Length of Duke Rotation under Coach K

[Kedsy]

Kind of depends on what you call "in the rotation" and also, what are the quality of minutes...I mean, is a ten MPG average made on 20 minutes in blow outs and zero or 1 in tight games?

I've moved this discussion from the 2019 recruiting thread. The above is a good point. As many of you know, my definition of "length of rotation" is how many players play 10+ minutes in close games (less than 20 points) after January 1.

I have compiled the number of players who played 10+ minutes in such games for every year Coach K has coached at Duke. Here's the table:



Year 5 6 7 8 9 10 Avg.
2018 4 4 7 1 6.31
2017 6 14 1 6.76
2016 1 18 3 0 6.09
2015 1 9 8 2 6.55
2014 2 8 6 1 7.35
2013 6 10 2 6.78
2012 1 10 9 2 7.55
2011 6 9 5 6.95
2010 2 5 8 6 6.86
2009 1 4 10 4 1 7.00
2008 3 5 8 3 7.58
2007 1 3 11 2 6.82
2006 4 17 0 6.81
2005 6 14 0 6.70
2004 3 14 1 6.89
2003 3 17 2 1 7.04
2002 8 3 0 6.27
2001 8 8 0 6.50
2000 11 5 0 6.31
1999 2 7 2 7.00
1998 0 9 4 7.31
1997 1 9 7 1 7.44
1996 4 13 1 6.83
1994 8 14 0 6.64
1993 7 8 4 6.84
1992 10 9 5 6.79
1991 0 13 0 4 7.47
1990 1 11 9 7.38
1989 1 9 7 7.35
1988 3 12 6 2 7.30
1987 3 11 6 1 7.24
1986 2 16 4 1 7.22
1985 10 7 2 6.58
1984 8 12 1 6.67
1983 2 6 5 1 7.36
1982 7 5 4 1 6.94
1981 7 6 3 1 6.88


As you can see, the numbers people cite showing 8 (or more) players with 10+ mpg are skewed by injuries, blowouts, and November/December games. For example, this past season we had 8 guys who played more than 10mpg, but if you factor out blowouts and early season games and things like AOC playing 20 minutes in a game that Marques sat out with an injury, it turns out that in non-blowouts after January 1, we only had one game in which eight players played 10+ minutes, compared to four games when we only had five players who played 10+ minutes. Using my definition, our rotation was a lot closer to 6 players (an average of 6.3) than it was to 8 players.

Over the course of Coach K's tenure at Duke (again, using my definition), the rotation has ranged from 6 to 7.5, with an average just under 7. Since K's first season in 1981 (not including 1995, because Coach K didn't coach the team after January 1 that season), we have played 8+ players for 10+ minutes only 20% of close games after January 1. Since 2000, we've played 8+ players for 10+ minutes less than 16% of post-January 1 close games. In the last four seasons, we've only had four games in which 8 or more players played 10+ minutes in such games. Clearly under this definition an 8-man rotation has not been the norm.

Obviously, nobody is forced to accept my definition. You might think a player who gets 6 or 8 minutes should count in the rotation for a particular game, or you might think garbage time should count, or you might want to include November/December games.

But one thing I hope we can agree on is it's difficult to have an intelligent conversation about the length of Duke's rotation if we can't agree on what "rotation" means.

[budwom]

good stuff. And you should also mention that minutes per game needs to be defined (and I think you have) as minutes played divided by total number of Duke games (or games for which the player was available.)
i.e. according to a lot of stats, if Duke plays 40 games and player A plays a total of 100 minutes in 10 games, he's averaging 10 minutes per game, whereas I (and you I believe) would say he averages only 2.5 minutes/game.

Having said that, I think we can agree that this is the year we play nine guys. Have to. It's obvious.

[BD80]

I thought this would be a bibliography of all of the other threads that have discussed, nay -thrashed and beaten until naught remains - this topic.

[Kedsy]

good stuff. And you should also mention that minutes per game needs to be defined (and I think you have) as minutes played divided by total number of Duke games (or games for which the player was available.)
i.e. according to a lot of stats, if Duke plays 40 games and player A plays a total of 100 minutes in 10 games, he's averaging 10 minutes per game, whereas I (and you I believe) would say he averages only 2.5 minutes/game.

Yes, I agree. DNP-CD should be considered equal to zero minutes in one game.


I thought this would be a bibliography of all of the other threads that have discussed, nay -thrashed and beaten until naught remains - this topic.

I know you hate this topic. I was just trying to get it out of the 2019 recruiting thread and into it's own thread.

[luvdahops]

I've moved this discussion from the 2019 recruiting thread. The above is a good point. As many of you know, my definition of "length of rotation" is how many players play 10+ minutes in close games (less than 20 points) after January 1.

I have compiled the number of players who played 10+ minutes in such games for every year Coach K has coached at Duke. Here's the table:



Year 5 6 7 8 9 10 Avg.
2018 4 4 7 1 6.31
2017 6 14 1 6.76
2016 1 18 3 0 6.09
2015 1 9 8 2 6.55
2014 2 8 6 1 7.35
2013 6 10 2 6.78
2012 1 10 9 2 7.55
2011 6 9 5 6.95
2010 2 5 8 6 6.86
2009 1 4 10 4 1 7.00
2008 3 5 8 3 7.58
2007 1 3 11 2 6.82
2006 4 17 0 6.81
2005 6 14 0 6.70
2004 3 14 1 6.89
2003 3 17 2 1 7.04
2002 8 3 0 6.27
2001 8 8 0 6.50
2000 11 5 0 6.31
1999 2 7 2 7.00
1998 0 9 4 7.31
1997 1 9 7 1 7.44
1996 4 13 1 6.83
1994 8 14 0 6.64
1993 7 8 4 6.84
1992 10 9 5 6.79
1991 0 13 0 4 7.47
1990 1 11 9 7.38
1989 1 9 7 7.35
1988 3 12 6 2 7.30
1987 3 11 6 1 7.24
1986 2 16 4 1 7.22
1985 10 7 2 6.58
1984 8 12 1 6.67
1983 2 6 5 1 7.36
1982 7 5 4 1 6.94
1981 7 6 3 1 6.88


As you can see, the numbers people cite showing 8 (or more) players with 10+ mpg are skewed by injuries, blowouts, and November/December games. For example, this past season we had 8 guys who played more than 10mpg, but if you factor out blowouts and early season games and things like AOC playing 20 minutes in a game that Marques sat out with an injury, it turns out that in non-blowouts after January 1, we only had one game in which eight players played 10+ minutes, compared to four games when we only had five players who played 10+ minutes. Using my definition, our rotation was a lot closer to 6 players (an average of 6.3) than it was to 8 players.

Over the course of Coach K's tenure at Duke (again, using my definition), the rotation has ranged from 6 to 7.5, with an average just under 7. Since K's first season in 1981 (not including 1995, because Coach K didn't coach the team after January 1 that season), we have played 8+ players for 10+ minutes only 20% of close games after January 1. Since 2000, we've played 8+ players for 10+ minutes less than 16% of post-January 1 close games. In the last four seasons, we've only had four games in which 8 or more players played 10+ minutes in such games. Clearly under this definition an 8-man rotation has not been the norm.

Obviously, nobody is forced to accept my definition. You might think a player who gets 6 or 8 minutes should count in the rotation for a particular game, or you might think garbage time should count, or you might want to include November/December games.

But one thing I hope we can agree on is it's difficult to have an intelligent conversation about the length of Duke's rotation if we can't agree on what "rotation" means.

Just to be clear, I think your definition is a perfectly reasonable and objective standard. My point was that I don’t think it necessarily reflects the way coaches or players think about being “in the rotation”. But I don’t have a better standard to offer up, because I think that the true definition tends to be a little subjective and idiosyncratic to particular teams.

[BD80]

...

I know you hate this topic. I was just trying to get it out of the 2019 recruiting thread and into it's own thread.

I don't hate the topic, I hate that it rears its head in each and every thread.

I do appreciate the effort to sequester the discussion into one thread.

I am bemused by the fact that you have authoritatively made your point, with supporting data, and yet the debate roils on ...

[budwom]

The thread is clearly labeled, so those who don't want to participate can easily avoid it, so I'm not sure why one could object. And the data appear to be sound.

[CDu]

Personally, my feelings on the matter have evolved over time to where I really don’t care about defining what is or is not “in the rotation”. Ultimately, “in the rotation” is a largely subjective term that comes up (a) tangentially in discussing minutes or (b) when someone complains about Coach K shortening the bench in tight games. In the case of (a) the term really doesn’t matter because we are discussing minutes. In the case of (b) it doesn’t matter because Coach K is gonna do whatever Coach K does.

One can make an objective argument using Kedsy’s approach looking at competitive games. One can make an objective argument based average minutes played for the season. One can make an objective argument based on total games played on the season. One could make an objective argument on the basis of ANY games played in a season. And all of them can be correct in the eye of the beholder, because the term is subjective.

But we almost never are actually debating whether or not so-and-so is or is not in the rotation. It is pretty much only ever a tangential aspect of a discussion. Occasionally, we discuss it as an abstract construct of how deep Coach K’s bench is or has been. So it doesn’t really seem to matter to me what the definition is. And, sense folks aren’t likely to ever uniformly agree on what the definition is, I try not to worry too much about it.

The only time it comes up for me is offhand in minutes prognostication. But for me, in those circumstances I really just care about the minutes estimates. I know that Coach K tends to play just 6-7 guys significant minutes in tight games. I know that we also typically end up with 8-10 guys averaging 6+ mpg on the season, and 8 guys averaging ~10+ mpg on the season in aggregate. Beyond that, “in the rotation” to me is just noise.

[Bluedog]

Wow, major props for putting all that together. Interesting that the much derided Rivers-led 11-12 squad had among the deepest rotation under K (basically equivalent with the '08 and early 90s team). Not that I want to add work to your plate, but I wonder how this would compare to other teams. It's not like Duke is short of talent compared to most teams so it's clearly a coaching philosophy. In particular, I'd be curious how it compares to UNC (perhaps just a couple seasons there...) as Roy gets the flak from his fanbase that he plays the bench too much such that the best players don't have enough time on the court.

I personally think K's rotation is pretty optimal in trying to win each and every game. (And is what is optimal for the NCAA tournament.) A deeper bench might be nice to help other players develop their game more and "keep them happy" which could pay dividends in future seasons or in case of extreme foul trouble.

[Kedsy]

Not that I want to add work to your plate, but I wonder how this would compare to other teams. It's not like Duke is short of talent compared to most teams so it's clearly a coaching philosophy. In particular, I'd be curious how it compares to UNC (perhaps just a couple seasons there...) as Roy gets the flak from his fanbase that he plays the bench too much such that the best players don't have enough time on the court.

Two seasons of UNC box scores is all I could bear to go through before my stomach curdled:



UNC 5 6 7 8 9 AVG
2017 0 0 6 12 2 7.80
2015 0 1 5 11 5 7.91


I chose their championship season to see if Roy played a shorter rotation than normal, and it was a little shorter than 2015 but not much. Either way, it seems Roy plays at least a full player more than K in his rotation (again, using my definition).

[House P]

Since K's first season in 1981 (not including 1995, because Coach K didn't coach the team after January 1 that season), we have played 8+ players for 10+ minutes only 20% of close games after January 1. Since 2000, we've played 8+ players for 10+ minutes less than 16% of post-January 1 close games. In the last four seasons, we've only had four games in which 8 or more players played 10+ minutes in such games.

Thanks for the compelling evidence that the 8th man for Coach K doesn’t typically play 10 minutes in competitive games after Jan 1. This brings up a related question:

“If the 8th man doesn’t typically play 10 minutes (in close games after Jan 1), how much does the 8th man typically play?”

To find out, I used your definition of a close game (final margin of less than 20 points) and calculated the average minutes per game each year for the 8th man in close games after Jan 1. Here is what I found.

It seems that, in a “typical year” during Coach K’s tenure at Duke, the 8th man plays about 6.5 minutes in close games after Jan 1.

Using the same approach, the 9th man plays 1.9 minutes in a typical year in close games after Jan 1.


So, if you use 10 minutes as a threshold for being “in the rotation”, these results are pretty consistent with the idea that Coach K has most commonly played a 7.5 man rotation. In 27 of Coach K’s 38 seasons at Duke, the 8th man averaged between 4.5 and 9.5 minutes per game (in close games after Jan 1).

Here is a plot showing the average mpg that the 8th man (in blue) and 9th man (in green) played each year in close games after Jan 1. The solid line represent the median for each spot on the depth chart. By the way, when I used the term “typical year” above, I am referring to the median of all the yearly values between 1981 and 2018. So 50% of the seasons were above the "typical year" and 50% were below.

http://forums.dukebasketballreport.com/forums/attachment.php?attachmentid=8440&stc=1

A couple notes:

- The 2007 team probably comes the closest in representing a “typical year” rotation-wise for Coach K. In 2007 the 7th man averaged 12.5 min, the 8th man averaged 6.4 min and the 9th man averaged 1.9 min in close games after Jan 1. These are all within a couple tenths of the median for each rotation spot over the course of Coach K’s tenure at Duke


- The 2008 team was probably Coach K’s deepest team. In close games that season after Jan 1, the 8th man averaged 10.3 mpg, the 9th man averaged 4.8 mpg, and the 10th man averaged 2.4 mpg. In 13 of 20 close games that year, at 10 or more guys played multiple minutes.


- The 2016 team almost certainly represents the shortest bench for Coach K. In 22 close games that season after Jan 1, Coach K gave a total of 3 minutes to guys outside the top 7 on the depth chart (1 min for Sean Obi, 1 for Antonio Vrankovic, and 1 for Nick Pagliuca).


- While Coach K almost never plays 9 guys for significant (8+ min) minutes in close late seasons games, there have been a handful of seasons where the 9th man regularly played 3-5 minutes in such games. This appears to be less common since 2000 and has only happened twice since 2010.


- As Kedsy noted, it is reasonable to exclude the Guadet-coached games from 1995. Who knows what Coach K would have done with that team, but the 1995 team was trending towards a deeper-than-average team even before Coach K stepped away. That season, the 8th man played between 11-15 minutes in 6 of the 7 close games with Coach K on the bench. If you exclude 1995, the median yearly mpg for the 8th man in close games after Jan 1 is 6.4.

[HereBeforeCoachK]

I personally think K's rotation is pretty optimal in trying to win each and every game. (And is what is optimal for the NCAA tournament.) A deeper bench might be nice to help other players develop their game more and "keep them happy" which could pay dividends in future seasons or in case of extreme foul trouble.

Love K, agree he's the GOAT, period...end of discussion....but even the GOATS in every pursuit have blind spots. I respectfully submit the theory that depth might be his. You don't develop deeper bench merely to keep them happy, though that's important, you do it so when your two best 3 point shooters go stone cold in a regional final, you have someone ready to come in and take big minutes potentially.

Grayson's 16 versus Wisconsin doesn't happen without his minutes extending late in the season. As it was, it was kind of a fluke given how little he had played all year.