great article in chronicle

[uh_no]

for those of you who might not check the chronicle frequently, there was what I consider a very well done article about the 'tired legs' myth

http://dukechronicle.com/article/duke-does-decline-objectively-speaking

the center of the argument is that

1, there is a clear trend of decreasing performance as the season progresses
2. this trend can be measured each season
3. there is no correllation between the amount of minutes starters play per game and the decreasing performance, indicating that the decrease in performance is likely due to some other variable rather than 'tired legs'

[Kedsy]

for those of you who might not check the chronicle frequently, there was what I consider a very well done article about the 'tired legs' myth

http://dukechronicle.com/article/duke-does-decline-objectively-speaking

the center of the argument is that

1, there is a clear trend of decreasing performance as the season progresses
2. this trend can be measured each season
3. there is no correllation between the amount of minutes starters play per game and the decreasing performance, indicating that the decrease in performance is likely due to some other variable rather than 'tired legs'

Interesting article. He could have proved his point even more by analyzing Duke's performance during the "good" years. I haven't run the offensive efficiency statistics for 1986 to 1994, or 1998 to 2004, and I'm not even sure if they're available, but based on W/L records Duke almost always "fades" in the last x games prior to the postseason tournaments. Some of those Final Four teams were truly outstanding, but many of them were just really good (similar to Duke's teams in recent years). The only significant difference is their W/L record in the NCAA tourney.

Too many people start with the fact that a team lost its last game and thus posit as an undeniable truth that therefore something was wrong with that team. And once there, they can draw all sorts of unsupportable conclusions. They can identify anything they didn't like about the team and say they lost because of xx. If you argue with them, they essentially respond with "scoreboard," except that was their original premise so logically it can't be used to prove itself. That kind of illogic drives me crazy, and it's good to see a writer for the Chronicle taking that sort of thing head on.

[Starter]

Nah, there's nothing really here. He compares Duke players to NBA guys who play big minutes -- well no kidding, NBA players are older and stronger by design than college players. (He brings up Tyreke Evans, who's clearly an outlier as a physical beast as he has not hit the rookie wall, though he still has time to do so. I hope not, for the sake of my fantasy team)

Plus, he presents data to show that Duke generally declines over the course of a season, and yet he denies the most obvious reason without explaining why. He refers to Redick's high minute total as "anecdotally," when it's anything but. And he makes it seem like the issue here is running, when the true issue is the physical nature of the game and the amount of pounding players take, which only seems to be getting more so as the ACC turns into a more physical league.

Just because you claim you're right about something doesn't mean you are.

[uh_no]

well what makes it interesting is that he uses data on HOW we won games, not just whether we won them, which makes a much better point that 'you went out early in the tournament'

[BD80]

for those of you who might not check the chronicle frequently, there was what I consider a very well done article about the 'tired legs' myth

http://dukechronicle.com/article/duke-does-decline-objectively-speaking

the center of the argument is that

1, there is a clear trend of decreasing performance as the season progresses
2. this trend can be measured each season
3. there is no correllation between the amount of minutes starters play per game and the decreasing performance, indicating that the decrease in performance is likely due to some other variable rather than 'tired legs'

A "trend" SINCE the 2004 season?

This "trend" includes only the ACC schedule for five seasons, which were years when unc generally had the best teams that unc has had for decades, including two national championships. We always play carolina at the end of the first half of the season and at the end of the season. Thus, the "trend" of decreasing performance may be skewed just by playing the toughest team on our schedule at the middle and at the end of the ACC schedule.

[El_Diablo]

We play UNC in the last game of the year every year. For the last few years, this has obviously been a major drag on our efficiency rating, and it always occurs at the same spot in the schedule. Switch the numbers from the first ACC game and the last ACC game and see what the trend looks like (nevermind the fact that the other UNC game is late in the season as well).

Then we play in the ACC tournament, which further skews the numbers for two reasons: (1) we play increasingly better teams as the rounds progress, and (2) unless we win the whole thing, we always end with a loss--meaning our efficiency is negative whereas it necessarily was in positive territory to get to that point.

[ice-9]

Plus, he presents data to show that Duke generally declines over the course of a season, and yet he denies the most obvious reason without explaining why. He refers to Redick's high minute total as "anecdotally," when it's anything but. And he makes it seem like the issue here is running, when the true issue is the physical nature of the game and the amount of pounding players take, which only seems to be getting more so as the ACC turns into a more physical league.

I thought he did explain why -- he compared players' heavy minutes with the decline in efficiency and found no relationship. Ergo, his conclusion that something else is driving the decline.

But what? Could it simply be that Duke teams overachieve and then play to its base capabilities once other teams got it together? This seems too romantic of an explanation to me...

Perhaps it has something to do with scheduling? That rival games tend to be scheduled later rather than earlier?

What would be really interesting is to apply the same process to successful Duke seasons...e.g. Duke's Final Four teams. Did those teams experience decline during the ACC season?

[Hermy-own]

Nah, there's nothing really here. He compares Duke players to NBA guys who play big minutes -- well no kidding, NBA players are older and stronger by design than college players. (He brings up Tyreke Evans, who's clearly an outlier as a physical beast as he has not hit the rookie wall, though he still has time to do so. I hope not, for the sake of my fantasy team)

Plus, he presents data to show that Duke generally declines over the course of a season, and yet he denies the most obvious reason without explaining why. He refers to Redick's high minute total as "anecdotally," when it's anything but. And he makes it seem like the issue here is running, when the true issue is the physical nature of the game and the amount of pounding players take, which only seems to be getting more so as the ACC turns into a more physical league.

Just because you claim you're right about something doesn't mean you are.

No, there is lots in that article. I was very impressed by it. He does compare star minutes vs declining performance, and found no correlation. So he does prove his point that tired legs aren't the problem - or at least he introduces evidence against that theory, even if the evidence isn't conclusive.

More importantly though, is why Duke declines late in the season. That is a really interesting subject. I would love a statistical comparison to
a: Other elite programs in recent years
b: Great (Final 4) Duke programs of the past
and see if their overall efficiency declines.

My personal theory as to why our efficiency has declined is that our offense has a tendency to stagnate in the same players and the same plays as the year wears on. In the beginning of the year more players are involved in the offense making it more flexible & unpredictable. As Coach K tries to tighten the ship and reduce mistakes, it loses a bit of creativity. On the other hand, there is no data to back up this theory, and I really hope someone can come up with a better idea.

[sagegrouse]

The Chronicle (of which I was a proud staffer/editor for four years) is conducting an autopsy of Duke teams that fall off at the end of the season and how that happens.

Being an optimist, it looks to me that this team is improving steadily through the season. Therefore, the analysis is irrelevant.

I say, lets don't give them any data points this year.

sagegrouse

[uh_no]

We play UNC in the last game of the year every year. For the last few years, this has obviously been a major drag on our efficiency rating, and it always occurs at the same spot in the schedule. Switch the numbers from the first ACC game and the last ACC game and see what the trend looks like (nevermind the fact that the other UNC game is late in the season as well).

Then we play in the ACC tournament, which further skews the numbers for two reasons: (1) we play increasingly better teams as the rounds progress, and (2) unless we win the whole thing, we always end with a loss--meaning our efficiency is negative whereas it necessarily was in positive territory to get to that point.

the article is not arguing whether we suffer a decline or not, it is CLEAR that the decline is there, and scheduling is definitely a possible explanation, and the article doesn't presuppose its not....it simply says that there is no correlation between minutes played and late season performance

he also points out that other top teams, like UNC and MSU, who also play in conference tournaments and also play in the ncaat, much like us, do not suffer from a similar decline

[El_Diablo]

the article is not arguing whether we suffer a decline or not, it is CLEAR that the decline is there, and scheduling is definitely a possible explanation, and the article doesn't presuppose its not....it simply says that there is no correlation between minutes played and late season performance

he also points out that other top teams, like UNC and MSU, who also play in conference tournaments and also play in the ncaat, much like us, do not suffer from a similar decline

Yeah, there's a decline, but he didn't explain why. I threw in my 2 cents, citing quality of opponents. Referencing UNC and MSU doesn't really negate the fact that we play tougher opponents later in the season.

And speaking of UNC...anyone care to run the numbers on UNC's efficiency ratings this year? I bet there's a negative trend...

[The Gordog]

No, there is lots in that article. I was very impressed by it. He does compare star minutes vs declining performance, and found no correlation. So he does prove his point that tired legs aren't the problem - or at least he introduces evidence against that theory, even if the evidence isn't conclusive.

More importantly though, is why Duke declines late in the season. That is a really interesting subject. I would love a statistical comparison to
a: Other elite programs in recent years
b: Great (Final 4) Duke programs of the past
and see if their overall efficiency declines.

My personal theory as to why our efficiency has declined is that our offense has a tendency to stagnate in the same players and the same plays as the year wears on. In the beginning of the year more players are involved in the offense making it more flexible & unpredictable. As Coach K tries to tighten the ship and reduce mistakes, it loses a bit of creativity. On the other hand, there is no data to back up this theory, and I really hope someone can come up with a better idea.

My personal theory is that the fact that Duke is on TV more than anyone, more hated than anyone, more analyzed than anyone, means that any coach playing us has the analysis of thousands to help him find our weaknesses.

I also think K is better at utilizing the talent he has so our our teams come together quickly, whereas other teams take all year to jell.

[91_92_01_10_15]

Plus, he presents data to show that Duke generally declines over the course of a season, and yet he denies the most obvious reason without explaining why.

I don't think he needs to explain 'why'. His article set out to debunk the assertion that Duke's performance declined over the courses of the last few seasons, at which he failed.

He did, however, present some pretty convincing evidence that the decline did not correlate with minutes played by key players. He does not have to say why or why not; he simply says what is. Sounds like good, objective reporting to me.

[loran16]

I've always felt the decline is because of how we play.

Duke always starts more prepared than anyone else. Are players are closer to their top potential at the start of the season. We have less room for growth.

Then you have teams like Louisville, which has, under Pitino, typically done pretty poorly at the start only to finish really strong (see last year, for example).

The end result is that other teams get better while duke remains mostly the same.

(Is this a problem? Maybe, Maybe not. If you have a top tier team, it won't be. If your team isn't that great to start, well that's another story).

[roywhite]

The Chronicle (of which I was a proud staffer/editor for four years) is conducting an autopsy of Duke teams that fall off at the end of the season and how that happens.

Being an optimist, it looks to me that this team is improving steadily through the season. Therefore, the analysis is irrelevant.

I say, lets don't give them any data points this year.

sagegrouse

Agree.

For the 2009-10 season
6-3 in January
4-0 so far in February

A reversal of the "trend" perhaps.

[El_Diablo]

Regarding the MSU and UNC comparisons...

He admits that MSU has a reputation for getting better as the season progresses. And they don't play the same team at the end of every season like we do (which also happened to be a national powerhouse). Despite this, they still have a negative trend over the same time period. Huh? It's not as pronounced as ours, but it's still there.

http://sports.chronicleblogs.com/wp-content/uploads/2010/02/Mich-State-300x266.jpg

UNC has been pretty dominant in the ACC recently, and we've been good but not great. The Duke game at the end of the year hasn't been as much of a burden for them as it has been for us--just look at the comparisons. If UNC had to close out against an elite team consistently, I can bet their numbers would show a stronger negative trend. As it stands though, they still have a slight negative trend. Some of UNC's years (like 2005) have a bigger dropoff than others.

http://sports.chronicleblogs.com/wp-content/uploads/2010/02/UNC-Final1-300x264.jpg

I'm not saying "quality of opponent" is the only reason for our late-season dropoff, but it's a definitely a contributing factor.

[Fish80]

My personal theory is that the fact that Duke is on TV more than anyone, more hated than anyone, more analyzed than anyone, means that any coach playing us has the analysis of thousands to help him find our weaknesses.

I also think K is better at utilizing the talent he has so our our teams come together quickly, whereas other teams take all year to jell.

Spot on. As the season progresses, there is a ton of data for opposing coaches to mine. And mine they do. Remember 1999? Calhoun kept note cards on Duke as the season progressed, just in case UConn bumped into us in the tourney.

[CrazieDUMB]

Regarding the MSU and UNC comparisons...

He admits that MSU has a reputation for getting better as the season progresses. And they don't play the same team at the end of every season like we do (which also happened to be a national powerhouse). Despite this, they still have a negative trend over the same time period. Huh? It's not as pronounced as ours, but it's still there.

http://sports.chronicleblogs.com/wp-content/uploads/2010/02/Mich-State-300x266.jpg

UNC has been pretty dominant in the ACC recently, and we've been good but not great. The Duke game at the end of the year hasn't been as much of a burden for them as it has been for us--just look at the comparisons. If UNC had to close out against an elite team consistently, I can bet their numbers would show a stronger negative trend. As it stands though, they still have a slight negative trend. Some of UNC's years (like 2005) have a bigger dropoff than others.

http://sports.chronicleblogs.com/wp-content/uploads/2010/02/UNC-Final1-300x264.jpg

I'm not saying "quality of opponent" is the only reason for our late-season dropoff, but it's a definitely a contributing factor.



Can I be a huge Econ dork for a second and point out how low the r squared's are? To the uninformed, Rsquare is a metric that explains how effieciently the trend reflects the data. BC these numbers are so low for UNC and MSU, one could say that this regressor is not very powerful. I strongly doubt that the negative trend would be significant at even the 90% level.

[RoyalBlue08]

I also think K is better at utilizing the talent he has so our our teams come together quickly, whereas other teams take all year to jell.

I have always thought our team comes out in Nov. and Dec. playing much better than the other top teams...be that because of good coaching or smarter players that pick things up quicker or something else I'm not sure. The years we have seemed to decline towards the end of the year, I always thought it was more the other teams catching up while we have already maxed out our potential as opposed to our skill or effort going in reverse. These things are hard to quantify though.

[greybeard]

"Trust me, it's science." I don't think so.

Singler's freshman year he got pounded by State and Clemson and whomever in three straight games near the end of the regular season and he was finished. Done. Anyone dispute that?

The next year anyone remember Nelson's injury and how he limped, literally to the finish line? And, wasn't Lance done before the end of one of those years.

And, earlier than that, Duke's lone point guard was playing with a broken foot. Broken feet tend to bother one more the more you use them and once teams got it that the guy could not move laterally that is all that they made him do. Zoubek that same year or was it another also had a broken foot.

And McRob's career at Duke were marred by a very bad back, that required surgery after his freshman year. What, that back loosened up with the wear and tear of an elongated regular season?

Now, you will say that other teams have had injuries. Yeap. And, when Lawson went down, so did Carolina. But, Duke has been thinner in terms of depth at some key positions and key players have been hurt or just plain worn out (see, in addition to Singler his first year, Scheyer last year) and have still had to carry the load. Over time, carrying a heavy load gets heavier if you are hurt or worn out, which key Duke players were. Science, no; fact, yes.

[BD80]

Yeah, there's a decline, but he didn't explain why. I threw in my 2 cents, citing quality of opponents. Referencing UNC and MSU doesn't really negate the fact that we play tougher opponents later in the season.

And speaking of UNC...anyone care to run the numbers on UNC's efficiency ratings this year? I bet there's a negative trend...

How can you trend downwardly from 0?

[Hermy-own]

How can you trend downwardly from 0?

Easily. A negative data point on those graphs indicates a game you lost. For UNC and Duke, the points are mostly, but not all, positive. This year UNC has had a number of negative data points recently.

[Starter]

He did, however, present some pretty convincing evidence that the decline did not correlate with minutes played by key players. He does not have to say why or why not; he simply says what is. Sounds like good, objective reporting to me.

Did he? He said he had convincing evidence, but never presented it. Plus, his crux is this: "Even in years where Coach K has played a deep bench and kept his stars rested, his team still declined."

What years were those, by chance? No love lost, he worked hard, but I still see nothing here.

I obviously know you get jumped on here just by bringing this up, and I certainly don't expect or even care to change any minds, but you can't possibly tell me players will be fresher (and healthier) at the end of the year playing 35 mpg than, say, 28. You just can't, period, with all the efficiency ratings or whatever in the world. (That's a term a lot of people here probably just learned anyway by reading the article) This also goes into the personnel you have at your disposal; in fact, I think it's more that than any substitution patterns.

The depth in our frontcourt this year (and our backcourt next year) hopefully indicates a trend that Krzyzewski has spotted. He has blamed himself, correctly or not, for not surrounding Redick with the sort of team that would have allowed him to succeed. I think he's mindful of it.

[91_92_01_10_15]

Did he? He said he had convincing evidence, but never presented it.


Maybe you missed the link to the "More on the Math" part of the article. Here it is:

http://sports.chronicleblogs.com/2010/02/17/more-on-the-math-behind-duke-does-decline-objectively-speaking/

Argue with this evidence as you like. My guess is that it will not be enough to convince you that your opinion is wrong. It is enough to convince me that mine is right, however, which is admittedly quite easy to do. ;)

[Kedsy]

...but you can't possibly tell me players will be fresher (and healthier) at the end of the year playing 35 mpg than, say, 28. You just can't, period...

Well, we could tell you, and frankly we'd be right, but it's certainly your prerogative to ignore us.

[camion]

With an R-square of 0.03 or 0.00 those slightly negative lines for UNC and MSU are virtually meaningless. Since I'm not a statistician I'll just copy and paste from a web page (http://www.graphpad.com/help/Prism5/r2_ameasureofgoodness_of_fitoflinearregression.htm ).

"The value r2 is a fraction between 0.0 and 1.0, and has no units. An r2 value of 0.0 means that knowing X does not help you predict Y. There is no linear relationship between X and Y, and the best-fit line is a horizontal line going through the mean of all Y values."

[CrazieDUMB]

"The value r2 is a fraction between 0.0 and 1.0, and has no units. An r2 value of 0.0 means that knowing X does not help you predict Y. There is no linear relationship between X and Y, and the best-fit line is a horizontal line going through the mean of all Y values."

As I posted before, this is correct. What it seems to suggest is that for UNC and MSU (I dont know as much about MSU, but UNC does run a fairly deep bench), for teams that use more subs the lateness of the season doesn't affect their outcome. That is, this isn't a problem for them, whereas for duke, not only is the trend negative, but much more a likely indicator of the reason why.

[Dukeface88]

"Trust me, it's science." I don't think so.

Singler's freshman year he got pounded by State and Clemson and whomever in three straight games near the end of the regular season and he was finished. Done. Anyone dispute that?

The next year anyone remember Nelson's injury and how he limped, literally to the finish line? And, wasn't Lance done before the end of one of those years.

And, earlier than that, Duke's lone point guard was playing with a broken foot. Broken feet tend to bother one more the more you use them and once teams got it that the guy could not move laterally that is all that they made him do. Zoubek that same year or was it another also had a broken foot.

And McRob's career at Duke were marred by a very bad back, that required surgery after his freshman year. What, that back loosened up with the wear and tear of an elongated regular season?

Now, you will say that other teams have had injuries. Yeap. And, when Lawson went down, so did Carolina. But, Duke has been thinner in terms of depth at some key positions and key players have been hurt or just plain worn out (see, in addition to Singler his first year, Scheyer last year) and have still had to carry the load. Over time, carrying a heavy load gets heavier if you are hurt or worn out, which key Duke players were. Science, no; fact, yes.

How does this refute the article? The analysis says that there is a correlation between when games are played and our efficiency margin, but that there is no evidence this correlation is caused by minutes played (in fact, the opposite appears to be the case). Nothing you say changes either of those.

[CrazieDUMB]

What we really need here is a regression of individual efficiency, regressed against number of games into the ACC season and weighted by minutes played. Any ambitious Dukies still have access to STATA or SAS and want to tackle that one?

[loldevilz]

Does anyone else feel that the decline isn't to do with Duke's stars getting exhausted, but rather bench players and other starters not improving as the season progresses. I know he said that this year our efficiency has been improving (only slightly). Anyone that has been watching the games knows that this is at least partially because of improved play by the big guys.

[greybeard]

How does this refute the article? The analysis says that there is a correlation between when games are played and our efficiency margin, but that there is no evidence this correlation is caused by minutes played (in fact, the opposite appears to be the case). Nothing you say changes either of those.

Singler, Nelson, Scheyer all at different times late in the season after playing way many minutes went into the tank. McRob fell far at the end of his second season. I saw these things happen. You didn't? I saw them impact on Duke's play? You didn't?

The author says that there is no correlation between the fall off and time played and I say, in the words of the Great Bard, "You don't need a weatherman to tell which way the wind blows." The corrolation to minutes played and these key players' going south seems obvious to me.

In short, "if we have learned anything from history," aside from the reality that anybody can be killed at anytime, it is that numbers can be used to prove just about anything, even things that just ain't so. Here, this guy's use of statistics did just that; the real-life facts prove it. ;)

[Acymetric]

I think one of the questions some poeple would like answered is this:

How do you know its extra minutes of game time wearing out the players? Isn't it just as likely, perhaps more so, that working out and doing practices is wearing out the players? Social lives could also come into play...why assume its entirely based on 5-6 extra minutes a game (or less)?

[Dukeface88]

Singler, Nelson, Scheyer all at different times late in the season after playing way many minutes went into the tank. McRob fell far at the end of his second season. I saw these things happen. You didn't? I saw them impact on Duke's play? You didn't?

The author says that there is no correlation between the fall off and time played and I say, in the words of the Great Bard, "You don't need a weatherman to tell which way the wind blows." The corrolation to minutes played and these key players' going south seems obvious to me.

In short, "if we have learned anything from history," aside from the reality that anybody can be killed at anytime, it is that numbers can be used to prove just about anything, whether even if what it proves is not so. Here, this guy's use of them did just that; the facts, not statistical analysis, prove it. ;)

The statistical analysis is the facts. Yes, key players have been asked to play big minutes. Yes, the team frequently trends downward during the season. But these two things do not occur together - teams that have given starters heavy minutes have not experienced a greater decline that teams that gave starters fewer minutes (in fact, it seems to be the opposite- the more the starters play, the less the team goes downhill, although there are correlation/causation problems there).

You may not need a weather man to tell you which way the wind is blowing - but you do need one to tell you what kind of weather that wind is going to bring.

[Kedsy]

Singler, Nelson, Scheyer all at different times late in the season after playing way many minutes went into the tank. McRob fell far at the end of his second season. I saw these things happen. You didn't? I saw them impact on Duke's play? You didn't?

As others have noted, the issue is one of causation. Just because these things may have happened doesn't necessarily mean they were caused by playing too many minutes. If you insist that a few individual datapoints prove something your chance of being right is not particularly high. Especially when there are plenty of counterpoints. All five of Duke's starters in 1992 played 30+ minutes per game, for example. Shane Battier played 35 minutes per game in 2001 without any noticeable dropoff.

When I was at Duke, my statistics professor told a story (which admittedly may have been apocryphal) about malaria. He showed us a map detailing which areas in the world had the highest occurrences of malaria in the early 20th century and a second map showing the largest consumption of Coca Cola at the time. The maps were almost identical, which led to an outcry accusing the soft drink manufacturer of causing malaria. Cooler heads eventually prevailed, pointing out that it must have been a third (unrelated) factor that underlay both observations, in this case hot temperatures which were both necessary for certain types of mosquitos to flourish and also made humans yearn for a cold drink.

My point is observing some Duke players play a lot of minutes and also noting that Duke teams have experienced dropoffs in efficiency toward the end of the year doesn't necessarily mean that one caused the other. If the dropoff is consistent across seasons but the amount of minutes played is not -- which is the point of the Chronicle article -- then the likely explanation is there are additional factors causing the dropoff unrelated to the minutes.

[greybeard]

DUKEFACE88 AND KEDSKY We're going to have to diasagree here. Singler was worn out; ask him. Everyone has said so for years now, well, two since he was worn out.

McRob played mucho minutes his second season and his offense declined and he just could not, as he did early, contain on defense and get key rebounds in a crowd.

Nelson literally looked like he ran into a wall. Did you see him play? How can you conclude otherwise? He had killed himself all season and it took it's toll on him.

Scheyer last year was the second key piece to a super pressurized defense, especially once Elliot took the wheel on defense. He couldn't throw it in the ocean as he got deep into the season. Had no apparent injury. Yeah, I know, just a coincidence. Bad luck. A shooting slump. Science my foot!

Battier is an extraordinary guy in every way. You are comparing these guys to him? Besides, he did have more than a few other serious, serious athletes on the floor with him, no? And, he was lucky.

How come Brand, Boozer, and Dunlevy have all been injury prone? I don't know. Maybe all those minutes and playing so hard has caught up with them. Maybe playing as hard as K gets "key" guys to play has its costs. Pay me now or pay me later. I'm not saying that that is so, but it might be.

I wonder, is the less intense defense Duke is presenting this year reducing the wear and tear, and also avoiding injury? What do the statistics say? That's what I thought.

[tbyers11]

Singler, Nelson, Scheyer all at different times late in the season after playing way many minutes went into the tank. McRob fell far at the end of his second season. I saw these things happen. You didn't? I saw them impact on Duke's play? You didn't?



Would 28.6 minutes per game for the entire season and 31.1 per game that season be more reasonable numbers for a Duke player? I assume that you are referring to Kyle's freshman year in 2007-08 as the year that he hit the wall. That is what Kyle averaged that year.

I'm not disagreeing that Kyle wore down that year. I just don't think it was due strictly to playing excessive minutes. I believe he wore down because he was asked to play inside too much against bigger players and that physical toll of the that combined with the mental toll of his first season of college ball weighed on him. These traits did affect him but I don't think that they would have affected him differently over the course of the season if he played 28 minutes/gm (conference) instead.

[Kedsy]

DUKEFACE88 AND KEDSKY We're going to have to diasagree here. Singler was worn out; ask him. Everyone has said so for years now, well, two since he was worn out.

The Board has been going round and round on this for years, so I don't want to go through it yet again now. But I will again say the issue is causation. Kyle was worn out at the end of his freshman year -- that much is undisputed -- but there are plenty of potential reasons for that beyond the 28.6 minutes per game he played that year. For one thing, 28 minutes isn't really all that many (and he played 15% more minutes last year without appearing worn out). For another, he was playing out of position and banging against bigger players, which can take a lot out of you no matter how many minutes you play. Probably most importantly he was a freshman, 3000 miles away from home, playing a much longer season against much bigger and better players, with a great deal more practice time and a lot more pressure/stress and requiring a much higher intensity level (both in games and at practice) than he'd ever had before. Do you really think he would have been OK if he'd played 25 minutes a game instead of 28?

[Dukeface88]

DUKEFACE88 AND KEDSKY We're going to have to diasagree here. Singler was worn out; ask him. Everyone has said so for years now, well, two since he was worn out.

McRob played mucho minutes his second season and his offense declined and he just could not, as he did early, contain on defense and get key rebounds in a crowd.

Nelson literally looked like he ran into a wall. Did you see him play? How can you conclude otherwise? He had killed himself all season and it took it's toll on him.

Scheyer last year was the second key piece to a super pressurized defense, especially once Elliot took the wheel on defense. He couldn't throw it in the ocean as he got deep into the season. Had no apparent injury. Yeah, I know, just a coincidence. Bad luck. A shooting slump. Science my foot!

"My friend won the lottery. Therefore, the lottery is an excellent investment plan."

You can find isolated examples to support almost any claim. Those examples don't make them right (unless they are countering a universal statement), and they don't help to predict future results. This is exactly why statistics exist. "Did Kyle Singler get tired his freshman year?" may (or may not) be an interesting question. Either way, it isn't a question that is actually important if we want to discuss the merits of Coach K's bench managment. "Will giving starters more rest help the team?" is, and the answer is "On average, no."

[greybeard]

Would 28.6 minutes per game for the entire season and 31.1 per game that season be more reasonable numbers for a Duke player? I assume that you are referring to Kyle's freshman year in 2007-08 as the year that he hit the wall. That is what Kyle averaged that year.

I'm not disagreeing that Kyle wore down that year. I just don't think it was due strictly to playing excessive minutes. I believe he wore down because he was asked to play inside too much against bigger players and that physical toll of the that combined with the mental toll of his first season of college ball weighed on him. These traits did affect him but I don't think that they would have affected him differently over the course of the season if he played 28 minutes/gm (conference) instead.

This is a serious answer?

The other guys didn't wear down?

There is a point at which players play in pain because their cardio systems cannot keep up and they literally eat themselves up. There is a point at which legs turn to rubber, backs stay in perennial spasm, and the more minutes the deeper the spasm's impact, etc. Once you go over the edge, the longer you stay, the more the damage. It is cumulative.

3 1/2 minutes. Is this a serious question? Sure they mattered. How could it have been otherwise. Had they gotten rests instead of playing in the red zone, maybe they never break down. But, these guys were it; they had to play over-the-top hard, no gliding, no rest, not at key times. In big games, unless in foul trouble, the minutes were longer, the strain enormous; they were the go-to guys, usually on both ends and the boards; playing extended minutes wore them out. We all saw it. Well, everyone who didn't have his head in a calculator.

[tbyers11]

You weren't addressing me but sometimes I think throwing facts out to reference claims made is worthwhile.



Scheyer last year was the second key piece to a super pressurized defense, especially once Elliot took the wheel on defense. He couldn't throw it in the ocean as he got deep into the season. Had no apparent injury. Yeah, I know, just a coincidence. Bad luck. A shooting slump. Science my foot!

Scheyer's shooting slump was in the middle, not the end, of the season last year. In the last 12 games of the year he shot 42.3% (FG) and 43.2% (3PT) compared to season averages of 39.5 and 38.3, respectively. He did shoot poorly against Villanova 3-18 (2-10 3pt). Maybe, he just got really tired before that game.


How come Brand, Boozer, and Dunleavy have all been injury prone? I don't know. Maybe all those minutes and playing so hard has caught up with them. Maybe playing as hard as K gets "key" guys to play has its costs. Pay me now or pay me later. I'm not saying that that is so, but it might be.

I really don't think minutes played in approximately 80 (Brand) and 120 (Boozer and Dunleavy) college games really affected their injury status years down the road in the NBA. They played as many games after a year and a half in the NBA as they did in their entire college career.

[greybeard]

You weren't addressing me but sometimes I think throwing facts out to reference claims made is worthwhile.



Scheyer's shooting slump was in the middle, not the end, of the season last year. In the last 12 games of the year he shot 42.3% (FG) and 43.2% (3PT) compared to season averages of 39.5 and 38.3, respectively. He did shoot poorly against Villanova 3-18 (2-10 3pt). Maybe, he just got really tired before that game.



I really don't think minutes played in approximately 80 (Brand) and 120 (Boozer and Dunleavy) college games really affected their injury status years down the road in the NBA. They played as many games after a year and a half in the NBA as they did in their entire college career.

I didn't think Scheyer played well once he went into a slump. My memory could be faulty but I just didn't see a difference maker out there, not when it was needed, not on offense. But, maybe I'm wrong about him. McRob, Singler, Nelson, you do agree; they all hit the wall and were done. Right? I apoligize if my recollection about Scheyer proves incorrect.

My reference to the endemic injuries to Boozer, Brand, and Dunleavy in the pros was a bit of a cheap shot, but fit with my point about the deliberate pull back on defense this season--K can't afford to lose any of the big three or have them worn out, so the defensive pressure has been ratcheted back. I guess K is less concerned with statistics than experience. And, did anybody but me notice how in the last game, way before it was out of reach, there were some guys getting minutes from Scheyer and Singler who normally wouldn't. Coincidence I suppose.

The cheap shot is that I think that K's approach to the game, the intensity that he asks from players, especially his stars, might well have a cost. Brand, Boozer, and Dunleavy, all play way above their heads as pros, just as they learned to do in college and all are injury prone. I prefer to see players develop a style of play that is safe. I like very much how Duke is playing this year. I think Smith is playing much safer and more effectively than in the past, and that Scheyer's control of the pace that he plays at on offense is terrific. Lately, some of his drives, while seemingly under control, have begun to scare me. I'd like to see him cut it out.

[BD80]

The Board has been going round and round on this for years, so I don't want to go through it yet again now. But I will again say the issue is causation. Kyle was worn out at the end of his freshman year -- that much is undisputed -- but there are plenty of potential reasons for that beyond the 28.6 minutes per game he played that year. For one thing, 28 minutes isn't really all that many (and he played 15% more minutes last year without appearing worn out). For another, he was playing out of position and banging against bigger players, which can take a lot out of you no matter how many minutes you play. Probably most importantly he was a freshman, 3000 miles away from home, playing a much longer season against much bigger and better players, with a great deal more practice time and a lot more pressure/stress and requiring a much higher intensity level (both in games and at practice) than he'd ever had before. Do you really think he would have been OK if he'd played 25 minutes a game instead of 28?

This is the point Coach K has specifically addressed. He does not believe there is a difference.

How many minutes/hours per week does a starter spend practicing, working out, stretching, warming up, studying film, or otherwise preparing for games? Would all of those hours change if he were to play six minutes less per week?

Is six minutes of playing time relevant compared to the dozens of hours of basketball activity each week?

[Dukeface88]

3 1/2 minutes. Is this a serious question? Sure they mattered. How could it have been otherwise. Had they gotten rests instead of playing in the red zone, maybe they never break down. But, these guys were it; they had to play over-the-top hard, no gliding, no rest, not at key times. In big games, unless in foul trouble, the minutes were longer, the strain enormous; they were the go-to guys, usually on both ends and the boards; playing extended minutes wore them out. We all saw it. Well, everyone who didn't have his head in a calculator.

"You can see everything. You fail, however, to reason from what you see."

People see what they look for. People who look for tired players will see tired players. They won't see those who are playing better because extra minutes have left them better prepared for the intensity of March. That's not even getting in to things like cognitive dissonance, availability bias, confirmation bias and the like.

Edit: For example...


I didn't think Scheyer played well once he went into a slump. My memory could be faulty but I just didn't see a difference maker out there, not when it was needed, not on offense. But, maybe I'm wrong about him. McRob, Singler, Nelson, you do agree; they all hit the wall and were done. Right? I apoligize if my recollection about Scheyer proves incorrect.


Starting with the Miami game (i.e., the second of half of ACC play), through the end of year, Scheyer averaged 19.5 points per game. That's pretty good by anyone's standards. But you weren't looking for it, so you didn't see it.

[tbyers11]

This is a serious answer?

The other guys didn't wear down?

There is a point at which players play in pain because their cardio systems cannot keep up and they literally eat themselves up. There is a point at which legs turn to rubber, backs stay in perennial spasm, and the more minutes the deeper the spasm's impact, etc. Once you go over the edge, the longer you stay, the more the damage. It is cumulative.

3 1/2 minutes. Is this a serious question? Sure they mattered. How could it have been otherwise. Had they gotten rests instead of playing in the red zone, maybe they never break down. But, these guys were it; they had to play over-the-top hard, no gliding, no rest, not at key times. In big games, unless in foul trouble, the minutes were longer, the strain enormous; they were the go-to guys, usually on both ends and the boards; playing extended minutes wore them out. We all saw it. Well, everyone who didn't have his head in a calculator.

Yeah, it is a serious answer. I agree that something looked wrong with Nelson against WVA and Belmont. There were rumors that he was sick. Even if he wasn't, he looked fine against against GT and Clemson in the ACC tourney the week before. When does this fatigue kick in? Does it come and go? In addition to Scheyer's shooting getting better at the end of last year I thought he looked great overall in the ACC tourney. I honestly don't remember McBob being more tired as the season wore on. He might have been

You seem to think that the difference between 28 and 31 minutes/gm occurs during a flatout sprint race at a 5-minute per mile pace. With all the stoppages of play in a college basketball game due to timeouts, deadballs and free throws that 3 1/2 minute difference isn't remotely close to the difference that you presume. The emotional stress and strain of being the go-to-guy is going to be there whether you play 28 or 31 minutes.

As for the pull back to a less aggressive defense, I think it was due to not have great defensive quickness on the perimeter and having a better shot-blocking presence. It might save the players a little bit, but I don't think it really decreases their likelihood of injury. A lot of basketball injuries are freak occurrences. Coming down on a defenders ankle after a jumper or Lance knocking knees with Drew against UNC.

In the end, I do agree that some of our players have looked tired, but I don't think it is due to a difference in 2-3 minutes a game. Kyle was asked to do too much as a freshman. I think Nelson felt the pressure to do too much in the tourney his senior year. I think the emotional pressure got to JJ at the end of 2006. I do agree that several of our players have not looked their best at the end recently but I think it is simply due to the team not having enough "viable options" to carry the load recently due to early defections from the Deng, Livingston, Humphries recruiting class and the Paulus/McRoberts recruiting class generally not panning out and not due to 3 minutes/gm during the season. Even great players aren't going to play great every game in the stress of the NCAA tourney. Laettner was gassed in the 92 final against Michigan, but Grant came through with a big game.

Hopefully this year, we have enough options to make a big tourney run. I think we also agree to beat Miami tonight. Cheers.

[greybeard]

Yeah, it is a serious answer. I agree that something looked wrong with Nelson against WVA and Belmont. There were rumors that he was sick. Even if he wasn't, he looked fine against against GT and Clemson in the ACC tourney the week before. When does this fatigue kick in? Does it come and go? In addition to Scheyer's shooting getting better at the end of last year I thought he looked great overall in the ACC tourney. I honestly don't remember McBob being more tired as the season wore on. He might have been

You seem to think that the difference between 28 and 31 minutes/gm occurs during a flatout sprint race at a 5-minute per mile pace. With all the stoppages of play in a college basketball game due to timeouts, deadballs and free throws that 3 1/2 minute difference isn't remotely close to the difference that you presume. The emotional stress and strain of being the go-to-guy is going to be there whether you play 28 or 31 minutes.

As for the pull back to a less aggressive defense, I think it was due to not have great defensive quickness on the perimeter and having a better shot-blocking presence. It might save the players a little bit, but I don't think it really decreases their likelihood of injury. A lot of basketball injuries are freak occurrences. Coming down on a defenders ankle after a jumper or Lance knocking knees with Drew against UNC.

In the end, I do agree that some of our players have looked tired, but I don't think it is due to a difference in 2-3 minutes a game. Kyle was asked to do too much as a freshman. I think Nelson felt the pressure to do too much in the tourney his senior year. I think the emotional pressure got to JJ at the end of 2006. I do agree that several of our players have not looked their best at the end recently but I think it is simply due to the team not having enough "viable options" to carry the load recently due to early defections from the Deng, Livingston, Humphries recruiting class and the Paulus/McRoberts recruiting class generally not panning out and not due to 3 minutes/gm during the season. Even great players aren't going to play great every game in the stress of the NCAA tourney. Laettner was gassed in the 92 final against Michigan, but Grant came through with a big game.

Hopefully this year, we have enough options to make a big tourney run. I think we also agree to beat Miami tonight. Cheers.

Very excellent. Much better than statistics! BTW, I enjoyed the heck out of watching those guys give it their all and then some. Did I mention that I thought your post excellent? :o

[mehmattski]

I dug a little deeper into the statistics to address some of the tangible criticisms raised in this thread. For example: Duke does play tougher competition as the season goes on, and that is found to be a statistically significant trend. And Duke does play significantly worse against good teams across 2004-2010. Finally, home court advantage plays a very small role in the trend.

Link: http://www.immaculateinning.com/2010/02/duke-fade-response.html

Correlation does not imply causation, and I have not attempted to attach a causal link to the "Duke Fade." However, it is worth pointing out that in 2004, there was a statistically significant "Duke Fade." If the Fade is bad, how then did that team manage to make the Final Four with such "tired legs?"

[NovaScotian]

Can I be a huge Econ dork for a second and point out how low the r squared's are? To the uninformed, Rsquare is a metric that explains how effieciently the trend reflects the data. BC these numbers are so low for UNC and MSU, one could say that this regressor is not very powerful. I strongly doubt that the negative trend would be significant at even the 90% level.

i'm with you. r^2=.12? that means nothing. just look at the scatterplot - there's almost no relationship. to be fair, he says that the r^2 value means 12% of the variation is due to the game played, but seriously that can't be enough to go on. I think it's a really cool way to look at things, but maybe we should look at the trends for a few more schools to see if it's an accurate metric.

[mehmattski]

i'm with you. r^2=.12? that means nothing. just look at the scatterplot - there's almost no relationship. to be fair, he says that the r^2 value means 12% of the variation is due to the game played, but seriously that can't be enough to go on. I think it's a really cool way to look at things, but maybe we should look at the trends for a few more schools to see if it's an accurate metric.

An R^2 of 0.12 does not explain a lot of the variance, but it is significant at a very low p-value. The linear regression has a slope that is significantly different than zero. By definition, that is "enough to go on."

Correlation, of course, does not imply causation. Only a small amount of the variance in Duke's efficiency margin is explained by the number of ACC games played. As I showed in my post above, some more of the variance is explained by the quality of the opponent and the home/road disparity. In total there's about 80% of the variance unexplained. That's a lot, but statistically speaking, the r-squared value is still significant.

One method would be to improve the model beyond a linear regression, but that's taking things a little further than even I am comfortable with.

[Starter]

Well, we could tell you, and frankly we'd be right, but it's certainly your prerogative to ignore us.

Amazingly, this argument right here convinced me, along with all that stuff about efficiency ratings. Let the 35+ mpg seasons continue indiscriminately, despite how the players often look -- and play -- at the end of our seasons. That said...


Maybe you missed the link to the "More on the Math" part of the article. Here it is:

I did miss that! My apologies to you, and to the writer of the article. And THAT said...

Let's go Duke. Great win tonight. I think we should get the Chronicle guy to chart Zoubek's rapid ascent into stardom.

[BD80]

Let's see. If our key players were wearing down from playing too many minutes, they would be wearing down during games.

Their performance would decline in the second half of games.

How did our key players do in the second half last night vs the first half?

To my eye they did visibly better. Statistics back that up.

Maybe we can't draw conclusions from such small statistical samples.

[CrazieDUMB]

Not to pat ourselves on the back or anything, but I love that at the Duke message board the argument relies on the intricacies of linear regression models. Never seen that on ESPN conversation...

mehmattski, excellent blog post

[Starter]

Maybe we can't draw conclusions from such small statistical samples.

I mean, sure you can. Depends which statistics you want to use. Scheyer, for example, shot 3-for-15 for the game -- he began 1-for-12 -- while most of his misses were short. To be fair, Krzyzewski said Scheyer's been dealing with a sore back (wonder how that happened).

Listen, I'd absolutely love to be shown that this was the right way to go as the season wanes. If the Big 3 has this brilliant tournament and they go deep, and look relatively fresh the whole way, I'll be the first to come on here and say, "BD80, nice job, these guys held up great playing all those minutes." You can hold me to that, I'm a stand-up dude -- ask Julio. But the memories of Redick shooting 36% over his final 10 games in 2005 and 37% over his final 10 games in 2006 cut deep for me. They really do, because those teams had so much potential, mostly because of Redick's early-season brilliance. We'll see how it goes.