01-25-2013, 04:06 PM
Except for Harrison, Taylor, Lincoln, Garfield, McKinley, Harding, Roosevelt, and Kennedy, the President never dies in office.

Kedsy
01-25-2013, 04:06 PM
We're at 7 of the past 8 years at the moment.

But we also fell short compared to expected number of wins relative to seeding in 5 of the previous 8 years, and that's using the most favorable cutoff for the argument.

If you take the past 20 years and cut it into two equal 10 year periods (still subjective, but at least we're using round numbers and even time periods), we fell short relative to seeding 6 times from 1993 to 2002 (and didn't make the tournament once) and 7 times from 2003 to 2012. Not much of a difference.

Ultimately I stand by my assessment that other than the amazing period from 1986 to 1994, we've performed pretty close to the same in the NCAAT for the other seasons of K's tenure.

Incidentally, the reason I keep trying to separate out 1986 to 1994 is because if you recognize that stretch as exceptional and probably unrepeatable, the question becomes "what did we do so well in that nine year period?" instead of "what are we doing so poorly in the past eight years?" Which I believe is an important distinction. The first question may be an interesting point of discussion. Personally, I don't think the second question is fair or reasonable.


.

freshmanjs
01-25-2013, 04:25 PM
Again, you're being subjective. Personally I'd characterize many of these seasons differently. For example, we were the pre-season #1 team in the country in 1988-89 -- how could reaching the Final Four and losing the semifinal be considered exceeding expectations? And I think a lot of people around here would argue we fell short in 1999, and I doubt many would consider losing to a lower seeded directional school in 1996 should be considered "on target." In 1998 we fell short just as much as we did in 2009; in 2000 we fell short as much as we did in 2011; and in 2005 and 2008, while we lost to lower seeds, I personally didn't expect much (if at all) better than we did. So, in my subjective opinion, even using your arbitrary cutoff, the years from 1995 to 2004 included four or five "fall shorts" (depending on how you view 1999) and four or five "on targets." The years from 2005 to 2012 included four or five "fall shorts" (depending on how you view 2005) and three or four "on targets" or "exceeded." Not so much of a difference.

Also, you're completely ignoring 1981 to 1985 (or at least 1984 to 1985 if all you care about is the NCAAT), but we "fell short" in those years too, and they're "pre 2004," right?

Your earlier post spoke about pre-2001 vs. post and now you're talking about pre-2004 vs. post, so to me it seems like you're moving the goalposts again. I know you've mentioned small sample size, but the later you go with your cutoff date the smaller the sample becomes. At this point you're comparing 24 years to 8, which doesn't make a lot of sense to me.

i'll try to address your points.

1) you can re-do everything i did with a non-arbitrary measure of expected wins per seed. if you do that, my point still holds

2) i disagree on your assessment of many of these seasons.

1988 - 89 - now you are the one moving the goalposts. preseason ranking has nothing to do with what i said at all. duke was a 2 seed and got to the final 4. that is not falling short of expected performance for seed.

1999, we won more than expected number of games for seed

1996, we lost an 8-9 game. those are essentially toss-ups so i said on target.

1998, we won 3 games as a #1 seed - about typical performance.

2009, maybe -- i could see a case for on target there vs fell short.

2000 vs 2011, i think are very different cases because of the dunleavy illness as i pointed out. you could say the return of kyrie offsets that. maybe. but i don't think so. duke was a #1 seed without kyrie and had the option to play without him if that was best for the team.

2005 and 2008 - if the discussion is how did the team do in the ncaat vs kedsy's ingoing expectations, that is a) totally uninteresting to me and b) not something i have any ability to evaluate. both of these teams did not win the expected number of games for seed

3) ok -- duke fell short twice before 1986. noted. does not change my conclusion

4) i see very little difference between 2001 cutoff and 2004 cutoff in terms of the conclusion. if the fact that i mentioned both is a problem for you, let's stick wiht 2004. i think the difference more striking after then.

5) yes!!! i am comparing 24 years to 8. and the 8 have not been at the same level in ncaa performance relative to seed (which reflects regular season). i am interested in a discussion about why that is. is it random, luck, something else? that would be a great conversation to have but for some reason you insist on denying that the 8 years have been different.

throatybeard
01-25-2013, 04:35 PM
1996, we lost an 8-9 game. those are essentially toss-ups so i said on target.

Indeed, I think since 1985, the Nines have a winning record over the Eights. Or at least that was the case for many years.

Also, with respect to the Eastern Michigan game, virtually the whole team was hurt in one fashion or another.

bedeviled
01-25-2013, 04:45 PM
1) you can re-do everything i did with a non-arbitrary measure of expected wins per seed.By nature, expectation is subjective, thus not a non-arbitrary measure.
2) i disagree on your assessment of many of these seasons. 1988 - 89 - now you are the one moving the goalposts. preseason ranking has nothing to do with what i said at all. duke was a 2 seed and got to the final 4. that is not falling short of expected performance for seed.I realize you are discussing this with Kedsy, but, to me, this IS the point. Just like choosing different timeframes, choosing a particular relative measures is subjective. (falling short of expected performance...of a seed, of preseason ranking, coach's poll, season win%, RPI, Pomeroy, of my own personal desires, etc.) I think, upthread, Kedsy compared the actual outcomes for comparative timeframes. If those outcomes didn't meet a particular set of expectations, then perhaps the problem is with the expectations.
5) yes!!! i am comparing 24 years to 8. and the 8 have not been at the same level in ncaa performance relative to seed (which reflects regular season). i am interested in a discussion about why that is. is it random, luck, something else? that would be a great conversation to have but for some reason you insist on denying that the 8 years have been different.After y'all establish the ground rules, I will be interested in hearing your reasoning for what you think is the difference in those years.

freshmanjs
01-25-2013, 04:47 PM
By nature, expectation is subjective, thus not a non-arbitrary measure

i meant compare number of ncaat wins actually achieved to the average number of ncaat wins achieved by previous teams with the same seed in the 64+ team era. there is a table upthread with those numbers.

bedeviled
01-25-2013, 04:52 PM
I get that and saw the table upthread. However, seeds are based on subjective expectations built on history, personal preference, ideas of what conferences/schedules are harder, etc.
EDIT: Those factors are not necessarily controlled / balanced across teams (like Duke!)

freshmanjs
01-25-2013, 04:54 PM
I get that and saw the table upthread. However, seeds are based on subjective expectations built on history, personal preference, ideas of what conferences/schedules are harder, etc.

i'm with you. it is possible that duke has been over-seeded in this recent period. that is one of several possible explanations i have in mind for the recent 8 year period. there is a counter balancing factor to consider, though, which is that over-seeding buys you easier opponents. #1 seeds do well in the tournament both because they are good teams AND because they have easier games.

OldPhiKap
01-25-2013, 04:56 PM
Any calculation based upon the assumption that a #1 seed is "supposed" to go to the final four, and that a #2 seed is "supposed" to get to the regional championship game, has little basis in reality and is fundamentally flawed. With all due respect, that simply ignores the nature of a six-game, one-loss elimination tournament. It also ignores the fact that, in each bracket, each of the top four teams are pretty darn good and in fact each of the teams that got there have a proven record sufficient to merit the invitation or earned bid. The seeding itself is highly subjective to start.

And to my mind, if we played out hearts out and lost -- no matter what round -- that is not below my expectations.

If someone else feels differently, well, we are all entitled to our subjective views. But they are exactly that.

freshmanjs
01-25-2013, 04:59 PM
Any calculation based upon the assumption that a #1 seed is "supposed" to go to the final four, and that a #2 seed is "supposed" to get to the regional championship game, has little basis in reality and is fundamentally flawed. With all due respect, that simply ignores the nature of a six-game, one-loss elimination tournament. It also ignores the fact that, in each bracket, each of the top four teams are pretty darn good and in fact each of the teams that got there have a proven record sufficient to merit the invitation or earned bid. The seeding itself is highly subjective to start.

And to my mind, if we played out hearts out and lost -- no matter what round -- that is not below my expectations.

If someone else feels differently, well, we are all entitled to our subjective views. But they are exactly that.


why do we talk on this board about "subjective" views as if there is something wrong with having subjective views. very odd.

personally, i don't have a view of "supposed to." but, i do have a view of typical performance of a team with a given seed and that is, to me, useful as a benchmark for comparison. fine if you disagree - feel free to ignore it.

fwiw, i agree about the play their hearts out point with respect to my expectations of the people on the team.

Wander
01-25-2013, 05:36 PM
Any calculation based upon the assumption that a #1 seed is "supposed" to go to the final four, and that a #2 seed is "supposed" to get to the regional championship game, has little basis in reality and is fundamentally flawed.

No one here has been using that assumption.

bedeviled
01-25-2013, 05:37 PM
why do we talk on this board about "subjective" views as if there is something wrong with having subjective views.It's likely related to the passion involved for AND against Duke basketball. It's a good thing. Truthfully, nearly all posts are opinions or subjective, but it is nice to try to distinguish "fact" from "feeling" when we can, especially when there is the potential of starting or perpetuating false negative Duke basketball memes. Take a look back at your post #69 as an example of a post that claims "fact" but could have been phrased a number of different, more palatable ways. ;):D The thread has many thoughtful, reasoned arguments for and against the claim, including the actual results pointed out by Kedsy and tourney performance of Blue Bloods by Devil84.
Most of all, I, personally, find nothing more annoying in a forum than the "Pudding Pops > Matt Ryan. FACT!" comments.:o

Duvall
01-25-2013, 05:42 PM
Indeed, I think since 1985, the Nines have a winning record over the Eights. Or at least that was the case for many years.

Also, with respect to the Eastern Michigan game, virtually the whole team was hurt in one fashion or another.

And the only NBA player on the floor wasn't wearing a Duke uniform.

Actually, looking back at some of these upsets with a better sense of the talent on both sides, it seems like the real scandal in these games is that the opposing coaches couldn't get more wins out of their elite talent during the regular season. Florida in 2000 and LSU in 2006 were particularly egregious.

freshmanjs
01-25-2013, 05:46 PM
It's likely related to the passion involved for AND against Duke basketball. It's a good thing. Truthfully, nearly all posts are opinions or subjective, but it is nice to try to distinguish "fact" from "feeling" when we can, especially when there is the potential of starting or perpetuating false negative Duke basketball memes. Take a look back at your post #69 as an example of a post that claims "fact" but could have been phrased a number of different, more palatable ways. ;):D The thread has many thoughtful, reasoned arguments for and against the claim, including the actual results pointed out by Kedsy and tourney performance of Blue Bloods by Devil84.
Most of all, I, personally, find nothing more annoying in a forum than the "Pudding Pops > Matt Ryan. FACT!" comments.:o

i do have a pretty strong view that the facts back up a difference in ncaat performance for duke in the recent 8 years vs the period that came before. there is obviously interpretation applied to those facts to reach that conclusion. Kedsy himself acknowledges (not in a manner that makes it at all enjoyable to be part of this board) over and over again that the 1986-1994 period was superior performance. the blue blood performance analysis is very interesting but does not comment on the comparison of duke pre 2001 (or 2004) to after.

bedeviled
01-25-2013, 06:07 PM
I don't share your view as stated - I think our recent performance has been on par with our near-recent performance. But, I do believe that our performance has been worse compared to expectations based on tournament seeding recently, which appears to be your supported assertion. Thus, if you believe that this is a systematic trend, I remain curious about your evidence of a systematic cause on the other side of the equation. Furthermore, I am interested in whether or not that systematic cause is meaningful (isolated to Duke, can be adjusted, etc).
The primary reason for this post, though, is to clarify my last post. I should have included the disclaimer that I do LOVE Pudding Pops, I just have a difficult time comparing them to Matt Ryan.

OldPhiKap
01-25-2013, 06:11 PM
No one here has been using that assumption.

I guess I am unclear what "expected number of wins relative to seeding" would mean other than that. But it may be my misunderstanding.

How many games is a #1 seed "expected" to win?

Kedsy
01-25-2013, 06:26 PM
here is how i'd assess ncaat relative to regular season


yes!!! i am comparing 24 years to 8. and the 8 have not been at the same level in ncaa performance relative to seed (which reflects regular season).

Before, you were talking about "ncaat relative to regular season," now you want to talk about underperforming seed. That's fine, but as someone said earlier, there are many ways to "reflect" the regular season. I would argue seeding in the NCAAT is not a particularly good way to do it, however. Be that as it may:



1984 - ??? (3 vs. 6 2nd round loss) -- should be fell short
1985 - ??? (3 vs. 11 2nd round loss) -- should be fell short
1986 - on target (1 vs. 7 made FF)
1987 - on target (5 vs. 1 S16 loss)
1988 - exceeded (2 vs. 1 made FF)
1989 - exceeded (2 vs. 1 made FF)
1990 - exceeded (3 vs. 1 made FF)
1991 - exceeded (2 vs. 1 made FF)
1992 - on target (1 vs. 2 made FF)
1993 - fell short (3 vs. 6 2nd round loss)
1994 - exceeded (2 vs. 1 made FF)
1995 - n/a
1996 - on target (8 vs. 9 1st round loss) -- should be fell short
1997 - fell short (2 vs. 10 2nd round loss)
1998 - on target (1 vs. 2 E8 loss) -- should be fell short
1999 - on target (1 vs. 6 made FF)
2000 - on target (1 vs. 5 S16 loss) -- should be fell short
2001 - on target (1 vs. 6 made FF)
2002 - fell short (1 vs. 5 S16 loss)
2003 - on target (3 vs. 2 S16 loss)
2004 - on target (1 vs. 7 made FF)

2005 - fell short (1 vs. 5 S16 loss)
2006 - fell short (1 vs. 4 S16 loss)
2007 - fell short (6 vs. 11 1st round loss)
2008 - fell short (2 vs. 7 2nd round loss)
2009 - fell short (2 vs. 3 S16 loss)
2010 - exceeded (1 vs. 3 made FF)
2011 - fell short (1 vs. 5 S16 loss)
2012 - fell short (2 vs. 15 1st round loss)


In parenthesis I have included either our last game or the game that got us into the Final Four.

First of all, if you're strictly talking about comparing performance against seed, you can't count 2 beating 1 as "exceeding" and 1 losing to 2 as "on target." You can't say 8 losing to 9 is "on target." You can't make injury excuses for one #1 losing to a #5 and not another. If you change 1996, 1998, and 2000 to "fell short," and it really does look like cherry picking.


i am interested in a discussion about why that is. is it random, luck, something else? that would be a great conversation to have but for some reason you insist on denying that the 8 years have been different.

Actually, what I've been insisting is that the years from 1986 to 1994 were different. The years before 1986 and after 1994, not so much.

We only have five post-1994 "on target" years. Here is our NCAAT path to the Final Four (or to a loss before the Final Four) for those five years:

1999: 16-9-12-6
2001: 16-9-4-6
2003: 14-11-2 (loss)
2004: 16-8-5-7
2010: 16-8-4-3

At the outset, why are 2003 and 2004 in the "we were good" days and 2005 and 2006 in the "something happened to us" days? That doesn't sound arbitrary to you?

Putting that aside, during all those years prior to the Final Four we only beat one team seeded higher than 4th, and that was in 2010 (during the "down" time). Other than 2010, prior to the Final Four we only beat two teams total seeded higher than 6th.

If the 2003 team had played a second round opponent as tough as 2009 West Virginia (instead of 11th seeded Central Michigan), would they have performed any better than our 2008 team? If 2006 LSU hadn't won on a last-second prayer to beat #12 Texas A&M, would our 2006 team fared just as well as the 2004 team? You may not be willing to entertain those sorts of questions as part of this debate, but when you have so few data points that luck factor makes the whole thing. Because if 2006 and 2008 are even with 2003 and 2004, then you're left with two Final Four visits by Shane Battier against the 2010 team, and to say that's a trend is silly.

Again, I say it ultimately comes down to what was different about 1986 to 1994, not what's been different about 2005 to 2012.

davekay1971
01-25-2013, 06:30 PM
I guess I am unclear what "expected number of wins relative to seeding" would mean other than that. But it may be my misunderstanding.

How many games is a #1 seed "expected" to win?

I think they were using a more reasonable calculation of average number of tournaments wins for each seed. i.e.: over a given period of time, 1 seeds win, on average, x.yz number of games per tournament. Then they compared Duke's performance to that and, if Duke won less games at a given seen than average, it was termed underperforming. Given the sample size (i.e.: Duke's number of times over a given period as a 1, 2, 3, etc seed), I'm having a hard time reaching any significant conclusions. I simply don't have the inclination to do a chi-square or linear regression or z-statistic or whatever would apply here, but I feel safe in assuming that, with such a small sample size, our standard deviation is pretty huge and Duke's performance over the last 10 years would be the perfect definition of a statistically insignificant trend.

However, accepting the argument that Duke has underperfomed the expected number of wins for our given seed over the last 10 years, I would offer a few possible explanations

1) It's a small sample size and we're simply seeing random variation (i.e.: kids have bad games, kids on other teams have great games, injuries, dumb luck, etc)
2) Coach K forgot how to coach the NCAA tournament
3) Duke may have a tendency to be overseeded given the program's, and K's, well earned tournament reputation
4) The greater dispersal of talent and quality of coaching has made the tournament a more fundamentally random event.
5) The NCAA's seeding system sucks and underrates less well established programs like Lehigh, VCU, etc. Frankly, both Lehigh and VCU were criminally underrated when they bounced us.

I personally think reason number 2 is complete bunk and the other 4 probably are all factors...particularly number 1.

freshmanjs
01-25-2013, 06:34 PM
I don't share your view as stated - I think our recent performance has been on par with our near-recent performance. But, I do believe that our performance has been worse compared to expectations based on tournament seeding recently, which appears to be your supported assertion. Thus, if you believe that this is a systematic trend, I remain curious about your evidence of a systematic cause on the other side of the equation. Furthermore, I am interested in whether or not that systematic cause is meaningful (isolated to Duke, can be adjusted, etc).
The primary reason for this post, though, is to clarify my last post. I should have included the disclaimer that I do LOVE Pudding Pops, I just have a difficult time comparing them to Matt Ryan.

i have a lot of hypotheses and no way to prove them. i also think there is a decent chance that this is just random variation from small sample size, as i've stated many times in this thread.

some ideas that i think are worth exploring (although some here seem to want to cut me off before this point and deny permission to discuss this topic):

1) the roll of constant favorite takes a toll. for some reason playing as a #1 seed or even the overall #1 seed has made Duke more tentative in the NCAA-T vs prior periods when Duke was actually seeded lower but had better ncaa-t performance. some of the great runs included somewhat surprising wins and I think coach K was at his very best preparing for games we were expected to lose (Purdue, Temple, Georgetown, UNLV 2). we haven't had many games we were expected to lose in the ncaat recently.

2) perhaps the change in quality of other ACC teams for the worse has made Duke less prepared for the NCAA-t than back in the 88-94 timeframe when the conference was so good. duke's sos is usually quite good as evaluated by the rpi. however a lot of this is because we don't play cupcakes and not because we play a lot of really tough teams. this varies by year and i'd love to see a history of our record vs top 10, 25, 50 teams by year going back to 1986. i don't have the data. i think this has also coincided with fewer really tough home/home non conference games (michigan, arizona, ucla, lsu (shaq) vs more recently st johns).

3) coach k has actually gotten better overall as a coach. this has led, in particular, to the incredibly good early season records. couple that with his recent scheduling philosophy which has seemed to be designed to optimize RPI, and maybe there is an over-seeding phenomenon happening with Duke

4) personnel. recent duke teams have not had the ingredients that the great coach k teams had. now, i'm not talking about absolute talent level here. obviously, overall talent level for duke and for everyone else was much higher 20 years ago than it is today. what i mean, though is on a relative basis, we have not had key ingredients. we have not had balance between perimeter and post scoring in many of the recent seasons and we have not had a relatively great point guard in many of these seasons. contrast to the earlier period when we had both of those items. i think when it comes to ncaa-t play, it is very difficult to win without both.

5) a lot of this is really a reflection of one recruiting class that just didn't work out. paulus, mcroberts, pocius, boateng, boykin. i like them all, especially paulus. but this class was not the juggernaut that it was predicted to be. this led directly to several years of the issues in #4.

6) coach k has changed his priorities, putting more emphasis on each regular season game and the ACCT than he used to. it is REALLY interesting that our ACCt performance was so good during this timeframe. maybe that wears down the team as well.

anyway, i'm not sure i believe any of these. also to state it again because i expect some folks to bite my head off about it -- i am not at all sure there is a trend here beyond randomness.

freshmanjs
01-25-2013, 06:39 PM
At the outset, why are 2003 and 2004 in the "we were good" days and 2005 and 2006 in the "something happened to us" days? That doesn't sound arbitrary to you?

Again, I say it ultimately comes down to what was different about 1986 to 1994, not what's been different about 2005 to 2012.

no, including our 2004 final 4 team in a list of good years does not seem arbitrary to me.

we must be talking past each other totally. "we were good" in ALL of these eras except for 1995-96. and maybe i'd say the 07 team was not very good. i would not remotely agree, AT ALL, with characterizing these eras in to "we were good" and "something happened to us." not at all. not remotely close to what i think.

what i'm talking about is different NCAA-T performance.

freshmanjs
01-25-2013, 06:40 PM
I think they were using a more reasonable calculation of average number of tournaments wins for each seed. i.e.: over a given period of time, 1 seeds win, on average, x.yz number of games per tournament. Then they compared Duke's performance to that and, if Duke won less games at a given seen than average, it was termed underperforming. Given the sample size (i.e.: Duke's number of times over a given period as a 1, 2, 3, etc seed), I'm having a hard time reaching any significant conclusions. I simply don't have the inclination to do a chi-square or linear regression or z-statistic or whatever would apply here, but I feel safe in assuming that, with such a small sample size, our standard deviation is pretty huge and Duke's performance over the last 10 years would be the perfect definition of a statistically insignificant trend.

However, accepting the argument that Duke has underperfomed the expected number of wins for our given seed over the last 10 years, I would offer a few possible explanations

1) It's a small sample size and we're simply seeing random variation (i.e.: kids have bad games, kids on other teams have great games, injuries, dumb luck, etc)
2) Coach K forgot how to coach the NCAA tournament
3) Duke may have a tendency to be overseeded given the program's, and K's, well earned tournament reputation
4) The greater dispersal of talent and quality of coaching has made the tournament a more fundamentally random event.
5) The NCAA's seeding system sucks and underrates less well established programs like Lehigh, VCU, etc. Frankly, both Lehigh and VCU were criminally underrated when they bounced us.

I personally think reason number 2 is complete bunk and the other 4 probably are all factors...particularly number 1.

has anyone ever suggested that #2 is a reasonable possible explanation?

Kedsy
01-25-2013, 06:47 PM
no, including our 2004 final 4 team in a list of good years does not seem arbitrary to me.

OK, I'll agree we're talking past each other. Absent something major happening at that juncture (e.g., a change in conference or a coaching change), saying that something that happened in 2004 is part of a 24 year trend but something that happened in 2005 is not but instead is part of a more recent 8 year trend seems to me to be the height of arbitrariness.

davekay1971
01-25-2013, 06:49 PM
has anyone ever suggested that #2 is a reasonable possible explanation?

I was listing possible explanations and clarifying that I thought it would not be a good one. However, there have certainly been better worded suggestions on DBR over recent years that some of Coach K's decisions in managing the development of the team throughout the season may contribute to some earlier than desired exits in the NCAAT (as did you about 4 or 5 posts up the thread). I'm being charitable in entertaining the notion that our 2005-now NCAAT performance is disappointing in any way other than as-compared-to-inflated-expectations (personally, I think this entire thread, from title on down, is bunk...but interesting and intelligently discussed bunk). However, assuming that Duke has indeed underperformed during that time frame, and it is a significant trend with 2010 as a significant outlier, coaching should certainly be considered as a possible cause, no?

freshmanjs
01-25-2013, 06:57 PM
OK, I'll agree we're talking past each other. Absent something major happening at that juncture (e.g., a change in conference or a coaching change), saying that something that happened in 2004 is part of a 24 year trend but something that happened in 2005 is not but instead is part of a more recent 8 year trend seems to me to be the height of arbitrariness.

i don't disagree, but by that logic you can't ever draw a cutoff anywhere to categorize things into groups. i really feel that you are just in this to try to win the internet. i tried to post a poll to see what others think, but i didn't do it correctly. i'll try again.

NSDukeFan
01-25-2013, 08:04 PM
i don't disagree, but by that logic you can't ever draw a cutoff anywhere to categorize things into groups. i really feel that you are just in this to try to win the internet. i tried to post a poll to see what others think, but i didn't do it correctly. i'll try again.

I agree with davekay in that the title of the thread immediately turned me off and perhaps that is me being too defensive. Fortunately, I thought Udaman made some good points, and have very much enjoyed this discussion. I have enjoyed reading your(freshmanjs) posts as well, but am not quite understanding your seeming to feel persecuted when others disagree with you. It's been an entertaining discussion, IMO, and I certainly don't mind a sarcastic remark every once in awhile.

freshmanjs
01-25-2013, 08:10 PM
I agree with davekay in that the title of the thread immediately turned me off and perhaps that is me being too defensive. Fortunately, I thought Udaman made some good points, and have very much enjoyed this discussion. I have enjoyed reading your(freshmanjs) posts as well, but am not quite understanding your seeming to feel persecuted when others disagree with you. It's been an entertaining discussion, IMO, and I certainly don't mind a sarcastic remark every once in awhile.

i enjoy the discussion as well with just about all of the posters here. i just don't like having my integrity questioned via an assertion that i have chosen cherry picked data to prove something that isn't true. several posters disagree with my conclusions and that is totally fine with me!

oddly enough, the one poster who seems to be so adamantly disagreeing with me is actually only debating when the lesser period began -1995 or 2005 (or maybe 2002?) and agrees on the fundamental premise that there was a greater and lesser period in ncaa-t performance relative to regular season performance.

davekay1971
01-25-2013, 08:15 PM
In the interest in exploring different ways of breaking down Duke's performance over the eras, I've decided to look at the effect of the Most Interesting Man in the World, Daniel Craig's inspiration for his take on James Bond AND Matt Damon's inspiration for his take on Jason Bourne, the man who's awesomeness Hugh Jackman tried to emulate portraying Wolverine and who's utter fearsomeness inspired the portrayal of the Hulk in the final battle scene in the Avengers. He doesn't drink beer very often, but when he does, he eats the bottle just because he can. Nate James.

James joined the staff as a full assistant in 2009-2010 season, at which time Duke promptly won the Natty. In the James years, Duke has won the National Championship 33% of the time and the ACC championship 67% of the time, easily outperforming the postseason performance of both Coach K and every other coach in the nation. K's performance pre-James: 3 natties in almost 30 years (and one of those was thanks to James, of course). After adding James to the staff: one natty every 3 years.

So forget the arbitrary breakdowns based on random years. Let's look at this according to the one factor that we all know can shift the fundamental laws of Life, the Universe, and Everything. Nate James. The results are clear.

freshmanjs
01-25-2013, 08:20 PM
In the interest in exploring different ways of breaking down Duke's performance over the eras, I've decided to look at the effect of the Most Interesting Man in the World, Daniel Craig's inspiration for his take on James Bond AND Matt Damon's inspiration for his take on Jason Bourne, the man who's awesomeness Hugh Jackman tried to emulate portraying Wolverine and who's utter fearsomeness inspired the portrayal of the Hulk in the final battle scene in the Avengers. He doesn't drink beer very often, but when he does, he eats the bottle just because he can. Nate James.

James joined the staff as a full assistant in 2009-2010 season, at which time Duke promptly won the Natty. In the James years, Duke has won the National Championship 33% of the time and the ACC championship 67% of the time, easily outperforming the postseason performance of both Coach K and every other coach in the nation. K's performance pre-James: 3 natties in almost 30 years (and one of those was thanks to James, of course). After adding James to the staff: one natty every 3 years.

So forget the arbitrary breakdowns based on random years. Let's look at this according to the one factor that we all know can shift the fundamental laws of Life, the Universe, and Everything. Nate James. The results are clear.

awesome. i think that's it.

it could also be my attendance at duke where my 6 years, 1991-94 and 99-00, accounted for 36% of K's final 4's and 50% of his national championships.

sagegrouse
01-25-2013, 08:37 PM
Without having read every thread in the post about the supposed downturn in Duke's post-season fortunes, I have seen this movie before. And I have posted something similar to what is below:

If you take 1995 as a watershed year, then K's pre-1995 NCAA results are far better than the post-1995 results. But why? Because if you look at ACC tournament results before and after, they have been far better afterwards. Overall Duke's post-season record is not much different if both NCAA and ACC games are counted. Fact is, we have been phenomenally successful in the ACC's, which is hardly ever remarked upon.

Not throwing blame, but just recognizing human nature, we like to focus on failures and not successes.

Here is my calculation of Duke's ACC record from 1996 through 2012: 36-7, with 10 championships.

Our NCAA record in this same period appears to be 40-15, excellent although not gaudy, with two NC's and four Final Fours.

Combined for postseason we are 76-22, a 0.776 success rate.

Yeah, I know we did better in 1986 through 2004, but that was amazing and amazingly lucky. And we were not nearly as successful in the ACC's, winning only three championships in a period where we had seven Final Fours and losing some ACC games we never should have lost.

sagegrouse
'More later -- good discussion'

Newton_14
01-25-2013, 08:43 PM
that is not how i look at it. i am interested in performance in the ncaa-t RELATIVE to performance in the regular season in the same year. since you are accusing me (rudely) of cherry picking time frames to make a false point (a remarkably annoying accusation), i will choose 1 year chunks.

here is how i'd assess ncaat relative to regular season

1986 - on target
1987 - on target
1988 - exceeded
1989 - exceeded
1990 - exceeded
1991 - exceeded
1992 - on target
1993 - fell short
1994 - exceeded
1995 - n/a
1996 - on target
1997 - fell short
1998 - on target
1999 - on target
2000 - on target (given dunleavy sickness)
2001 - on target or exceeded
2002 - fell short
2003 - on target
2004 - on target / exceeded

2005 - fell short
2006 - fell short
2007 - fell short
2008 - fell short
2009 - fell short
2010 - exceeded
2011 - fell short
2012 - fell short


perhaps this sheds some light on why i drew the time periods the way i did. as I've said several times already, maybe this is explained by randomness or small sample size. but, i really don't see how you could say there is no difference in pre 2004 vs post.

In the years I bold-ed, Coach is on record as stating those teams "Maxed Out Based On Their abilities", so in K's eyes those teams did not "fall short", but instead were "on target". The problem with this thread and argument, is that it is 100% based on the expectations of fans, media, and a NCAA Selection committee that is made up of lots of folks who know squat about basketball.

K has said many times over the years that he cares not what the expectations of outsiders are for His Team. He and his staff decide how good their teams are and what the expectations are for each "different" team each year. I support that philosophy 100%.

Each team is different each year, and deal with their own unique challenges. (The 08 team played the two games in the tourney with injuries, and 5 guys sick with the Flu for example, and had no low post game and presence at all) Plus rankings are not the end all be all or always accurate. For example, it was ridiculous that DUke was placed back in the Number 1 ranking this week as the current version of the team is not the best team in the country by a long shot. Miami is also rated several spots to low. So in reality, a Miami team that should be ranked in the Top 15 if not higher, beat a Duke team that should be ranked in the 8 to 12ish range, on Miami's home floor.

Same thing with tournament seedings. They are not infallible. Lehigh was much better than a 15 seed last year. Duke was not a true 2 seed without Kelly. Yet history will show that some poor little 15 seed that did not belong, beat a "juggernaut" 2 seed in some monumental upset. Not even close to truth. Duke should have won that game for sure, but it was not an upset of the magnitude it was made out to be.

throatybeard
01-25-2013, 09:36 PM
Man, I'd hate to see what this board would do to the 2005-present New England Patriots.

Check that, I'd love it. I don't like the Patriots.

freshmanjs
01-25-2013, 10:37 PM
another way of comparing 2 time periods is to count the # of ncaat wins for each team vs the number expected (meaning average # of wins for that seed). instead of a binary on exceeded or fell short, this adds a measure of magnitude. it still has the issue, of course, that seeding may just be a poor reflection of the quality of the teams.

1995 - 2004

1995 - NA
1996 - wins 0, expected wins 0.65. -0.65
1997 - wins 1, expected wins 2.43. -1.43
1998 - wins 3, expected wins 3.42. -0.42
1999 - wins 5, expected wins 3.42 - +1.58
2000 - wins 2, expected wins 3.42 - -1.42
2001 - wins 6, expected wins 3.42 - +2.58
2002 - wins 2, expected wins 3.42 - -1.42
2003 - wins 2, expected wins 1.84 - +0.16
2004 - wins 4, expected wins 3.42 - +0.58

Average per year = -.048


2005 - 2012
2005 - wins 2, expected wins 3.42 - -1.42
2006 - wins 2, expected wins 3.42 - -1.42
2007 - wins 0, expected wins 1.22 - -1.22
2008 - wins 1, expected wins 2.43 - -1.43
2009 - wins 2, expected wins 2.43 - -0.43
2010 - wins 6, expected wins 3.42 - +2.58
2011 - wins 2, expected wins 3.42 - -1.42
2012 - wins 0, expected wins 2.43 - -2.43

Average per year = -.899

This suggests that the latter era had a considerably larger average gap in ncaat performance vs seed than the prior era.

BlueDevilBrowns
01-25-2013, 10:48 PM
I know I'm more than a little late to the party on this topic but here's my view:

I split the K era in to 3 time periods - 1986-1994(9 seasons, 3 ACC's, 7 ff's, 2 NC's); 1998-2006(9 seasons, 7 ACC's, 3 ff's, 1 NC); 2007-2012(6 seasons, 3 ACC's, 1 ff, 1 NC).

I don't include 1981-1985 and 1995-1997 as I consider these transitional years due to completely building or rebuilding the program. The 2nd era ends in 06 due to the end of the JJ/Sheldon years, JJ especially becoming the "face" of 21st century Duke Basketball as a brand. Also, to me, there was a definite shift in recruiting philosophy as well, targeting more 3 and 4 year-type players due to the advent of the "1 and done" rule.

Having laid out the time periods, the numbers speak for themselves. Era 1, of course, wildly successful on a national scale but obviously less so on a conference scale. Era 2 seems to be the exact opposite, historically successful in the ACC but somewhat less so nationally). Era 3, so far, appears to be trending similarly with about 2 to 3 times as much conference success as national.

Why is this so?

I believe the answer is primarily because the ACC as a conference whole was entirely stronger during the 80's and early 90's. I don't think anyone in this thread would even debate that fact. So while battling day in and day out in a much stronger conference, by the time Duke reached the postseason, they were better prepared for the challenge of the ncaat.

As time wore on, moving into the 2nd era, The ACC began to weaken, especially after VTech and Miami were added, causing postseason success to dwindle in the NCAAT while Duke was able to dominate the conference tournament on an unprecedented scale. Most recently in the 3rd era, this conference erosion has continued or, at best, leveled off so the postseason results have continued the same.

To a lesser extent, just looking at the overall talent of the rosters during each era, it's easy to see a drop-off in talent from 2005 onward to the present. Much of this is the "Livingston effect" or the 1 and done rule causing Duke to target 3/4 year players as opposed to NBA-ready talent. Of course, in recent years, K has shifted back to attracting such talent and it remains to be seen if this improves our NCAAT results.

For 99% of schools, the last 6 years would be a Golden Age fans/alumni would tell there children about for years but, because we are Duke, the standard is much higher, right or not. The last 6 years simply haven't lived up to expectations but could change depending on what the next 3 years brings.

sagegrouse
01-25-2013, 10:55 PM
that is not how i look at it. i am interested in performance in the ncaa-t RELATIVE to performance in the regular season in the same year. since you are accusing me (rudely) of cherry picking time frames to make a false point (a remarkably annoying accusation), i will choose 1 year chunks.

here is how i'd assess ncaat relative to regular season

1986 - on target
1987 - on target
1988 - exceeded
1989 - exceeded
1990 - exceeded
1991 - exceeded
1992 - on target
1993 - fell short
1994 - exceeded
1995 - n/a
1996 - on target
1997 - fell short
1998 - on target
1999 - on target
2000 - on target (given dunleavy sickness)
2001 - on target or exceeded
2002 - fell short
2003 - on target
2004 - on target / exceeded

2005 - fell short
2006 - fell short
2007 - fell short
2008 - fell short
2009 - fell short
2010 - exceeded
2011 - fell short
2012 - fell short


perhaps this sheds some light on why i drew the time periods the way i did. as i've said several times already, maybe this is explained by randomness or small sample size. but, i really don't see how you could say there is no difference in pre 2004 vs post.

I am certain my comments will be viewed as rude, because anything short of total agreement with your analysis seems to elicit the charge. (Just kidding.)

First of all, you have detected a pattern in the data which seems to have meaning to you. Now, exactly why does a "point in time" suggest a change in underlying behavior? Can't see why, especially since the 2005 and 2006 teams had the same two key players as the 2004 team. Your change of sea state looks like a coincidence to me, since we had the same coach and many of the same players overlapping the supposed time-point.

Second, no fair censoring the data wrt 2000 and 1984 and 1985. By any standard, the 2000 team underperformed, although you chose to adjust the underlying expectations due to Dunleavy's unavailability. Every statistican's hair stands on end. There are lots of injuries; how about 2011? How about Jay's injuries in 1986 and Cherokee's ankle in 1992? How about overcoming Carlos's in 2001? Also, you left out the years 1984 and 1985, where Duke underperformed its seed. Can't do that -- the data speak for themselves, or they stay mute.

Third, "runs tests," such as looking at a sequence of events and divining a pattern, are very weak statistical tests because there is no underlying hypothesis to test. Now I made the same error (to make a different point) by using 1995 as a breakpoint -- the nine years before we had phenomenal success in the NCAA's; the sixteen years after we had phenomenal success in the ACC's (where we didn't before). And BTW we haven't done so badly in the NCAA's of late.

Fourth, is the Jim Sumner-Sage Grouse hypothesis: Duke has been overseeded fairly often in the past ten years, mostly because it has won the ACC tournament so often. Other conference reg. season champs frequently lose in their conference tournament (and this is true in spades of UNC). So the supposed high-seeded Duke has often backed into a high seed when the regular season record didn't merit it. This is true in 2003 (#12 AP and then a #3 seed) , 2005 (#6 AP and then a #1 seed), and 2009 (#9 AP and then a #2 seed).

Fifth, and I suppose this is "randomness," we need to give credit to the other teams. Arizona in 2011, for example. Those guys hit two outrageous threes, perfectly defended, at the end of the first half to narrow the margin to six points. Sir Charles says, "You see how Arizona raced off the court. Those guys are pumped. They're gonna win this game." "No way," I say to myself, but he was right. They miss both of those threes and Duke wins. Well, the NCAA is an interesting combination of whimsical outcomes and sheer torture, depending on the year and your team.

Anyway, didn't mean to pull your chain, but these are some random thoughts.

sagegrouse

freshmanjs
01-25-2013, 11:03 PM
I am certain my comments will be viewed as rude, because anything short of total agreement with your analysis seems to elicit the charge. (Just kidding.)

First of all, you have detected a pattern in the data which seems to have meaning to you. Now, exactly why does a "point in time" suggest a change in underlying behavior? Can't see why, especially since the 2005 and 2006 teams had the same two key players as the 2004 team. Your change of sea state looks like a coincidence to me, since we had the same coach and many of the same players overlapping the supposed time-point.

Second, no fair censoring the data wrt 2000 and 1984 and 1985. By any standard, the 2000 team underperformed, although you chose to adjust the underlying expectations due to Dunleavy's unavailability. Every statistican's hair stands on end. There are lots of injuries; how about 2011? How about Jay's injuries in 1986 and Cherokee's ankle in 1992? How about overcoming Carlos's in 2001? Also, you left out the years 1984 and 1985, where Duke underperformed its seed. Can't do that -- the data speak for themselves, or they stay mute.

Third, "runs tests," such as looking at a sequence of events and divining a pattern, are very weak statistical tests because there is no underlying hypothesis to test. Now I made the same error (to make a different point) by using 1995 as a breakpoint -- the nine years before we had phenomenal success in the NCAA's; the sixteen years after we had phenomenal success in the ACC's (where we didn't before). And BTW we haven't done so badly in the NCAA's of late.

Fourth, is the Jim Sumner-Sage Grouse hypothesis: Duke has been overseeded fairly often in the past ten years, mostly because it has won the ACC tournament so often. Other conference reg. season champs frequently lose in their conference tournament (and this is true in spades of UNC). So the supposed high-seeded Duke has often backed into a high seed when the regular season record didn't merit it. This is true in 2003 (#12 AP and then a #3 seed) , 2005 (#6 AP and then a #1 seed), and 2009 (#9 AP and then a #2 seed).

Fifth, and I suppose this is "randomness," we need to give credit to the other teams. Arizona in 2011, for example. Those guys hit two outrageous threes, perfectly defended, at the end of the first half to narrow the margin to six points. Sir Charles says, "You see how Arizona raced off the court. Those guys are pumped. They're gonna win this game." "No way," I say to myself, but he was right. They miss both of those threes and Duke wins. Well, the NCAA is an interesting combination of whimsical outcomes and sheer torture, depending on the year and your team.

Anyway, didn't mean to pull your chain, but these are some random thoughts.

sagegrouse

i don't disagree with anything you wrote --- except that i would find it rude. i don't see anything rude in your post.

sorry -- editing to add one thing as food for thought. i think the replacement of chris duhon with dockery/paulus at pg made a massive difference between 2004 and 2005 and, therefore, think it is actually a reasonable break point to look at. our lack of an excellent pg is one of the things that I think made it tougher for us in the NCAAT in many of the seasons post 2004.

bedeviled
01-25-2013, 11:06 PM
I'm having a hard time reaching any significant conclusions...[SNIP], but I feel safe in assuming that, with such a small sample size, our standard deviation is pretty huge and Duke's performance over the last 10 years would be the perfect definition of a statistically insignificant trend.I’m not a statistician, but, if you forced me to declare the quickest dirty way to test, I say do a t-test of average expected #wins versus actual #wins to first see if we significantly underperform.

T-tests(Done for sets of 8 years as the last 8 years appears to be the timeframe in question)

There is no statistical difference between number of games won in 1997-2004 tournaments versus number of games expected to win.
There is no statistical difference between number of games won in 2005-2012 tournaments versus number of games expected to win.
If we group all of 1997-2012 together, there is no statistical difference between games won and games expected to win.
There is no statistical difference between difference in games won (Actual minus expected) for 1997-2004 versus 2005-2012.
It’s definitely too small of a sample to statistically analyze based on individual seeds.
There is no statistical difference between seeding in 1997-2004 vs 2005-2012.
There is no statistical difference between seed of team lost to in 1997-2004 vs 2005-2012.

ANOVA
My analysis is limited to Excel which has some requirements I can’t break, so I again can’t use seedings as a factor. But, if Time1 and Time2 are 1997-2004 and 2005-2012, respectively, with Group1 and Group2 being ActualDuke (tournament wins each year) and ExpectedDuke (wins predicted by history of seed), then
Time, Group, and TimeXGroup are all statistically insignificant. I'm not certain this makes sense, but it was the best I could do.

As you conjectured, we likely have too small of a data set for formal statistics. Looks like this is still up for debate...but I guess that's better than being significantly worse!!

throatybeard
01-25-2013, 11:10 PM
another way of comparing 2 time periods is to count the # of ncaat wins for each team vs the number expected (meaning average # of wins for that seed). instead of a binary on exceeded or fell short, this adds a measure of magnitude. it still has the issue, of course, that seeding may just be a poor reflection of the quality of the teams.

1995 - 2004

1995 - NA
1996 - wins 0, expected wins 0.65. -0.65
1997 - wins 1, expected wins 2.43. -1.43
1998 - wins 3, expected wins 3.42. -0.42
1999 - wins 5, expected wins 3.42 - +1.58
2000 - wins 2, expected wins 3.42 - -1.42
2001 - wins 6, expected wins 3.42 - +2.58
2002 - wins 2, expected wins 3.42 - -1.42
2003 - wins 2, expected wins 1.84 - +0.16
2004 - wins 4, expected wins 3.42 - +0.58

Average per year = -.048


2005 - 2012
2005 - wins 2, expected wins 3.42 - -1.42
2006 - wins 2, expected wins 3.42 - -1.42
2007 - wins 0, expected wins 1.22 - -1.22
2008 - wins 1, expected wins 2.43 - -1.43
2009 - wins 2, expected wins 2.43 - -0.43
2010 - wins 6, expected wins 3.42 - +2.58
2011 - wins 2, expected wins 3.42 - -1.42
2012 - wins 0, expected wins 2.43 - -2.43

Average per year = -.899

This suggests that the latter era had a considerably larger average gap in ncaat performance vs seed than the prior era.

I'm not sure in which of the three threads on this topic I'm supposed to be posting.

One of the problems with the above is that in actuality, a team has to have an integer between zero and six as its win total, and in a single elimination tournament, the marginal chances of getting an additional win get harder and harder as you go along. When you sometimes win or lose by a single possession, it's kind of ridiculous to categorize, for example, the 1998 team (32-4) who lost to Kentucky by one possession as an "underperformer." If even one of those onslaught of threes that cut the game from 71-54 to 71-70 doesn't go in, suddenly that team is a significant "overperformer" instead of an underperformer. That just doesn't make any sense.

The difference between 2008's -1.43 and 2012's -2.43 is one Gerald Henderson drive against Belmont.

I'm not sure this analysis is telling us much, except that we maybe should have beaten Lehigh, Michigan State, and Indiana.

freshmanjs
01-25-2013, 11:10 PM
I’m not a statistician, but, if you forced me to declare the quickest dirty way to test, I say do a t-test of average expected #wins versus actual #wins to first see if we significantly underperform.

T-tests(Done for sets of 8 years as the last 8 years appears to be the timeframe in question)

There is no statistical difference between number of games won in 1997-2004 tournaments versus number of games expected to win.
There is no statistical difference between number of games won in 2005-2012 tournaments versus number of games expected to win.
If we group all of 1997-2012 together, there is no statistical difference between games won and games expected to win.
There is no statistical difference between difference in games won (Actual minus expected) for 1997-2004 versus 2005-2012.
It’s definitely too small of a sample to statistically analyze based on individual seeds.
There is no statistical difference between seeding in 1997-2004 vs 2005-2012.
There is no statistical difference between seed of team lost to in 1997-2004 vs 2005-2012.

ANOVA
My analysis is limited to Excel which has some requirements I can’t break, so I again can’t use seedings as a factor. But, if Time1 and Time2 are 1997-2004 and 2005-2012, respectively, with Group1 and Group2 being ActualDuke (tournament wins each year) and ExpectedDuke (wins predicted by history of seed), then
Time, Group, and TimeXGroup are all statistically insignificant. I'm not certain this makes sense, but it was the best I could do.

As you conjectured, we likely have too small of a data set for formal statistics. Looks like this is still up for debate...but I guess that's better than being significantly worse!!

what confidence level are you requiring to determine statistical significance?

freshmanjs
01-25-2013, 11:18 PM
I'm not sure in which of the three threads on this topic I'm supposed to be posting.

One of the problems with the above is that in actuality, a team has to have an integer between zero and six as its win total, and in a single elimination tournament, the marginal chances of getting an additional win get harder and harder as you go along. When you sometimes win or lose by a single possession, it's kind of ridiculous to categorize, for example, the 1998 team (32-4) who lost to Kentucky by one possession as an "underperformer." If even one of those onslaught of threes that cut the game from 71-54 to 71-70 doesn't go in, suddenly that team is a significant "overperformer" instead of an underperformer. That just doesn't make any sense.

The difference between 2008's -1.43 and 2012's -2.43 is one Gerald Henderson drive against Belmont.

I'm not sure this analysis is telling us much, except that we maybe should have beaten Lehigh, Michigan State, and Indiana.

of course what you say is true. it's also true that duke's 1992 championship would not have happened without an extremely unlikely miracle play. yet, we ascribe a ton of value to that championship. perhaps that doesn't make a lot of sense either, but that's how the sport is judged.

bedeviled
01-25-2013, 11:24 PM
i have a lot of hypotheses and no way to prove them.
I started to address this post point-by-point, but, honestly, #2 was the only one I felt was worth exploring in more depth. Here's some data (pre-Tournament) for your giggles.

2) perhaps the change in quality of other ACC teams for the worse has made Duke less prepared for the NCAA-t...[SNIP]...i'd love to see a history of our record vs top 10, 25, 50 teams by year going back to 1986. i don't have the data. i think this has also coincided with fewer really tough home/home non conference gamesObviously, you can deduce ACC numbers by subtracting Out-of-Conference from the totals. And, you can get the rest of you exploration back to 1986 by looking at the schedule/results section of goduke.com.



Top25/Yr
AvgTop25Rank
Top25Away/Yr
Top25OOC/Yr
Top25OOCAway/Yr


1997-2004
9.125
11.7260274
3.75
2.875
1


2005-2012
7.5
12.91666667
2.625
2.75
0.625











Top10/yr
AvgTop10Rank
Top10Away/Yr
Top10OOC/Yr
Top10OOCAway/Yr


1997-2004
4.125
5.454545455
1.75
1.5
0.625


2005-2012
2.875
5.043478261
1.5
0.875
0.5


Top25/Yr = avg # of games against Top25 opponents
Rank is the opponents average rank
Top25Away/Yr = avg # of away games against Top25 opponents
Top25OOC/YR = avg # of games against out-of-conference Top25 opponents
Top25OOCAway/Yr = avg # of away games against out-of-conference Top25 opponents (Note: I included Stanford in Oakland and Kansas in Kansas City as aways)

EDIT: In answer to your question about my last post, alpha was 0.05...and none were close.

freshmanjs
01-25-2013, 11:32 PM
I started to address this post point-by-point, but, honestly, #2 was the only one I felt was worth exploring in more depth. Here's some data for your giggles.
Obviously, you can deduce ACC numbers by subtracting Out-of-Conference from the totals. And, you can get the rest of you exploration back to 1986 by looking at the schedule/results section of goduke.com.



Top25/Yr
AvgTop25Rank
Top25Away/Yr
Top25OOC/Yr
Top25OOCAway/Yr


1997-2004
9.125
11.7260274
3.75
2.875
1


2005-2012
7.5
12.91666667
2.625
2.75
0.625











Top10/yr
AvgTop10Rank
Top10Away/Yr
Top10OOC/Yr
Top10OOCAway/Yr


1997-2004
4.125
5.454545455
1.75
1.5
0.625


2005-2012
2.875
5.043478261
1.5
0.875
0.5


Top25/Yr = avg # of games against Top25 opponents
Rank is the opponents average rank
Top25Away/Yr = avg # of away games against Top25 opponents
Top25OOC/YR = avg # of games against out-of-conference Top25 opponents
Top25OOCAway/Yr = avg # of away games against out-of-conference Top25 opponents (Note: I included Stanford in Oakland and Kansas in Kansas City as aways)

EDIT: In answer to your question about my last post, alpha was 0.05...and none were close.

Thanks -- this is really interesting. looks like the schedule has gotten weaker, but maybe not materially so. hard to tell.

all in all, i suspect that any variance in performance that we've seen is really just randomness.

also in case it comes across differently, i think our program has an amazing track record of success in all of the time periods being discussed.

Kedsy
01-25-2013, 11:39 PM
I don't include 1981-1985 and 1995-1997 as I consider these transitional years due to completely building or rebuilding the program.

I agree that 1983 to 1985 and 1995 to 1997 can be considered transitional/rebuilding years. (IMO, 1981 and 1982 were more or less lost years due to Bill Foster's abrupt departure, so the rebuilding couldn't really start until K's first true recruiting class came on board in 1982-83.) I'd go further and say 2007 to 2009 follow the same exact pattern as '83 to '85 and '95 to '97, and thus could also be considered separately.

If you accept my hypothesis, then the third nine-year era would be counted from 2010 to 2018 (assuming K stays that long), and we won't be able to judge the relative successes until then. Nonetheless, I bet the "third era" will end up looking at least as rosy as 1998 to 2006. From an NCAAT standpoint, it's hard to imagine any team in the forseeable future having Duke's 1986 to 1994 level of success over any nine year period.

throatybeard
01-26-2013, 12:42 AM
Consider how mid-majors may or may not have gotten stronger compared to the 80s. Yeah you had a couple Cleveland States and Bradleys in the 1980s, but no Butlers.

How much of this swings on our perception of VCU and Lehigh specifically? And how much of that, in turn, has really been compared to every other top program? Say you sat down and did this expected-wins analysis with, say, UNC, UK, KU, MSU, UConn, and a few others you want to name (is IU back?). Would they look way better than us, especially over a cherry-picked period like 2005-present?

For example, take Carolina. Let's use Dean's retirement as an arbitrary marker. (I'm aware of the problem of this cherry-picking). This excludes his 1993 title but also K's in 1991 and 1992, since we're so focused on national titles. Two titles apiece. But but it also excludes our 1995 grease-fire, I guess. But, UNC has missed the tournament in 2002, 2003 and 2010. Shouldn't missing the tournament conceivably result in even greater negative points than losing to VCU? And why don't ACCT expected wins factor into the discussion? Some of the more amazing postseason feats in K's tenure were the 2000, 2003, 2005, 2009 and 2011 ACCTs. I've never seen stats on expected wins by seed in the ACCT. I do know we've won quite a few from the 2 or 3 spot.

Remind me why we're whining about the postseason record of roughly the second best college basketball coach ever? I'm slow and I need things explained to me.

BlueDevilBrowns
01-26-2013, 08:40 AM
I agree that 1983 to 1985 and 1995 to 1997 can be considered transitional/rebuilding years. (IMO, 1981 and 1982 were more or less lost years due to Bill Foster's abrupt departure, so the rebuilding couldn't really start until K's first true recruiting class came on board in 1982-83.) I'd go further and say 2007 to 2009 follow the same exact pattern as '83 to '85 and '95 to '97, and thus could also be considered separately.

If you accept my hypothesis, then the third nine-year era would be counted from 2010 to 2018 (assuming K stays that long), and we won't be able to judge the relative successes until then. Nonetheless, I bet the "third era" will end up looking at least as rosy as 1998 to 2006. From an NCAAT standpoint, it's hard to imagine any team in the forseeable future having Duke's 1986 to 1994 level of success over any nine year period.


I agree that 2007-2009 could be considered transitional years. However, the reason I included them in the current era was because there wasn't a complete rebuild due to unprecedented circumstances, such as building a program from scratch in the 80's or having to leave the program for basically an entire season in 1995. In fact, you could argue that Coach K had to entirely rebuild Duke's brand and reputation with recruits after the '95 disaster.

If 2007 looked like a rebuild, to me, that's on the coaching staff as it's their responsibility to anticipate turnover of the roster. Further, I don't believe anyone from the 2006 roster left or transferred unexpectedly, as Sheldon, Dockery, and JJ were all seniors. It's my contention that the players the coaches were bringing in, while high-level, weren't the NBA-level talents of previous eras and the overall results in the NCAAT prove it. Now, if we could project out 5 years in the future, then certainly the 3rd era could extend beyond a 9 year term, but I believe the 3rd era has to begin in 2007.

BlueDevilBrowns
01-26-2013, 08:46 AM
Consider how mid-majors may or may not have gotten stronger compared to the 80s. Yeah you had a couple Cleveland States and Bradleys in the 1980s, but no Butlers.

How much of this swings on our perception of VCU and Lehigh specifically? And how much of that, in turn, has really been compared to every other top program? Say you sat down and did this expected-wins analysis with, say, UNC, UK, KU, MSU, UConn, and a few others you want to name (is IU back?). Would they look way better than us, especially over a cherry-picked period like 2005-present?

For example, take Carolina. Let's use Dean's retirement as an arbitrary marker. (I'm aware of the problem of this cherry-picking). This excludes his 1993 title but also K's in 1991 and 1992, since we're so focused on national titles. Two titles apiece. But but it also excludes our 1995 grease-fire, I guess. But, UNC has missed the tournament in 2002, 2003 and 2010. Shouldn't missing the tournament conceivably result in even greater negative points than losing to VCU? And why don't ACCT expected wins factor into the discussion? Some of the more amazing postseason feats in K's tenure were the 2000, 2003, 2005, 2009 and 2011 ACCTs. I've never seen stats on expected wins by seed in the ACCT. I do know we've won quite a few from the 2 or 3 spot.

Remind me why we're whining about the postseason record of roughly the second best college basketball coach ever? I'm slow and I need things explained to me.

Good points. This reminds of the old MLB debate about whether you'd want to be the Florida Marlins and have 2 WS titles(97 and 03) and NOTHING else or the Atlanta Braves with 1 WS title but WAY WAY WAY more overall success. It's a fascinating argument, really.

freshmanjs
01-26-2013, 09:20 AM
I agree that 2007-2009 could be considered transitional years. However, the reason I included them in the current era was because there wasn't a complete rebuild due to unprecedented circumstances, such as building a program from scratch in the 80's or having to leave the program for basically an entire season in 1995. In fact, you could argue that Coach K had to entirely rebuild Duke's brand and reputation with recruits after the '95 disaster.

If 2007 looked like a rebuild, to me, that's on the coaching staff as it's their responsibility to anticipate turnover of the roster. Further, I don't believe anyone from the 2006 roster left or transferred unexpectedly, as Sheldon, Dockery, and JJ were all seniors. It's my contention that the players the coaches were bringing in, while high-level, weren't the NBA-level talents of previous eras and the overall results in the NCAAT prove it. Now, if we could project out 5 years in the future, then certainly the 3rd era could extend beyond a 9 year term, but I believe the 3rd era has to begin in 2007.

Interesting thoughts about "eras." Why is 2007 the break point for the 3rd era though? Is it because it was a (relatively) weak team for Duke? Otherwise, couldn't you point to any other transition at which key personnel turned over? Why wouldn't 2003 be just as much of a break point with Dunleavy, Boozer, JWil all gone? I'm not sure why the JJ/Shelden teams would be grouped with 2000-02 instead of with the Paulus/McRoberts teams. JJ/Shelden actually played together with Paulus/McRoberts and never with the 2002 guys. You could also argue the era should extend through to 2011 with Nolan/Kyle leaving and that 2012 was a break point.

Kedsy
01-26-2013, 11:24 AM
I agree that 2007-2009 could be considered transitional years. However, the reason I included them in the current era was because there wasn't a complete rebuild due to unprecedented circumstances, such as building a program from scratch in the 80's or having to leave the program for basically an entire season in 1995. In fact, you could argue that Coach K had to entirely rebuild Duke's brand and reputation with recruits after the '95 disaster.

If 2007 looked like a rebuild, to me, that's on the coaching staff as it's their responsibility to anticipate turnover of the roster. Further, I don't believe anyone from the 2006 roster left or transferred unexpectedly, as Sheldon, Dockery, and JJ were all seniors. It's my contention that the players the coaches were bringing in, while high-level, weren't the NBA-level talents of previous eras and the overall results in the NCAAT prove it. Now, if we could project out 5 years in the future, then certainly the 3rd era could extend beyond a 9 year term, but I believe the 3rd era has to begin in 2007.


Interesting thoughts about "eras." Why is 2007 the break point for the 3rd era though? Is it because it was a (relatively) weak team for Duke? Otherwise, couldn't you point to any other transition at which key personnel turned over? Why wouldn't 2003 be just as much of a break point with Dunleavy, Boozer, JWil all gone? I'm not sure why the JJ/Shelden teams would be grouped with 2000-02 instead of with the Paulus/McRoberts teams. JJ/Shelden actually played together with Paulus/McRoberts and never with the 2002 guys. You could also argue the era should extend through to 2011 with Nolan/Kyle leaving and that 2012 was a break point.

Obviously this is all subjective to an extent, but let me explain my position. I have been talking about the "era" thing since before the 2009-10 season and my reasoning was basically that we seem to get into 12 year "cycles," with three building years followed by nine golden years.

The three building years start with a "down" year (1983, 1995, 2007). After that, Coach K has to basically teach the kids how to win. The 2nd year is a pretty good year, but the NCAA tournament proves too much (1984, 1996, 2008), the third year has us solidly in the top 10 (1985, 1997, 2009), but still not able to get over the NCAAT hump, and then the nine "golden years" can begin. The parallels seemed striking to me back in the summer of 2009.

Why is 2007 included with the other two? Well, for starters, if you believe in the 12 year cycle (3 followed by 9), we'd just finished our second 12 year cycle, because if 1995 started the three year cycle and 1998 was the first year of the nine year era, then that era ended in 2006. Second, the 2006-07 team was by far the youngest team Coach K had since 1982-83. Third, it was the first official season of the "one and done era." Perhaps relatedly, it was the first season where Coach K's focused recruiting strategy showed some need for revamping. He had obviously expected at least some of the huge 1997-98 freshman class to be gone in two seasons, because he brought in a large and very talented class in 1999-2000. Similarly he brought in a super (and super-large) class in 2002-03. He only brought in one or two freshman in 1998-99, 2000-01, 2001-02, 2003-04, and 2004-05, followed by a big class in 2005-06. The strategy appeared to be to bring in big, talented classes every few years and supplement with one or two very talented recruits in the middle. When Deng left early, Humphries reneged on his commitment, and Livingston jumped straight to the NBA, that strategy fell apart. The 2006-07 recruiting class joined a team with no seniors and just one recruited junior. The team started pretty well (similarly to 1995) but its inexperience and lack of upperclass talent caught up to it. To me it's almost as much of a "start over" year as 1995 (which, after all, had a pretty decent freshman class of Trajan, Wojo, and Ricky Price to aid the subsequent rebuilding process).

Put another way, the 1999-2000 freshmen had Carawell, Battier and James as upperclassmen to show them what it meant to play for Duke. The 2002-03 freshmen had several upperclassmen to show them the ropes, including Duhon, Casey Sanders, and Dahntay Jones. The 2005-06 freshmen had JJ, Shelden, Sean Dockery, and Lee Melchionni. But the 2006-07 freshmen had just DeMarcus Nelson, who had never been a starter and had been injured much of the previous season. So just like 1983 and 1995, Coach K had to teach that team how to win and it took the same three years.

Therefore, in my opinion 2007 was not just another year after a big senior class left (like 2003 or 2010 or 2011), it was a year we had to start over and revamp our whole recruiting strategy. Anyway, that's my story and I'm sticking with it.

freshmanjs
01-26-2013, 11:40 AM
Obviously this is all subjective to an extent, but let me explain my position. I have been talking about the "era" thing since before the 2009-10 season and my reasoning was basically that we seem to get into 12 year "cycles," with three building years followed by nine golden years.

The three building years start with a "down" year (1983, 1995, 2007). After that, Coach K has to basically teach the kids how to win. The 2nd year is a pretty good year, but the NCAA tournament proves too much (1984, 1996, 2008), the third year has us solidly in the top 10 (1985, 1997, 2009), but still not able to get over the NCAAT hump, and then the nine "golden years" can begin. The parallels seemed striking to me back in the summer of 2009.

Why is 2007 included with the other two? Well, for starters, if you believe in the 12 year cycle (3 followed by 9), we'd just finished our second 12 year cycle, because if 1995 started the three year cycle and 1998 was the first year of the nine year era, then that era ended in 2006. Second, the 2006-07 team was by far the youngest team Coach K had since 1982-83. Third, it was the first official season of the "one and done era." Perhaps relatedly, it was the first season where Coach K's focused recruiting strategy showed some need for revamping. He had obviously expected at least some of the huge 1997-98 freshman class to be gone in two seasons, because he brought in a large and very talented class in 1999-2000. Similarly he brought in a super (and super-large) class in 2002-03. He only brought in one or two freshman in 1998-99, 2000-01, 2001-02, 2003-04, and 2004-05, followed by a big class in 2005-06. The strategy appeared to be to bring in big, talented classes every few years and supplement with one or two very talented recruits in the middle. When Deng left early, Humphries reneged on his commitment, and Livingston jumped straight to the NBA, that strategy fell apart. The 2006-07 recruiting class joined a team with no seniors and just one recruited junior. The team started pretty well (similarly to 1995) but its inexperience and lack of upperclass talent caught up to it. To me it's almost as much of a "start over" year as 1995 (which, after all, had a pretty decent freshman class of Trajan, Wojo, and Ricky Price to aid the subsequent rebuilding process).

Put another way, the 1999-2000 freshmen had Carawell, Battier and James as upperclassmen to show them what it meant to play for Duke. The 2002-03 freshmen had several upperclassmen to show them the ropes, including Duhon, Casey Sanders, and Dahntay Jones. The 2005-06 freshmen had JJ, Shelden, Sean Dockery, and Lee Melchionni. But the 2006-07 freshmen had just DeMarcus Nelson, who had never been a starter and had been injured much of the previous season. So just like 1983 and 1995, Coach K had to teach that team how to win and it took the same three years.

Therefore, in my opinion 2007 was not just another year after a big senior class left (like 2003 or 2010 or 2011), it was a year we had to start over and revamp our whole recruiting strategy. Anyway, that's my story and I'm sticking with it.

There definitely has been a 3/9 pattern. Isn't that probably just coincidence though? Is there any reason to think that pattern would repeat into the future? In any case, I certainly hope we have 6 more excellent years in a row!

Kedsy
01-26-2013, 12:38 PM
There definitely has been a 3/9 pattern. Isn't that probably just coincidence though? Is there any reason to think that pattern would repeat into the future? In any case, I certainly hope we have 6 more excellent years in a row!

Yeah, who knows. Prior to the 2010 season I saw the pattern, noticed we'd just been through what would have been the three rebuilding years, and predicted we'd go to the Final Four in 2010. Since that did actually happen, I decided maybe there was something to it. But of course it could just be coincidence. I like to think of it the other way, though.

Wander
01-26-2013, 12:46 PM
For example, take Carolina. Let's use Dean's retirement as an arbitrary marker. (I'm aware of the problem of this cherry-picking). This excludes his 1993 title but also K's in 1991 and 1992, since we're so focused on national titles. Two titles apiece. But but it also excludes our 1995 grease-fire, I guess. But, UNC has missed the tournament in 2002, 2003 and 2010. Shouldn't missing the tournament conceivably result in even greater negative points than losing to VCU? And why don't ACCT expected wins factor into the discussion? Some of the more amazing postseason feats in K's tenure were the 2000, 2003, 2005, 2009 and 2011 ACCTs. I've never seen stats on expected wins by seed in the ACCT. I do know we've won quite a few from the 2 or 3 spot.

You are missing the point of the entire discussion. No one is saying that missing the tournament isn't worse than getting into the tournament and getting upset early. The point here is to compare a team's NCAAT record to a team's regular season record in a relative sense, not to compare programs across the board. Hell, I don't think anyone on this thread has even said that Duke isn't the single best program in the country. And not every discussion of a team or program's characteristics/weak spots/whatever is "whining."

Wander
01-26-2013, 01:00 PM
3) Duke may have a tendency to be overseeded given the program's, and K's, well earned tournament reputation

5) The NCAA's seeding system sucks and underrates less well established programs like Lehigh, VCU, etc. Frankly, both Lehigh and VCU were criminally underrated when they bounced us.


You're not the only person to suggest this, but I think these ideas are completely off. The selection does a fine job of seeding the teams (I'd say "excellent job," except that would imply selecting and seeding teams is difficult, which it's not), and seeds do a reasonably good job of reflecting the quality of a team's regular season. The only time I can ever remember Duke being overseeded was in 2007, when IMO we deserved and 8 rather than a 6. Lehigh had zero good wins last year before the tournament and played in a terrible conference. Maybe you could argue a 14 seed instead of a 15, I guess. But that's it.

I'll put forth one idea that I don't think has been mentioned so far (but I skipped over some of the recent stat posts). We have a table of each seed's expected number of wins - I wonder what the variability is for each seed though. Duke has so many 1 and 2 seeds compared to other programs - maybe the standard deviation in the average number of wins is high enough for 1 and 2 seeds that our record doesn't really fall short as significantly as it looks. This might also explain Kansas, which I think is the most comparable program in the past decade to Duke.

bedeviled
01-26-2013, 07:40 PM
More top teams lose to unranked opponents today. You know, maybe it's just plain harder to win nowadays. In the old days, maybe the outcomes were more along the lines of the chalk...thus, "performing to expectation." With increased parity, there are more "upsets," hence more perceived "underperforming" and "overperforming." Fortunately, we are nearly invariably a top seed, so the only direction the perception of our performance can go is down.
I don't have any data for the above conjecture, but I will point out that the seed-expectation chart we've been using is for the last 26 years all combined.

NSDukeFan
01-26-2013, 08:10 PM
I believe early entry into the NBA draft has made it so that the most talented teams tend to be less experienced, which increases parity.
On another note, I don't believe Duke has been over seeded based on their season's accomplishments though I could see an argument that the team has on occasion been overseeded based on talent. But that just goes with having a coach that has his team consistently play hard, well and smart (minus the occasional outlier blowout) and uses his talent more effectively than others. At least that's my theory.
I would also add that it is very difficult to exceed your seed when you are #1 and I would always rather be at the top.

Kedsy
01-26-2013, 08:38 PM
More top teams lose to unranked opponents today. You know, maybe it's just plain harder to win nowadays. In the old days, maybe the outcomes were more along the lines of the chalk...thus, "performing to expectation." With increased parity, there are more "upsets," hence more perceived "underperforming" and "overperforming." Fortunately, we are nearly invariably a top seed, so the only direction the perception of our performance can go is down.
I don't have any data for the above conjecture, but I will point out that the seed-expectation chart we've been using is for the last 26 years all combined.

That's a great point. We ought to do a similar chart for just the past 8 years and see what it looks like. If I find the time, maybe I'll try it. If anybody else has the time, I'd gladly cede the glory.

bedeviled
01-26-2013, 09:27 PM
Kedsy, I PM'd to tell you I'll do the chart, but now I'm worried that you are furiously working on it while not logged in! Anyway, I knew I should have tested my theory before posting, but I was too lazy. I am at fault, I feel guilty, and I have begun work on the chart. (You can still have the glory because I wouldn't have done it without you!)

Kedsy
01-26-2013, 10:14 PM
Kedsy, I PM'd to tell you I'll do the chart, but now I'm worried that you are furiously working on it while not logged in! Anyway, I knew I should have tested my theory before posting, but I was too lazy. I am at fault, I feel guilty, and I have begun work on the chart. (You can still have the glory because I wouldn't have done it without you!)

Don't feel guilty at all. I did start it but didn't get very far before I got pulled away. Once I got your PM, I stopped. And it was your idea so I don't want the glory.

bedeviled
01-27-2013, 03:35 AM
Well, so much for that idea! Here is the chart for average number of wins achieved by each seed in the NCAA-T, years 2004-2012 (not much different than the fist 26 years combined):


Seed
1985-2010 Avg Wins
2004-12 Avg Wins


1
3.42
3.5


2
2.43
2.38


3
1.84
2.03


4
1.46
1.41


5
1.18
1.22


6
1.22
0.88


7
0.83
0.84


8
0.65
0.66


9
0.59
0.5


10
0.65
0.59


11
0.49
0.72


12
0.52
0.56


13
0.25
0.31


14
0.17
0.09


15
0.04
0.06


16
0
0

bedeviled
01-27-2013, 03:54 AM
Since I had compiled all the raw data, I decided to take a peek at the performance of other blue blood programs (and Butler, just for kicks). Here is how other programs are performing compared to the expectations (expectations based on average number of wins for a certain NCAA-T seed):




2004-2012 NCAA-T
Butler
Florida
Arizona
Mich St
Kentucky
UCLA
Louisville
UNC
Indiana
UConn
OSU
Kansas
Michigan
Duke
Syracuse


1
Total Appearances
5
6
6
8
7
6
7
7
4
6
6
8
3
8
6


2
Average Seed
6.8
4.5
7.67
5.38
5
4.83
4.43
1.43
6.25
3.33
2.67
2
7.33
2
2.83






















3
Total Expected Wins
4.44
10.75
5.72
11.51
12.83
10.7
11.14
21.91
3.79
13.32
14.8
21.85
2.66
22.02
13.69


4
Total Wins
13
19
9
18
18
15
15
26
4
14
14
20
2
15
8






















5
Expected Wins/Appearance
0.89
1.79
0.95
1.44
1.83
1.78
1.59
3.13
0.95
2.22
2.47
2.73
0.89
2.75
2.28


6
Actual Wins/Appearance
2.6
3.17
1.5
2.25
2.57
2.5
2.14
3.71
1
2.33
2.33
2.5
0.67
1.88
1.33


7
% of Expected Wins Achieved
292.79
176.74
157.34
156.39
140.3
140.19
134.65
118.67
105.54
105.11
94.59
91.53
75.19
68.12
58.44


It's past my brain's bedtime, but a couple quick observations popped into my head.

Note that, across the 2004-12 time span, most blue bloods are outperforming the seed expectations (line 7). This suggests, conversely, that non-blue bloods are the ones weighing down the seed averages. Again, so much for parity.
Duke had the highest overall expectations for the time period (line 3), but only because UNC missed the tournament one year. If you look at expected wins per tournament appearance (line 5), UNC's is remarkably high due to their high seeding at each appearance.
I had questioned if Duke's consistent high seeding was the cause of underperforming compared to seed averages. Unfortunately, the expectations are based on seeding, so I think it is invalid to test for a correlation between seed (line 2) and ability to meet expectations (line 7). However, looking at the numbers (line 2 or line 5), it does look like blue bloods with lower seeding (and, hence, lower expectations) better outperform their seed's average. If my brain is correct in looking at this, it may affect my choices come bracket time - betting on the blue bloods who are seeded low (I'm looking at you, UCLA....and Kentucky in the NIT ;)).

bedeviled
01-27-2013, 04:00 AM
Finally, I decided to take a look at some non-blue bloods, so I chose our ACC colleagues. Here, the majority of teams do not meet the expectations of seed average. I wouldn't say that NCSU gets the most out of the tournament (I think our trophy case can attest to that), but it is interesting to see who helps boost the ACC image and who is "Clemsoning."




2004-2012 NCAA-T
NCSU
Miami
UNC
BC
Maryland
GT
VT
FSU
Duke
UVA
Wake
Clemson


1
Total Appearances
3
1
7
4
3
3
1
4
8
2
3
4


2
Average Seed
10.33
7
1.43
5.5
6
8.33
5
6.75
2
7
5
7.75



















3
Total Expected Wins
1.9
0.84
21.91
4.5
3.41
2.4
1.22
4.34
22.02
2
4.29
3.46


4
Total Wins
5
1
26
4
3
2
1
3
15
1
2
0



















5
Expected Wins/Appearance
0.63
0.84
3.13
1.13
1.14
0.8
1.22
1.09
2.75
1
1.43
0.87


6
Actual Wins/Appearance
1.67
1
3.71
1
1
0.67
1
0.75
1.875
0.5
0.67
0


7
% of Expected Wins Achieved
263.16
119.05
118.67
88.89
87.98
83.33
81.97
69.12
68.12
50
46.62
0

freshmanjs
01-27-2013, 08:22 AM
Since I had compiled all the raw data, I decided to take a peek at the performance of other blue blood programs (and Butler, just for kicks). Here is how other programs are performing compared to the expectations (expectations based on average number of wins for a certain NCAA-T seed):




2004-2012 NCAA-T
Butler
Florida
Arizona
Mich St
Kentucky
UCLA
Louisville
UNC
Indiana
UConn
OSU
Kansas
Michigan
Duke
Syracuse


1
Total Appearances
5
6
6
8
7
6
7
7
4
6
6
8
3
8
6


2
Average Seed
6.8
4.5
7.67
5.38
5
4.83
4.43
1.43
6.25
3.33
2.67
2
7.33
2
2.83






















3
Total Expected Wins
4.44
10.75
5.72
11.51
12.83
10.7
11.14
21.91
3.79
13.32
14.8
21.85
2.66
22.02
13.69


4
Total Wins
13
19
9
18
18
15
15
26
4
14
14
20
2
15
8






















5
Expected Wins/Appearance
0.89
1.79
0.95
1.44
1.83
1.78
1.59
3.13
0.95
2.22
2.47
2.73
0.89
2.75
2.28


6
Actual Wins/Appearance
2.6
3.17
1.5
2.25
2.57
2.5
2.14
3.71
1
2.33
2.33
2.5
0.67
1.88
1.33


7
% of Expected Wins Achieved
292.79
176.74
157.34
156.39
140.3
140.19
134.65
118.67
105.54
105.11
94.59
91.53
75.19
68.12
58.44


It's past my brain's bedtime, but a couple quick observations popped into my head.

Note that, across the 2004-12 time span, most blue bloods are outperforming the seed expectations (line 7). This suggests, conversely, that non-blue bloods are the ones weighing down the seed averages. Again, so much for parity.
Duke had the highest overall expectations for the time period (line 3), but only because UNC missed the tournament one year. If you look at expected wins per tournament appearance (line 5), UNC's is remarkably high due to their high seeding at each appearance.
I had questioned if Duke's consistent high seeding was the cause of underperforming compared to seed averages. Unfortunately, the expectations are based on seeding, so I think it is invalid to test for a correlation between seed (line 2) and ability to meet expectations (line 7). However, looking at the numbers (line 2 or line 5), it does look like blue bloods with lower seeding (and, hence, lower expectations) better outperform their seed's average. If my brain is correct in looking at this, it may affect my choices come bracket time - betting on the blue bloods who are seeded low (I'm looking at you, UCLA....and Kentucky in the NIT ;)).


Thanks for doing this...very cool. You included 2004, which helps Duke's numbers a bit vs. starting with 2005 or 2002. Of course, this is arbitrary and going back to 2001 would help the numbers again.

I still hold my strong view that Duke has underperformed relative to seed since 2005. Per my earlier analysis, this underperformance is quite a bit more pronounced than it was in 1997-2004. I agree the cutoff between those periods is arbitrary, that the sample size is too small to be statistically significant, and that it could just be randomness.

Recognizing that the following is subjective:

Duke #1/2 seeds without excellent point guard play with # of wins:
2012 - 0
2011 - 2
2009 - 2
2008 - 1
2006 - 2
2005 - 2
1997 - 1

Duke #1/2 seeds with excellent point guard play with # of wins:
2010 - 6
2004 - 4
2002 - 2
2001 - 6
1999 - 5
1998 - 3

i'm not sure where to put 2000....i think a freshman JWil goes in the top group, but this was the hardest one for me to categorize. 2 wins.

Mtn.Devil.91.92.01.10.15
01-27-2013, 09:12 AM
Since I had compiled all the raw data, I decided to take a peek at the performance of other blue blood programs (and Butler, just for kicks). Here is how other programs are performing compared to the expectations (expectations based on average number of wins for a certain NCAA-T seed):


I had questioned if Duke's consistent high seeding was the cause of underperforming compared to seed averages. Unfortunately, the expectations are based on seeding, so I think it is invalid to test for a correlation between seed (line 2) and ability to meet expectations (line 7). However, looking at the numbers (line 2 or line 5), it does look like blue bloods with lower seeding (and, hence, lower expectations) better outperform their seed's average. If my brain is correct in looking at this, it may affect my choices come bracket time - betting on the blue bloods who are seeded low (I'm looking at you, UCLA....and Kentucky in the NIT ;)).


First of all, thanks for pulling together all this data - it's very interesting. However, I think this quoted part is more significant than you are giving credit to. Let's say Butler is a 13 seed and is expected to win 0 games. Every win the pull off - while impressive - skews the "expectation percentage" infinitely. As a 1 or 2 seed, it's incredibly difficult to exceed expectations, because there are more opportunities for you to do so. If a 1 seed meets expectations in the first round they are rewarded (in these calculations) with another opportunity to fail to meet expectations.

On the other hand, that does make UNC's figures in this table even more impressive (having such a high average seed and an impressive percentage). Since this analysis seems to praise the Tar Heels, I shall move on to other topics. **(edit: though, as you pointed out, the absence of a certain team in the tournament for their run the NIT Final does skew the data in their direction)

To address original post and point, I think that Duke excels in early season due to stellar coaching in pre-season and excellent preparation. I think that the reason there's been a moderation of success in March (though we did happen to cut down the nets three short years ago) is the lack of "great" teams. I think due to one-and-dones, transfers, coaching changes, and many many other factors - the era of completely dominant teams is behind us. No more UNLV 1990, and probably no more Duke 1999. The mid-majors have become much better, as their experience gives them a bag advantage over teams with one and two year players, and days of elite talent playing together at elite programs for several years in a row is history. I wonder if other fan bases of these "blue blood programs" - if answering honestly - would say they feel they have underachieved lately as well? I don't mean Louisville this week, or UNC and Kentucky this year, but how do Syracuse fans and Arizona fans and Michigan fans feel about the last decade of their program?

But, I think this is fine for college basketball fans. We like the Butlers, the Gonzagas, the Creightons - just not the Lehighs. And as much as I'm cautious about expansion, I think the ACC is desperate for an influx of basketball talent. How long has it been since we had four teams in the Top 25 for several weeks? For whatever reasons independent of those stated above, the ACC has suffered a dearth of tournament success, talented players, and exceptional coaches outside of the two shades of blue.

So yes, we are the kings of November and December. I'm fine with that. I think we do tend to dominate those months on a national scale. However, we also sometimes come home with the hardware in March and April, so I'm fine with that too.

Kedsy
01-27-2013, 10:15 AM
Thanks for doing this...very cool. You included 2004, which helps Duke's numbers a bit vs. starting with 2005 or 2002. Of course, this is arbitrary and going back to 2001 would help the numbers again.

I still hold my strong view that Duke has underperformed relative to seed since 2005. Per my earlier analysis, this underperformance is quite a bit more pronounced than it was in 1997-2004. I agree the cutoff between those periods is arbitrary, that the sample size is too small to be statistically significant, and that it could just be randomness.

Recognizing that the following is subjective:

Duke #1/2 seeds without excellent point guard play with # of wins:
2012 - 0
2011 - 2
2009 - 2
2008 - 1
2006 - 2
2005 - 2
1997 - 1

Duke #1/2 seeds with excellent point guard play with # of wins:
2010 - 6
2004 - 4
2002 - 2
2001 - 6
1999 - 5
1998 - 3

i'm not sure where to put 2000....i think a freshman JWil goes in the top group, but this was the hardest one for me to categorize. 2 wins.

How can 2011 not be in the category of "excellent point guard play"? By the time the tournament rolled around we had the NBA first pick and a leading NPOY candidate who both played PG. And in 1997, our PG had 5+ apg, a 3 to 1 a/to ratio, shot nearly 40% from three and had 2.5 spg (and would win the NDPOY a year later). That's not "excellent"? As much as people like to rag Greg Paulus around here, he was 1st team all freshman in the ACC and a 2nd team All American freshman in 2006. Those three and 2000 (Jason Williams had 6.5 apg and 2.4 spg, though he did have a lot of turnovers) belong on the "excellent" list at least as much (if not more) than 1999 (Will Avery had 5.0 apg, 1.5 spg, and an a/to ratio less than 2). I think a lot of times how the team finished the season colors our later perception of the excellence (or lack thereof) of its play.

I don't want to get into another knock down/drag out with you, but I don't see how this categorization tells us very much.

freshmanjs
01-27-2013, 10:24 AM
How can 2011 not be in the category of "excellent point guard play"? By the time the tournament rolled around we had the NBA first pick and a leading NPOY candidate who both played PG. And in 1997, our PG had 5+ apg, a 3 to 1 a/to ratio, shot nearly 40% from three and had 2.5 spg (and would win the NDPOY a year later). That's not "excellent"? As much as people like to rag Greg Paulus around here, he was 1st team all freshman in the ACC and a 2nd team All American freshman in 2006. Those three and 2000 (Jason Williams had 6.5 apg and 2.4 spg, though he did have a lot of turnovers) belong on the "excellent" list at least as much (if not more) than 1999 (Will Avery had 5.0 apg, 1.5 spg, and an a/to ratio less than 2). I think a lot of times how the team finished the season colors our later perception of the excellence (or lack thereof) of its play.

I don't want to get into another knock down/drag out with you, but I don't see how this categorization tells us very much.

my mistake on 2011. absolutely should have been in the other group. just an error. i really do think the pg thing makes a difference though. look at how much better we are this year when quinn cook plays well (and compared to last year).

it's funny how any hypothesis about why duke is underperforming relative to seed must be imaginary or rooted in selection bias, but a hypothesis that tries to show that duke is actually not underperforming is worth testing (despite being at least as far off the mark, as shown by the data).

Troublemaker
01-27-2013, 10:33 AM
Here are my thoughts on this:


The thread title doesn't give us enough credit. We've got to at least be the kings of Jan, Feb and half of March, too. I seem to recall lots of ACC regular season and tournament championships (which Nov/Dec wins don't count towards ) even when we don't advance as far as we'd hope in the NCAAs. Also, as pointed out in the all the "playing to seed" posts, we've received lots of 1 seeds. Nov/Dec DO count towards achieving those, but it'd be quite a feat to receive 1 seeds by playing well for only two months.
But, yes, the playing to seed thing. We "perform poorly" there. But here's the thing. We rack up so many 1 and 2 seeds that it's statistically likely that we would "perform poorly" at playing to seed. No other program receives those high seeds at the rate that we do (because of our Dec thru mid-March prowess, as mentioned). In the past 15 years, we've been a one or two seed 13 times, including being a one seed 10 times, including a streak of 8 one seeds in 9 years. Incredible. If any other program received one or two seeds at the rate that we do, especially all the one seeds, that program would likely "underperform," too. Why? Because the odds are typically against any one team winning 3 or 4 consecutive games against quality competition.
Check the Vegas odds after the NCAA tournament brackets are revealed. The 1 seeds are almost always "plus money" to reach the Final Four (meaning they are regarded as less than 50/50 shots to do so). I've seen many a one seed fall into the +150 to +250 range to reach the Final Four (meaning they are given a 29% to 40% shot of doing it [and keep in mind, Vegas has to make money, so they will overestimate the team's chances]). To put it simply, it's much tougher to "play to seed" when you receive mostly ones and some twos.
NSDukefan stated above that Duke has sometimes been overseeded relative to talent. I agree with this. We tend to overachieve in the regular season relative to our talent. Why? As others have mentioned, Coach K has his teams reach their ceilings much sooner than other coaches. He is the quickest coach to properly tweak his offense and defense to match the talent available in any particular season. Also, I really do believe as a program we play harder game-in and game-out than other programs. We just put so much energy, effort, and passion into each and every game (with some exceptions, of course). In fact, I would accept theories that we could do a better job managing our energy during a season. That maybe we'd be fresher for March if we didn't play so hard every single game. But that's just a theory, and I would think the coaching staff could scale back practices to offset the amount of energy our guys expend each game.
Getting back to the point. Yes, we peak early, i.e. we reach our ceiling early. But when people say we peak early, almost everyone is imagining a negative parabola, right? You're picturing this in your head: http://sprott.physics.wisc.edu/fractals/bookfigs/fig1-03.gif , meaning we reach our ceiling and then we start playing below our ceiling. I, on the other hand, believe our seasons generally look like a fast-rising plateau curve: http://www.devchakraborty.com/images/FROC%20Curve.jpg , meaning we reach our ceiling early and then we MAINTAIN it. But, because other teams are hitting their ceilings later in the season, it seems like, relative to them, that we are slipping from peak when, in fact, we are just maintaining something we had achieved earlier than them. Generally speaking, I would say we plateau more than we parabola, but it always feels like a parabola because everyone else hits their ceiling later.
If a team parabolas because of injury, I don't find fault with coach/program, and we've had bad luck with injuries recently.

bedeviled
01-27-2013, 10:50 AM
You included 2004, which helps Duke's numbers a bit vs. starting with 2005 or 2002.OMG!! I mislabeled those three charts. The data IS from 2005-2012.Sorry about that. It was late, and I was seduced by the '2004' in the 2004-2005 season. The data, though, is from the last 8 years. Sorry.

Troublemaker
01-27-2013, 10:55 AM
One more thing. I agree with Kedsy that, health permitting, we're about to enter another golden age of NCAA tourney success for Duke (although I don't necessarily buy the all-too-neat 3/9 thing).

Our recruiting is in as good a shape as it's been in a long time. The Olympics are finally paying off in that area, and there seems to be more top-level recruits interested in Duke and interested in becoming well-rounded people (my theory on that: the Obama effect).

What I'm saying is, we're going to cease the overachieving/overseeding relative-to-talent phenomenon that we've experienced the past few years. When we get 1 seeds from now on, it'll be because of pure nasty talent, not just from playing hard and peaking early.

In summary: if we get Ryan Kelly back at a reasonable time for him to re-integrate and re-gain conditioning, we will three-peat. I had always thought it was going to be the second championship that was going to be the most difficult, but with all the injuries this season (Seth, Ryan, and now even Mason's left thumb [that's our Big 3!]), man, it's been rough for our guys this season. If we can just be healthy for once going into the tournament....

(Yes, 3-peat!!!).

freshmanjs
01-27-2013, 11:04 AM
How can 2011 not be in the category of "excellent point guard play"? By the time the tournament rolled around we had the NBA first pick and a leading NPOY candidate who both played PG. And in 1997, our PG had 5+ apg, a 3 to 1 a/to ratio, shot nearly 40% from three and had 2.5 spg (and would win the NDPOY a year later). That's not "excellent"? As much as people like to rag Greg Paulus around here, he was 1st team all freshman in the ACC and a 2nd team All American freshman in 2006. Those three and 2000 (Jason Williams had 6.5 apg and 2.4 spg, though he did have a lot of turnovers) belong on the "excellent" list at least as much (if not more) than 1999 (Will Avery had 5.0 apg, 1.5 spg, and an a/to ratio less than 2). I think a lot of times how the team finished the season colors our later perception of the excellence (or lack thereof) of its play.

I don't want to get into another knock down/drag out with you, but I don't see how this categorization tells us very much.

we are in the realm of subjective again, but i don't agree at all that greg paulus was as good of a pg as will avery. evidence for my view includes their relative nba draft positions and the fact that paulus did not even start at pg at duke at the end of his career, despite that fact that we had no one else who was ready to play the position. paulus did have good assist numbers when he played with jj, but that does prove he was a great pg in my view.

also, you stated upthread that you knew before the ncaat that the 2005 and 2008 teams weren't very good and, so, not much should have been expected of those teams. i could pull out stats to show that those teams were excellent. that wouldn't make it so, just like you posting assist numbers does not prove that paulus was a better player than avery.

Kedsy
01-27-2013, 05:57 PM
my mistake on 2011. absolutely should have been in the other group. just an error. i really do think the pg thing makes a difference though. look at how much better we are this year when quinn cook plays well (and compared to last year).

College basketball is a guards' game. Of course it's better to have a good PG than to not have a good PG.


it's funny how any hypothesis about why duke is underperforming relative to seed must be imaginary or rooted in selection bias, but a hypothesis that tries to show that duke is actually not underperforming is worth testing (despite being at least as far off the mark, as shown by the data).

When you propose a subjective test and then apply your criteria inconsistently, you should expect doubters.


we are in the realm of subjective again, but i don't agree at all that greg paulus was as good of a pg as will avery. evidence for my view includes their relative nba draft positions and the fact that paulus did not even start at pg at duke at the end of his career, despite that fact that we had no one else who was ready to play the position. paulus did have good assist numbers when he played with jj, but that does prove he was a great pg in my view.

Jon Scheyer didn't get drafted either, but he ended up on your "excellent point guard play" list. So I'm not sure how draft position plays into this.

Will Avery got drafted on potential and unfortunately never reached that potential and never even got a second contract. At Duke he was a really quick guard who played good but not outstanding defense and turned the ball over more than he should have. Greg Paulus was at best an adequate defender who despite that got a lot of steals. But in 2005-06 he also got a lot of assists and was voted to the 2nd team All American freshman team and the 1st team All ACC freshman team. His later struggles shouldn't take away that he was a pretty good PG in 2005-06. Perhaps not "excellent," but all I said was he was at least as excellent as Avery. Meaning, in my opinion, 2006 and 1999 should be on the same side of the ledger, whichever side that might be.


also, you stated upthread that you knew before the ncaat that the 2005 and 2008 teams weren't very good and, so, not much should have been expected of those teams. i could pull out stats to show that those teams were excellent. that wouldn't make it so, just like you posting assist numbers does not prove that paulus was a better player than avery.

Actually, what I stated upthread (in a conversation about expectations) was I personally didn't expect very much from the 2005 and 2008 teams, which is different from saying they weren't very good. And I have never said Paulus was a better player than Avery. I merely argued that putting 2006 on the "not excellent" list and 1999 on the "excellent" list appears to be an inconsistent application of whatever criteria you're using.

bedeviled
01-29-2013, 03:30 AM
I think this quoted part is more significant than you are giving credit to.Word. I included the last 2 charts for fun. Maybe I shouldn’t have put them in a thread that tempts us to conclude more than is there. There are a lot of half-truths and nuances in viewing “expectation” as based on average wins for a given seed. However, as evidenced in this thread, it may be a good representation of people’s subjective sense of how a team performs. Thus, I find it interesting. In case anyone is contemplating this manner of data manipulation, I’ll say a few words based on my present (admittedly often confused) thoughts. If you are not interested, then you’ll need to skip ahead a few posts; this could get tedious (I actually intended just to address freshmanjs, but it seems ridiculous to PM all this stuff). The main topics I want to address are:

Part I: It is more difficult to achieve expectations as a 1- or 2-seed, which makes Duke look worse due to Duke’s typically high seeding.
Part II: It is more difficult to meet expectations when a team has repeated high-seeded tournament appearances, thereby also making Duke appear worse.
Part III: Lower seeded teams get a disproportionate boost in expectation by winning a small number of games while not punished in an equal manner for losing when expected to do so. And the converse, higher seeded teams get a disproportionate decrease in expectation by losing an early game while not being rewarded in an equal manner for winning a game when expected to do so.

I’ve tried to reason out the rationales, but certainly could have made some mistakes or bad assumptions I didn’t realize I was making. Shout out any errors you find.


Part 1:
Hard to achieve expectation as a 1- or 2-seed
I’m talking about the “expectation” as termed recently in this thread – the expectation that a team, with a specific seeding, should achieve the same number of wins as the average number of wins by like-seeded teams. While some will immediately dismiss this measure as poppycock, I think it is an important one to consider. I think it might be a good surrogate variable to describe viewer’s subjective expectations. We have a feeling that 1-seeds should win and 13-seeds should lose. While this feeling is based on real experience, our brain probably does something akin to averaging out our experience to provide us with an expectation. I don’t know; it seems plausible to me. This data point is precisely that – an average of real life experience (historical tournament results) of how each seed has performed.

Since the “expectation” we are talking about is really just the seed’s average number of wins, the %achievement above expectation is actually equal to the %achievement below expectation for each seed. Thus, across the seed’s field, there is as much overperformance as there is underperformance. Consequently, there is the perception, both in basic data presented during tourney time and in personal gut/intuition, that expectation based on averages is justified and valid. However, what a seed field does as a group is not necessarily true for an individual team. A deep run by an 8-seed Butler may skew the average for 8-seeds so that 1. the seed’s number of games won above average does, indeed, equal the number of games lost below the average, but 2. the likelihood that any one team meets the average games won is not 50%.

Example for illustration: Suppose that, in the 32 teams seeded X from 2005-2012, six of those teams won the Championship (winning 6 games apiece), but the other 26 teams failed to win their opening game. The average wins for X-seeds would be 1.125 (36 games won / 32 seed appearances). Thus, according to this measure, you “expect” each X-seed team to win 1.125 games. Those that win 0 or 1 games are labeled ‘underachievers’ and those that win 2 or more games are ‘overachievers.’ The teams that won 6 games have achieved 533.33% of expectation (6/1.125) while those that lost their opening game have %achievement of 0% (0/1.125). Since, expectation is 100%, the winners overachieved expectation by 433.33% while the losers underachieved expectation by 100%. At first glance, it appears that the teams are rewarded more handsomely for wins than they are punished for losses (433 vs 100). Mathematically, though, the performances must balance out based on the odds. Accordingly,
433.33 * 6 (overachievement multiplied by the number of overachieving teams) = 2600 100 * 26 (underachievement multiplied by the number of underachieving teams) = 2600. BUT, VERY IMPORTANTLY, it is incorrect to think that any team that gets an X-seed should win 1.125 games. In fact, the likelihood is much greater that an X-seed doesn’t win any games at all!! Unfortunately, this way of looking at the data labels 81.25% (26/32) of the X-seeds as ‘underachievers’ when, here, they are performing the most common way for an X-seed to perform.

Thus, we see that the use of seed win averages as an expectation of how many games a single team should win is problematic.

The fluctuation in performance of the seed field can make it “harder” or “easier” for a given team to meet such an expectation. It is less likely (harder) for a given team to meet expectations if that seed’s average is skewed by outstanding individual performances (such as Butler’s deep run). 1-seeds are the most likely seed to advance and win championships, so the 1-seed group pretty much always has teams running past their 3.5 game average. Indeed, when we think about “special” teams with the mentality to make runs deep into the tournament, they are more likely to be high-seeded teams – particularly the 1-Seed, which has won 6 out of the last 8 Championships. The inspirational low-seed runs, on the other hand, are more sporadic and varied in their distribution (Butler (8), G Mason (11), VCU (11), Ohio (13), Davidson (10)).

However, the reverse is also true: one team’s surprisingly early exit will skew the seed’s average lower, resulting in a greater likelihood that any given team of that seed will supercede the seed’s average. As lower seeds aren’t expected to win as many games, there is more opportunity for high seeds to skew the data this way, making it more likely (easier) to meet expectation.

Thankfully, this is historical data, so we can actually look at look at how these factors played out in real life. It turns out that, for this data set, 62.5% of 1-Seeds performed below the 1-Seed average, thus making 1-Seeds seem like underachievers.


% of Teams “Underachieving” Their Seed-Based Average


Seed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15


%
62.5
53.1
68.8
59.4
65.6
43.8
40.6
46.9
53.1
59.4
56.3
62.5
75
90.6
93.8


**It looks like the 2-seed data was dramatically skewed from the general trend by the Duke & Missouri losses in addition to having no Champions during this timeframe. Also of note, I wonder if the effective cap on games (Champions can’t continue adding more wins to the end of their run) hinders the top seed’s percentages from even better mirroring of the bottoms seeds’ in this U-shaped trend.

Anyway, there you have it – the “expectation” is a moving target and more complex than “it’s an average, so half the teams do well and half don’t.” Most importantly, you can’t actually expect the expectation. In fact, in this historical sample, high seed teams are more likely to have less wins than the seed’s average than more. Nonetheless, this might be how our mind thinks, so it is important to understand our inner expectations.

bedeviled
01-29-2013, 03:33 AM
Part 2:
More difficult to achieve expectations due to repeated high-seed appearances

At the end of the last post, I charted the percent of teams in a given seed that appear to underachieve according to the expectation based on average seed wins. Because none of the seeds’ win averages calculated out to integers, each tournament appearance can be classified as “underachieving” or “overachieving” (making mathematic life easier). Thus, if 62.5% of 1-seeds are not meeting the 1-seed win average, then 37.5% are exceeding the average. Hence, we see that there is a greater likelihood that any random 1-seed will appear to underperform as opposed to overperform. Moreover, by nature of event odds, the more you play, the more certain it is that you will live up the billing of “underachiever.” If we calculate the odds for additional appearances, we get the following (disclaimer: I’m a lover, not a mathematician):


2 Appearances (How often do you look bad (“underachieve”)?)


2/2 Appearances
1/2 Appearances
0/2 Appearances


39%
47%
14%


So, now the likelihood of looking bad is double that of looking good.


3 Appearances (How often do you look bad?)


3/3 Appearances
2/3 Appearances
1/3 Appearances
0/3 Appearances


24%
44%
26%
5%


Well, you aren’t coming out looking gleaming after 3 appearances! Just try to make it respectable; the combined “underachievement” percentage has increased to 68%. Notice that the odds of totally floundering are basically the same as looking good once.
I’ll post one more iteration since Duke was a 1-seed four times during the period.


4 Appearances (How often do you look bad?)


4/4 Appearances
3/4 Appearances
2/4 Appearances
1/4 Appearances
0/4 Appearances


15%
37%
33%
13%
2%


Okay, now take a look at how Duke did and compare it to the chart.

Of course, those weren’t Duke’s only appearances during the timeframe. Unfortunately, the combinatorial odds get overwhelming when looking at 8 games with 3 different seeds and probabilities. I’m not going to list out the different possibilities of performance records, but my best efforts calculate the most likely single scenario for the seeds Duke had as:
Underperforming 3/4 times as 1-seed, 2/3 times as 2-seed, and overperforming as the 6-seed. The chances of that scenario were a whopping 8.2%. Duke’s own performance, using this criteria, was a 2.4% likelihood occurrence. Throatybeard pointed out that a basket here or there would have pushed us to defying expectations, which is interesting given the likelihood of looking bad.

What’s in store for the future? You could look at this:


5 Appearances (How often do you look bad?)


5/5 Appearances
4/5 Appearances
3/5 Appearances
2/5 Appearances
1/5 Appearances
0/5 Appearances


10%
29%
34%
21%
6%
1%


and bemoan that the combined “overall underachiever” status categories now comprise a glorious 73%! Or, you could realize that the past is the past and does not affect the odds of our beating our next tourney opponent. I still hope for a one seed this year and every year thereafter. Be aware, though, that “expectations” may need to be challenged.

bedeviled
01-29-2013, 03:35 AM
Part 3:
Low seeds get a disproportionate boost from expectations while high seeds get a disproportionate punishment
Once again, I’m talking about the expectations for a given seeded team to achieve the same number of wins as the average number of wins for that seed. While the last couple posts have pointed out some problems using this approach, I actually think it is still worthwhile to consider and use. Why? Because I suspect it is akin to our real-life personal, subjective expectations. So, maybe such data points and massage can’t clearly tell you about how Duke really performed, but they may help in understanding the perceptions of how Duke performed (both your own and the sports-entertainment community’s).

As an alternative method, it is also reasonable to consider each win as equal, no matter the teams or round – a win is a win. In that sense, even if it lost in the second round, a 1-seeded Duke has clearly outperformed 15-seeded NE Bumble State who lost in the first round. Duke was clearly better in the tournament than they were. BUT, that’s not how we usually operate in this scenario. Instead, we feel disappointment. Perception-wise, Duke did not fulfill its “potential” while NEBS did. Our tendency is to weight wins according to our sense of things. If both Duke and NEBS win in the first round, it is an understandable viewpoint to see NEBS’ victory as more of an accomplishment than Duke’s first round victory. It’s the way we work.

So, is the weighting we ascribe to wins and losses disproportionate? Let’s go to the data to look at the boost/decrement in the % of expectation achieved for wins and losses. First, I’ll address whether the low-seed team’s boost is disproportionate to its punishment (and high-seed vice versa), and this will kind of expand to the whole tournament field. In my Part 1 post, I used an example and purposely included information on how the gain in %achievement (of expected wins) equals the loss for a given seed. As another simple example:
Consider a team that is expected to win 1 game. If they win 1 game, they achieve 100% (1/1) of expectation. If they win 2 games, they achieve 200% (2/1), but achieve 0% (0/100) if they lose the first game. So, a game won over the expected results in a achievement gain of 100% (200-100). This is balanced by a loss below expectation, which results in an achievement loss of 100% (100-0). This is the predominant trend of the data – every game won over the expected is balanced by the same number of corresponding losses under the expectation. Seems pretty fair, right? BUT, what about game 3 in this example?!? That’s just playing with house money, isn’t it? The corresponding loss would have to happen before game 1…and there is no game –1!

Indeed, there are significant barriers in the game-by-game-balancing pattern. Seeds 13, 14, and 15 (seed 16 data is mathematically invalid due to having a 0 win average) get MORE than 100% boost in perceived achievement per game despite only being able to lose 100% of expectation in the first round (15-seeds achieve a staggering 1600% of expectation if they win one game! – yet that team can only lose 100% of achievement at most). On the other end of the expectation spectrum, 1-seeds can never actually gain an additional 100% of expectation despite being able to lose 100% of expectation if they lost in the first round. This is because they are already expected to win 3.5 games while the tournament caps out at 6 games (6/3.5 = 171%). Yet, if a top-seed loses early, they lose the achievement points from all the subsequent games they were expected to compete in...so much so that even when they are winning the opening games, they still technically have a <100% achievement rate. What gives?! How can this possibly be fair? Low seeds get huge boosts which they can’t back up with losses! Top seeds can lose everything with an early loss but can’t make an equal gain even by winning the Championship!

If it is so unequal, how does it work out to make nice, pretty averages for our expectations? Well, as I said, there is some balance on a game-by-game basis for a particular team. What ultimately balances the numbers, though, is the play across the entire seed field. Take, for instance, the 15-seed that achieves 1600% with a first round win (an increase of 1500% from the 100% expectation). These values are based on the odds, a consequence of the fact that there are only 2 teams who have actually completed this task (shhhhhh!). Thus, together, those two teams excelled 3000% above expectation. This is balanced by the other 30 15-seed teams that did not achieve the expected 0.06 wins. Each of those 30 teams lost 100% of their expectation, for a total loss of 3000%. So, you see, the whole seed field picks up the slack so that the numbers work out perfectly. Unfortunately, single teams are the beneficiary or brunt of these number shifts….which, ultimately, are perceived achievement shifts. Again, we see that perceived achievement based on meeting the seed average is a complicated matter that depends as much on the other teams and math as on the performance of the team being examined.

You may have already noticed another inequality in the ability to boost/lose perceived achievement: in addition to meeting barriers regarding how much a low-seed can lose and how much a high-seed can win, the value of each win/loss differs according to seed. Lower-seeded teams have a greater change in achievement percentage for each game played. Compare the earlier example (team is expected to win (1), but wins (0 or 2) for loss/gain of 100%) to a team expected to win (3) but wins (2 or 4) (loss/gain of just 33%). The difference is due to expectation. Is this fair? Should a high-seed team have to put together numerous wins to reach the same perceived achievement that a low-seed team gets with a single win?

Well, this is just one way to look at performance. Perhaps other ways, like taking each win at face value, can give a different perspective on outcomes. But, if you are conscientious about assumptions and expectations, I think this is an extremely interesting tactic. For specific purposes, I think these disproportionate limits and boosts/deficits in perceived achievement are, not only fair, but also very appropriate. First, the numbers are based on the historical data. This is not conjectured expectations or wishes. The data is what really happened. It balances out this way because of the frequencies that events actually occurred. Only two 15-seed teams have won. Given the unlikelihood of a 15-seed winning, I think they did way-overachieve based on expectations. And, I think it’s reasonable that the perception of Duke’s achievement took a big hit due to last year. Was Duke a good team? Yes. Were they as good as other teams that won in the tournament? Yes. Were they as good as teams that were “overachievers?” Most certainly so. However, 93.75% of 2-seeds won their opening round.



% of Teams winning a game in each round
2005-2012


Seed
Rnd64
Rnd32
Sweet16
Elite8
Final4
Championship


1
100
93.75
71.875
37.5
28.125
18.75


2
93.75
65.625
46.875
21.875
9.375
0


3
90.625
59.375
31.25
9.375
6.25
6.25


4
75
40.625
12.5
12.5
0
0


5
59.375
34.375
12.5
9.375
3.125
0


6
56.25
25
6.25
0
0
0


7
59.375
18.75
6.25
0
0
0


8
53.125
3.125
3.125
3.125
3.125
0


9
46.875
3.125
0
0
0
0


10
40.625
15.625
3.125
0
0
0


11
43.75
15.625
6.25
6.25
0
0


12
37.5
18.75
0
0
0
0


13
25
6.25
0
0
0
0


14
9.375
0
0
0
0
0


15
6.25
0
0
0
0
0


16
0
0
0
0
0
0


The numbers are what they are. There is some validity to the unequal treatment because it is based on the odds of the actual outcomes. And, to me, I think it was reasonable to have an expectation that Duke should win. Secondly, important to me, this type of analysis makes some sense gut-wise. Averaging out past experiences may be how we develop our subjective sense of achievement. And, in this mode of thinking, I tend to think an 8-seed Butler has achieved more in a 3rd round win than a 1-seed Kansas, specifically because I do compare results to internal expectations.

This type of data set analysis emphasizes our internal expectations – low seeds have more achievement riding on a per-game basis while top seeds, due to the dependent nature of rounds, must string together wins to match our expectations. It’s a fun way to measure the shock-factor of results. And, there is some validity behind it, too, as the numbers are calculated from historical outcome. However, one must realize that the interpretation is based on expectations which are not as solid as we assume them to be (see Parts I and II also). Basing expectation on seed averages can be misleading for reasons beyond subjective seeding. And, it is only one way of looking at the data. It is not Truth. On the contrary, we have seen that there are many nuances involved with understanding what the data set really means.

Kedsy
01-29-2013, 12:34 PM
Thanks, bedeviled. Really interesting analysis and obviously a LOT of thought and work put into it. I'm giving you a big round of applause from my desk chair.

It's fascinating that the most likely result from having four #1 seeds is to "underachieve" three times and "overachieve" once (exactly what we've done between 2005 and 2012). Also, it makes our period between 1986 and 1994 seem that much more amazing, as (regardless of seed) we overachieved our seed expectation 8 times in 9 years.

Billy Dat
02-05-2013, 03:45 PM
Today's "Pat Forde's Take on Duke" thread, which is thematically similar to this one, made me think of an added wrinkle that I am not sure was explored in this thread - one that is harder to quantify but may be part of the trend.

On the twice-weekly ESPN College Basketball podcast featuring Andy Katz and Seth Greenberg, they recently interviewed Gonzaga HC Mark Few. Few was discussing the reality of being every WCC team's Superbowl because of the stature of the Gonzaga program and the potential for the win to improve the opponents at-large NCAA chances. Few said it takes a huge toll on the players emotions to "get up" that high for every game. Greenberg correctly pointed out that his teams never had to play that role, but that he respected the burden that it placed on teams like Gonzaga and Duke. He said that it was very hard to play that way and not end the year emotionally spent.

When do we think the full impact of Duke being every other team's Super Bowl, especially the level of hostility and anger from opposing fan bases and even casual fans, really took hold? K always says that the tide turned in 1992, but I think the graduations of Laettner and Hurley, and the dip the program took from 1995-1997, kept things neutral for a while. I feel like the start was the 2001 title team with Billy Packer and Gary Williams, let alone the entire Maryland fanbase, fanning the flames. Then, once JJ arrived, it fully flowered and really hasn't stopped. That kind of dovetails with the "Kings of Nov/Dec" era identified at the start of this post.

Maybe Duke "under performs" in the NCAAs because, since roughly 2001, there are more people then ever focusing their collective negative energy trying to bring about that very outcome.