Recruiting ranking, experience, and playing time at Duke

[Kedsy]

Every year, debate rages here at DBR over who will find playing time at Duke. The high level of talent we enjoy on an annual basis combined with Coach K's propensity for a short rotation makes it natural message board fodder. Those who believe the Michael Gbinijes of the world will set college basketball on fire point to players like Jon Scheyer (#28 in the RSCI in 2006), who started and played big minutes as a freshman. Those who feel the hotshot newcomers will have to wait their turn point to players like Michael Gbinije himself, who despite being a highly regarded recruit played just 3.3 mpg as a freshman and only 0.9 mpg once ACC play began.

So who's right? Is there even an answer to this question? I say there is, that we can with reasonable accuracy predict who will be in Duke's rotation based on a relatively simple formula.

First, here are my assumptions:

(1) Coach K generally plays a 7-man rotation, consisting of four perimeter players and three bigs (C/PF). Thus, he will play the "best" four perimeter players and the "best" three bigs.

(2) Who is "best" is determined by a blend of talent and experience. While Coach K clearly does not make decisions based on recruiting rankings, such rankings are the best evaluation of "talent" that we fans possess.

(3) Experience is based on how many years a player has been in college, including redshirt years since a player does mature both physically and basketball-wise during a redshirt year.

(4) Coach K will occasionally go to an 8-man rotation if we have five outstanding perimeter options. It is also possible to see an 8-man rotation if two of the top three interior options more or less can only play center (since the RSCI was invented in 1998, this has only happened once, in 2010, but it looks like it may happen again next season, so I'm mentioning it here).

(5) The recruiting rankings aren't perfect. It would be silly to argue that the #21 recruit is significantly better than the #23 recruit. But we have to group them somehow, and on a quantum level we can make some broad statements, e.g., in general a top 10 recruit is a better player than a guy rated 11 to 20 (even while admitting that a #10 recruit may or may not be better than a #11 recruit, we have to draw lines somewhere).

(6) It's easier to predict who will be in the rotation as opposed to who will start or the exact order of who will get the most-to-least minutes (e.g., who will gain the 3rd-most minutes vs. the 6th-most minutes). Because the "best" players will probably make the rotation, but who starts, etc., is partially governed by ability to blend with teammates' skillsets, factors like conditioning, and actual on-court performance, none of which is necessarily susceptible to prediction without actually seeing practice, etc. In other words, we should be able to predict with reasonable accuracy whether Grayson Allen will get rotation minutes, but it will be almost impossible to say with any certainty whether Justise Winslow starts or is the 6th or 7th man (at least until Coach K speaks on the subject, and maybe not even then).


OK, here's the formula:

Freshmen players are assigned a number from 1 to 4 (lower being better), based on their RSCI recruting ranking, as follows:

1 to 10: 1
11 to 20: 2
21 to 35: 3
36+: 4

Non-freshmen players take their freshman number and subtract half a point (0.5) for each year they've been in college. Redshirt years count as a year in college, although if the player is away from the team (like Andre Dawkins in 2013), this could be debatable. A redshirt year due to a transfer counts an extra half point, because if the player hadn't exceeded the expectations of his recruiting ranking, Coach K probably wouldn't have accepted him as a transfer.

So, for example, Amile Jefferson was #21 in the RSCI in his senior year of high school (2012), so he started as a 3.0. In 2014-15, his junior year of college, he'll be a 2.0, because two half points will be deducted for his two years spent in college.

Assuming a 7-man rotation, the theory is that the seven guys who play the most minutes will be the perimeter players with the four lowest numbers and the interior players with the three lowest numbers. If we have five perimeter guys with numbers of 2.5 or lower, the rotation should be 8 guys (5 perimeter, 3 interior). This has happened four times in the past 15 years (2008, 2009, 2011, 2014).

In cases of ties, I generally go with the player who has more experience. If two players from the same class are tied for the last spot in the rotation, it's impossible to predict in advance which will be chosen, but if history is our guide, whoever wins the competition will play rotation minutes and the other won't play very much.



Let's see how this theory has worked in the past, going back to the 1999-2000 season:

1999-2000:

Predicted perimeter rotation: Jason Williams (1.0), Nate James* (1.5), Chris Carrawell* (2.5), Mike Dunleavy (3.0)
Predicted interior rotation: Shane Battier* (0.0), Carlos Boozer (1.0), Casey Sanders (2.0)

Other perimeter players: Andre Buckner** (5.0)
Other interior players: Matt Christensen* (3.5), Nick Horvath (4.0)

Actual perimeter rotation: J Williams (1157 minutes for season), N James (970), C Carrawell (1212), M Dunleavy (724)
Actual interior rotation: S Battier (1206), C Boozer (807), M Christensen (295)

Others: N Horvath (269), C Sanders (144), A Buckner (68)

* - since RSCI didn't exist before the 1998 high school seniors, I had to estimate Battier, Carrawell, James, and Christensen
** - since Andre Buckner was basically a walk-on with a scholarship, I counted him as a "5," rather than a "4."

OK, in the first year we look at, the formula failed for the 8th man, as both Matt Christensen and Nick Horvath beat out Casey Sanders. It's arguable, since Christensen didn't even play as much as 10 mpg, that we only had a 6-man rotation, with the six players the formula predicted, but instead I'm going to say this season gives us an exception to the rule.


2000-01:

Predicted perimeter rotation: Jason Williams (0.5), Nate James* (1.0), Chris Duhon (1.0), Mike Dunleavy (2.5)
Predicted interior rotation: Shane Battier* (-0.5), Carlos Boozer (0.5), Casey Sanders (1.5)

Other perimeter players: Andre Buckner** (4.5)
Other interior players: Matt Christensen* (3.0), Reggie Love (4.0)

Actual perimeter rotation: Jason Williams (1246 minutes for season), Nate James (1085), Chris Duhon (1085), Mike Dunleavy (1137)
Actual interior rotation: Shane Battier (1363), Carlos Boozer (820), Casey Sanders (373)

Others: Matt Christensen (253), Reggie Love (130), Andre Buckner (83)

Exactly as predicted


2001-02:

Predicted perimeter rotation: Jason Williams (0.0), Chris Duhon (0.5), Dahntay Jones (2.0), Daniel Ewing (3.0)
Predicted interior rotation: Carlos Boozer (0.0), Casey Sanders (1.0), Mike Dunleavy (2.0)

Other perimeter players: Andre Buckner** (4.0)
Other interior players: Matt Christensen* (2.5), Nick Horvath (3.0), Reggie Love (3.5)

Actual perimeter rotation: Jason Williams (1175), Chris Duhon (1229), Dahntay Jones (1012), Daniel Ewing (636)
Actual interior rotation: Carlos Boozer (993), Mike Dunleavy (1134), Nick Horvath (247)

Others: Casey Sanders (242), Matt Christensen (133), Reggie Love (59)

Again Casey Sanders got beat out for 7th man, although again, the guy who beat him out played fewer than 10 mpg, and this time if Casey had received six (6) more minutes of playing time the rotation would have been properly predicted. Still, he didn't get six more minutes, so here's another exception.


2002-03:

Predicted perimeter rotation: Chris Duhon (0.0), JJ Redick (2.0), Daniel Ewing (2.5), Sean Dockery (3.0)
Predicted interior rotation: Casey Sanders (0.5), Shelden Williams (1.0), Dahntay Jones (1.5)

Other perimeter players: Andre Buckner (3.5), Lee Melchionni (4.0)
Other interior players: Shavlik Randolph (2.0), Nick Horvath (3.0), Michael Thompson (3.0)

Actual perimeter rotation: Chris Duhon (1188), JJ Redick (1013), Daniel Ewing (920), Sean Dockery (345)
Actual interior rotation: Casey Sanders (589), Shelden Williams (633), Dahntay Jones (1014)

Others: Nick Horvath (405), Shavlik Randolph (351), Lee Melchionni (71), Mike Thompson (59), Andre Buckner (37)

It's not entirely clear whether Dahntay Jones was a perimeter or an interior player in 2002-03. However, since none of the obvious big men on the team earned more than 633 minutes, I think we'll have to call Dahntay an interior player for rotation purposes.

That said, this was a very odd year for a Coach K Duke team, in which 9 players earned more than 10 mpg for the season. Note that despite nine players averaging more than 10 mpg, we didn't actually have a 9-man rotation. We played 23 games that season after January 1 in which the score was within 20 points; nine guys got 10 or more minutes in only one of those games and eight guys got 10 or more minutes in only two other games. In three games only six guys played 10 or minutes, and in the remaining 17 games we had seven guys playing 10 or more minutes. So really, due to Dahntay's ability to play both the perimeter and interior, what really happened was Coach K tried lots of different 7-man combinations, and so it's hard to say which of Dockery or Randolph or Horvath was the 7th man.


2003-04:

Predicted perimeter rotation: Chris Duhon (-0.5), JJ Redick (1.5), Daniel Ewing (2.0), Sean Dockery (2.5)
Predicted interior rotation: Shelden Williams (0.5), Luol Deng (1.0), Shavlik Randolph (1.5)

Other perimeter players: Lee Melchionni (3.5)
Other interior players: Nick Horvath (2.5)

Actual perimeter rotation: Chris Duhon (1311), JJ Redick (1152), Daniel Ewing (1131), Sean Dockery (571)
Actual interior rotation: Shelden Williams (963), Luol Deng (1149), Shavlik Randolph (709)

Others: Nick Horvath (218), Lee Melchionni (145)

Exactly as predicted.


2004-05:

Predicted perimeter rotation: JJ Redick (1.0), Daniel Ewing (1.5), Sean Dockery (2.0), DeMarcus Nelson (2.0)
Predicted interior rotation: Shelden Williams (0.0), Shavlik Randolph (1.0), Lee Melchionni (3.0)

Other perimeter players: none
Other interior players: Reggie Love (3.0???), Dave McClure (4.0)

Actual perimeter rotation: JJ Redick (1231), Daniel Ewing (1138), Sean Dockery (720), DeMarcus Nelson (634)
Actual interior rotation: Shelden Williams (1109), Shavlik Randolph (548), Lee Melchionni (716)

Others: Reggie Love (215), Dave McClure (184)

Reggie Love hadn't played organized basketball in three years, so I wasn't sure how to grade him. He wasn't really recruited for basketball anyway, so since they were tied I wouldn't have expected him to beat out a junior Melchionni.


2005-06:

Predicted perimeter rotation: JJ Redick (0.5), Sean Dockery (1.5), DeMarcus Nelson (1.5), Greg Paulus (2.0)
Predicted interior rotation: Shelden Williams (-0.5), Josh McRoberts (1.0), Lee Melchionni (2.5)

Other perimeter players: Martynas Pocius (4.0)
Other interior players: Eric Boateng (4.0), Jamal Boykin (4.0)

Actual perimeter rotation: JJ Redick (1336), Sean Dockery (1069), DeMarcus Nelson (517), Greg Paulus (1163)
Actual interior rotation: Shelden Williams (1198), Josh McRoberts (883), Lee Melchionni (717)

Others: Martynas Pocius (172), Eric Boateng (50), Jamal Boykin (70)

Exactly as predicted, despite three intriguing freshmen (rated #39, #53, and #60). Here's an example of an upperclassman (Lee Melchionni) beating out three freshmen who had higher recruiting rankings than he did, due to his experience. If we'd had a 6'7 freshman rated #18 (and thus earning a 2.0), such a player very well could have beaten out Lee for the 3rd big slot, but guys with a 4.0 score weren't able to do it.


2006-07:

Predicted perimeter rotation: DeMarcus Nelson (1.0), Gerald Henderson (1.0), Greg Paulus (1.5), Jon Scheyer (3.0)
Predicted interior rotation: Josh McRoberts (0.5), Lance Thomas (2.0), Dave McClure (3.0)

Other perimeter players: Marty Pocius (3.5)
Other interior players: Brian Zoubek (3.0, less experience than Dave McClure)

Actual perimeter rotation: DeMarcus Nelson (1052), Gerald Henderson (618), Greg Paulus (1068), Jon Scheyer (1112)
Actual interior rotation: Josh McRoberts (1164), Lance Thomas (463), Dave McClure (716)

Others: Brian Zoubek (235), Marty Pocius (193)

Exactly as predicted. Jon Scheyer's #28 RSCI rank probably wouldn't have gotten him a rotation spot in most years, but in 2006-07, his 3.0 was the 4th best perimeter score. Brian Zoubek had a solid #25 RSCI rank, but it only earned him a tie with Dave McClure (in his 3rd year in the program) and McClure had more experience.


2007-08:

Predicted perimeter rotation: DeMarcus Nelson (0.5), Gerald Henderson (0.5), Greg Paulus (1.0), Nolan Smith (2.0), Jon Scheyer (2.5)
Predicted interior rotation: Kyle Singler (1.0), Lance Thomas (1.5), Dave McClure (2.5)

Other perimeter players: Taylor King (3.0)
Other interior players: Brian Zoubek (2.5)

Actual perimeter rotation: DeMarcus Nelson (1051), Gerald Henderson (891), Greg Paulus (943), Nolan Smith (500), Jon Scheyer (963)
Actual interior rotation: Kyle Singler (972), Lance Thomas (593), Dave McClure (268)

Others: Taylor King (330), Brian Zoubek (262)

Exactly as predicted. This was a year that we had five outstanding (2.5 or better) perimeter options, so it's one of the few years we played an 8-man rotation. Again, McClure and Zoubek were tied (and McClure still had more experience), but the gap between them narrowed.


2008-09:

Predicted perimeter rotation: Gerald Henderson (0.0), Greg Paulus (0.5), Nolan Smith (1.5), Jon Scheyer (2.0), Elliot Williams (2.0)
Predicted interior rotation: Kyle Singler (0.5), Lance Thomas (1.0), Dave McClure (2.0)

Other perimeter players: Marty Pocius (2.5)
Other interior players: Brian Zoubek (2.0), Miles Plumlee (4.0), Olek Czyz (4.0)

Actual perimeter rotation: Gerald Henderson (1098), Greg Paulus (578), Nolan Smith (734), Jon Scheyer (1214), Elliot Williams (563)
Actual interior rotation: Kyle Singler (1193), Lance Thomas (689), Dave McClure (568)

Others: Brian Zoubek (427), Miles Plumlee (165), Marty Pocius (135), Olek Czyz (51)

As predicted. Again, we had five outstanding perimeter options, although in reality we never had an 8-man rotation (rather Elliot Williams, 5th on the perimeter totem pole, displaced Greg Paulus for the fourth perimeter slot, although for the season Paulus still had more minutes). Zoubek and McClure were still tied and once again experience prevailed, though Zoubek's minutes continued on the way up. Marty Pocius was down to a 2.5, but that still made him the sixth perimeter option.


2009-10:

Predicted perimeter rotation: Kyle Singler (0.0), Nolan Smith (1.0), Jon Scheyer (1.5), Andre Dawkins (3.0)
Predicted interior rotation: Lance Thomas (0.5), Brian Zoubek (1.5), Mason Plumlee (2.0) OR Ryan Kelly (2.0) (but not both)

Other perimeter players: none
Other interior players: Miles Plumlee (3.5)

Actual perimeter rotation: Kyle Singler (1436), Nolan Smith (1349), Jon Scheyer (1470), Andre Dawkins (477)
Actual interior rotation: Lance Thomas (1013), Brian Zoubek (746), Miles Plumlee (654)

Others: Mason Plumlee (480), Ryan Kelly (227)

You could argue that we played a 4-big rotation this season (8-man overall), with Mason Plumlee beating Ryan Kelly out of the tie for the 4th big spot, but that doesn't change the fact that Miles Plumlee leapfrogged both freshmen for the 7th spot, thus this season contains an exception to the rule. That said, I suspect the reason for this exception is Mason Plumlee's early season broken hand. It cost him the first six games of the season, and then he had to play catchup as a freshman. Considering how the staff was touting him in the pre-season, my guess is had he not gotten injured he would have won the third spot in the big rotation and this season would have gone exactly as predicted.


2010-11:

Predicted perimeter rotation: Kyle Singler (-0.5), Nolan Smith (0.5), Kyrie Irving (1.0), Andre Dawkins (2.5), Seth Curry (2.5)
Predicted interior rotation: Mason Plumlee (1.5), Ryan Kelly (1.5), Miles Plumlee (3.0)

Other perimeter players: Tyler Thornton (4.0)
Other interior players: Josh Hairston (3.0, less experience than Miles Plumlee)

Actual perimeter rotation: Kyle Singler (1286), Nolan Smith (1259), Kyrie Irving*** (303), Andre Dawkins (778), Seth Curry (924)
Actual interior rotation: Mason Plumlee (949), Ryan Kelly (743), Miles Plumlee (628)

Others: Tyler Thornton (337), Josh Hairston (165)

*** Kyrie obviously was a big part of the rotation while healthy.

Another season skewed by an injury. Again we see five outstanding perimeter options and a five man (counting Kyrie) perimeter rotation. So this season went as predicted as possible considering Kyrie's injury.


2011-12:

Predicted perimeter rotation: Austin Rivers (1.0), Seth Curry (2.0), Andre Dawkins (2.0), Quinn Cook (3.0) OR Michael Gbinije (3.0) (but not both)
Predicted interior rotation: Mason Plumlee (1.0), Ryan Kelly (1.0), Miles Plumlee (2.5)

Other perimeter players: Tyler Thornton (3.5)
Other interior players: Josh Hairston (2.5)

Actual perimeter rotation: Austin Rivers (1129), Seth Curry (1026), Andre Dawkins (760), Tyler Thornton (717)
Actual interior rotation: Mason Plumlee (964), Ryan Kelly (803), Miles Plumlee (697)

Others: Quinn Cook (387), Josh Hairston (246), Michael Gbinije (111)

Gbinije and Cook were tied and Cook beat Silent G out, and for awhile we played an 8-man (5-perimeter) rotation, despite Quinn's number being not quite as good as we usually needed for that. But in the end, Tyler Thornton leapfrogged them both and thus this season includes an exception. Again Miles's experience advantage broke his tie with Josh.


2012-13:

Predicted perimeter rotation: Seth Curry (1.5), Rasheed Sulaimon (2.0), Quinn Cook (2.5), Tyler Thornton (3.0)
Predicted interior rotation: Mason Plumlee (0.5), Ryan Kelly (0.5), Josh Hairston (2.0)

Other perimeter players: Alex Murphy (3.5)
Other interior players: Amile Jefferson (3.0), Marshall Plumlee (3.5)

Actual perimeter rotation: Seth Curry (1130), Rasheed Sulaimon (1050), Quinn Cook (1208), Tyler Thornton (791)
Actual interior rotation: Mason Plumlee (1248), Ryan Kelly (664), Josh Hairston (444)

Others: Amile Jefferson (405), Alex Murphy (194), Marshall Plumlee (50)

Exactly as predicted. Ryan Kelly's injury thrust the next guy in the pecking order (Amile) into the rotation for 13 games, which is why his minutes almost reached Josh's. Despite DBR clamoring for Alex and Marshall to get minutes, they clearly were the lowest rated.


2013-14:

Predicted perimeter rotation: Rasheed Sulaimon (1.5), Rodney Hood (1.5), Andre Dawkins (1.5), Quinn Cook (2.0), Tyler Thornton (2.5)
Predicted interior rotation: Jabari Parker (1.0), Josh Hairston (1.5), Amile Jefferson (2.5)

Other perimeter players: Matt Jones (3.0), Semi Ojeleye (3.0)
Other interior players: Marshall Plumlee (3.0)

Actual perimeter rotation: Rasheed Sulaimon (871), Rodney Hood (1150), Andre Dawkins (453), Quinn Cook (1042), Tyler Thornton (745)
Actual interior rotation: Jabari Parker (1073), Josh Hairston (284), Amile Jefferson (796)

Others: Marshall Plumlee (254), Matt Jones (235), Semi Ojeleye (80)

As predicted. I didn't move Andre up a half point for the year he spent away from the program, but he was still among five outstanding perimeter options and Coach K once again played all five in the rotation. Marshall almost but didn't quite catch Josh for the 8th most minutes. Despite DBR clamoring for Semi, Matt, and Alex (also 3.0) they were all tied for 7th among perimeter options and/or 4th among interior options.


Fifteen (15) seasons, 109 rotation spots, and only four (4) exceptions to the rule. And two of those four exceptions were due to Casey Sanders probably being overrated as a high school senior, while a third was likely due to injury. No system is perfect, but this one seems to cover most of the angles, explaining why some guys in the high 20s (e.g., Scheyer, Dunleavy, Ewing) played solid rotation minutes while others with similar or even better ratings (e.g., Gbinije, T King, Zoubek, Randolph, Kelly) didn't.



The new final RSCI just came out, so we can apply this theory to our upcoming 2014 rotation. Looks like Jahlil Okafor retained his #1 ranking, Tyus Jones came in at #7, Justise Winslow at #13, and Grayson Allen at #24. So what does our theory predict for the 2014-15 season? Let's take a look:

2014-15:

Predicted perimeter rotation: Rasheed Sulaimon (1.0), Tyus Jones (1.0), Quinn Cook (1.5), Justise Winslow (2.0)
Predicted interior rotation: Jahlil Okafor (1.0), Amile Jefferson (2.0), Marshall Plumlee (2.5)

Other perimeter players: Matt Jones (2.5), Grayson Allen (3.0)
Other interior players: Semi Ojeleye (2.5, less experience than Marshall Plumlee)

If we apply the "five outstanding perimeter player" rule, then Matt Jones (2.5) would appear in the predicted perimeter rotation (or if Semi Ojeleye is considered a perimeter player, since he's also 2.5 with the same experience as Matt, then possibly he could beat Matt out for that spot).

The only reason I didn't include the fifth perimeter guy is the fact that Marshall and Jahlil are both "pure" centers who are unlikely to be on the court together for more than a few minutes (if at all). Since Amile isn't going to play 40 minutes, presumably either Justise or Semi will have to fill the positional need of backup PF and, depending on which one it is, we may see a quirk. If it's Justise, then Matt (or possibly Semi) can fill in as the fifth outstanding perimeter option in an 8-man rotation. I guess it's also possible that Semi beats Marshall out for the third big man spot (they both rated at 2.5 but Marshall has been at Duke a year longer), in which case Matt could be the fifth outstanding perimeter guy and Marshall could be outside the rotation, looking in.

But if we have four bigs in the rotation, and the fourth big is Semi, then the "five outstanding perimeter option" rule will probably have to be ignored (meaning Matt Jones will not make the rotation), because it's very unlikely Coach K would go to a 9-man rotation. That said, we had a similar situation in 2003, so we might have a season like that in which Coach K tries a lot of different 7- or 8-man rotations before finally settling on one of them.

In the end, if the theory holds then we should be fairly confident that Rasheed, Tyus, Quinn, Justise, Jahlil, Amile and probably Marshall will be in the rotation. But Grayson Allen, despite earning a very respectable #24 in the RSCI, is very unlikely to be. There is likely to be an 8th guy earning around 10 mpg, and the only real question is whether that 8th man will be Matt or Semi.

Unless Coach K decides to completely abandon the substitution pattern he's used for at least the past 15 years. And while of course that's possible, I wouldn't bet on it.

[gumbomoop]

First-rate research, and not for the first time. Very valuable. Thank you.

Your conclusion re 2014-15 in general matches my intuitive sense. One's intuition -- mine and others -- surely does loosely incorporate the ratings and experience criteria that you detail.

I infer an unstated assumption: that 10 mpg is the floor for being "in the rotation." If so, I'd [intuitively] guess that, yes, there will be a battle for the eighth spot between Semi, Matt, and Grayson. Semi would win if Justise cannot give any backup minutes at the 4. Although I'd like to see a little experimenting with twin towers [Jahlil & Marshall] in pre-conference, I've no idea whether there will be any experimenting with that, and even if so, couldn't confidently predict the experiment would be so successful as to continue into ACC play. But absent some twin towers, someone has to back up Amile; and if Justise is no go there, Semi would surely play 10+ mpg and be in rotation.

But if Justise plays both some perimeter [wing/3] and some interior [stretch 4], that would move Matt into role of favorite to be 8th man.

I do assume Marshall is 7th man.

As to the definition of rotation, if we expand it to 5-7 mpg rather than 10+, then I'd guess we'd see a 9-man rotation, with both Semi and Matt each getting 5+. Even if the 8th guy gets, say, a mere 7 mpg, with the 9-10 guys DNP after ACC begins, we'd probably consider 8th guy in rotation, yes?

Finally, I infer another unstated assumption: your 2014-15 prediction is mostly about 2015, i.e., once ACC play begins.

[Kedsy]

As to the definition of rotation, if we expand it to 5-7 mpg rather than 10+, then I'd guess we'd see a 9-man rotation, with both Semi and Matt each getting 5+. Even if the 8th guy gets, say, a mere 7 mpg, with the 9-10 guys DNP after ACC begins, we'd probably consider 8th guy in rotation, yes?

I would agree with your 8th guy/7mpg statement if we only count relatively close games (within 20 points) after January 1. My post above lists minutes for the season, and with a few exceptions the rotation guys all played 10 mpg in that context. It's possible the 7th guy (or 8th guy in the few years we played an 8-man rotation) sometimes dipped to 7 or so in ACC and tournament play.

But I would also note that while I didn't explicitly research this, if we only count close games after January 1, I don't think Coach K has ever used a 9-man rotation even if you go down to 5 to 7 mpg. After January 1, the 9th and 10th guys (and usually the 8th guy) have only played meaningful minutes in cases of injury and garbage time.

Finally, I infer another unstated assumption: your 2014-15 prediction is mostly about 2015, i.e., once ACC play begins.

For the most part, yes. Although if you look at pre-ACC games against top 25 opponents the pattern looks pretty similar to what it becomes in ACC play. Sometimes Coach K uses November/December for experimentation, but in large part the reason we see more guys getting bigger minutes is due to the quality of the opponent and the (actual or expected) margin of victory. And even then, the same basic rotation can be discerned early in the game while the score remains relatively close.

[gumbomoop]

... while I didn't explicitly research this, if we only count close games after January 1, I don't think Coach K has ever used a 9-man rotation even if you go down to 5 to 7 mpg. After January 1, the 9th and 10th guys (and usually the 8th guy) have only played meaningful minutes in cases of injury and garbage time.

... if you look at pre-ACC games against top 25 opponents the pattern looks pretty similar to what it becomes in ACC play. Sometimes Coach K uses November/December for experimentation, but in large part the reason we see more guys getting bigger minutes is due to the quality of the opponent and the (actual or expected) margin of victory. And even then, the same basic rotation can be discerned early in the game while the score remains relatively close.

Thanks for these further clarifications, which temporarily persuade me that my attempts [on behalf of countless numbers of EK posters] to "force" K into employing a longer rotation -- by re-defining "rotation" -- may be fruitless.

Let me ask you, and anyone else, to respond to my guess that the likely battle between Semi and Matt for 8th guy will depend on whether Justise plays exclusively at wing/3 or also plays some at stretch 4.

This "Semi v. Matt battle for 8th" assumes, of course, that Justise and Marshall are in the 7-rotation. And it assumes that, as talented as he is, Grayson may not fit the particular need for the 8th slot: strong D.

The only "Grayson-as-8th" scenario would seem to necessitate (1) Justise getting backup minutes at 4, (2) a greater need for 3-bombs than for D on perimeter, and (3) Grayson as superior 3-bomber to both Matt and Semi. Yes?

[sagegrouse]

Outstanding work, Kedsy!

I believe there will be at least five perimeter players in 2014-15 that have ten minutes or more. I have two reasons:

First, I started with your assumptions and may have ended up at a different place: Jahlil and Marshall will very rarely play at the same time; therefore, their combined minutes will be less than 40 (don't ever misunderestimate the ability of Coach K to go small). Then, there will be 160+ minutes left to be shared. If it is primarily shared by five (Amile and four perimeter players), these five will average close to 30 MPG. It is much more likely that six players will get at least 10-15+ minutes per game.

Second, I believe Matt Jones is gonna play a lot. He averaged 7.3 mins. with four starts this year. Plus, he looks like he's 30 years old, which bespeaks a certain physical maturity and which will be a factor in getting on the court for defense. The last freshman who looked 30 years old was Carlos Boozer (who was only 17 when he enrolled). Also, I expect his shooting will improve substantially with an off-season of work.

Third, we have again a team with a big guy in the middle -- really big -- and eight other mobile players to choose from. I wouldn't be surprised at the end of the season to see that nine players will have averaged ten minutes per game, as the coaching staff tries lots of different combinations before settling on an eight player rotation by February.

Great work again, Kedsy.

[Kedsy]

Let me ask you, and anyone else, to respond to my guess that the likely battle between Semi and Matt for 8th guy will depend on whether Justise plays exclusively at wing/3 or also plays some at stretch 4.

I'd agree it depends largely on this. If Justise can't play PF, then the 8th man will be a big and Semi's the only big left so he'd win the competition by default.

That said, even if Justise turns out to be the backup PF, then Matt and Semi have the same "rating" (2.5) and it should really be a tossup. The only reason I'd assume Matt would win this competition is the fact that Matt played 3x as many minutes as Semi did in 2013-14, so he'd appear to have a big edge.

You never know how much improvement happens in the off-season, though. So while I expect Matt to beat out Semi for the 8th spot, my "system" doesn't give us any insight into this particular position battle.

One last thing on this, though. In 2009-10, Mason Plumlee and Ryan Kelly were "tied" with a 2.0 rating. Mason won the competition and played around 14 mpg. Ryan played only 6.5 mpg and hardly played at all after ACC play began. This was not an isolated instance, either. The player who loses the tie for the last rotation spot generally ends up all the way out.

This "Semi v. Matt battle for 8th" assumes, of course, that Justise and Marshall are in the 7-rotation. And it assumes that, as talented as he is, Grayson may not fit the particular need for the 8th slot: strong D.

If the theory espoused in this thread has validity, then Justise will be in the rotation. The only freshman ranked in Justise's ballpark that didn't make the rotation was Casey Sanders, and I'd be shocked if Justise had issues similar to Casey. Well, also Ryan Kelly, but the system predicted Ryan would only play if he could beat out Mason for 3rd big. This coming year the system predicts Justise safely as the 4th perimeter guy.

As for Marshall, the predictive system I've employed has he and Semi tied, so I guess it could be up in the air. But every tie over the past 15 years has gone to the player with more experience in the Duke system (with the possible exception of Reggie Love/Lee Melchionni, but Reggie hadn't played organized basketball in a couple years so that was a special case), and that would suggest Marshall has the upper hand over Semi (due to Marshall's redshirt year). That plus the fact that Marshall is more appropriately sized for the position. So it seems a reasonably safe assumption.

The only "Grayson-as-8th" scenario would seem to necessitate (1) Justise getting backup minutes at 4, (2) a greater need for 3-bombs than for D on perimeter, and (3) Grayson as superior 3-bomber to both Matt and Semi. Yes?

I honestly don't think there's any realistic scenario in which Grayson earns 8th man in the rotation. If you count Semi as a perimeter player (and if the theory in this thread has any merit), then Grayson is 7th in the perimeter rotation. Unless he comes out in October playing like a potential All ACC player (as a freshman), there are simply too many people ahead of him for him to get meaningful, non-garbage time minutes.

It's interesting to note that (counting Semi as a potential perimeter player) this will be only the 2nd season we've had six perimeter players rated 2.5 or better. The first one was 2008-09, and Marty Pocius (the 6th best perimeter guy at 2.5) only played 135 minutes for the season. I'd imagine a 7th perimeter option (like Grayson) will not even play that much. Even if Semi doesn't count as a perimeter player, meaning Grayson is the 6th perimeter option, it would make sense to look at Pocius-in-2009 as the guiding precedent.

[Mike Corey]

Just a tremendous post. The interest the content surely has for us as readers aside, I consider us lucky to have you committing the time and the thought into a product like this. Thanks, Kedsy.

[Kedsy]

I believe there will be at least five perimeter players in 2014-15 that have ten minutes or more.

I agree we should have five perimeter players with 10+ mpg. The system predicts that we will. The only wrench would be if Justise isn't capable of playing PF, and as you say Coach K loves to go small if he can so the odds of Justise-as-backup-PF seem reasonably good.

But if Justise can't or doesn't play backup PF, then I think the 8-man rotation will include four bigs (including Semi) like it did in 2010.

I also agree with you that we may have nine guys who average 10+ mpg (like we did in 2002-03), while really only playing a 7- or 8-man rotation.

[Mtn.Devil.91.92.01.10.15]

Just a tremendous post. The interest the content surely has for us as readers aside, I consider us lucky to have you committing the time and the thought into a product like this. Thanks, Kedsy.

Agreed. Post of the year.

And, now that Kedsy has cracked the code, I can save hours this off-season by not reading threads about line ups and minutes!

You, sir, are a gentleman and a scholar. See you in November!

[Billy Dat]

Great work.

I also supports the "Duke is a meritocracy" story that will continue to support the recent recruiting bonanza - if you are good enough, you will be in the rotation no matter your level of experience.

I wonder if any of K's talk about "doing things differently" will have consequences for one of his sacred cows...his 7-8 man rotation?

[Li_Duke]

Great analysis, and thanks for putting in all the time on this. Now I can start projecting this to the 2015-2016 season and beyond! (Only partially kidding.)

[Reilly]

What is the correlation between recruiting ranking, ultimate minutes played, and the decision to transfer?

I'm guessing there's the age-old correlation between minutes played and transferring ... wonder how expectations (the player's own, based on recruiting rankings) might play into it.

[Kedsy]

What is the correlation between recruiting ranking, ultimate minutes played, and the decision to transfer?

I'm guessing there's the age-old correlation between minutes played and transferring ... wonder how expectations (the player's own, based on recruiting rankings) might play into it.

An interesting question. If you'd equate the above analysis with reasonable expectations, then none of the players who transferred from Duke in the period should have reasonably expected to play more than they did, with the possible exception of Michael Gbinije, IF he realistically thought he was more deserving of playing time than *both* Quinn Cook and Tyler Thornton.

On the other hand, if the analysis is accurate and a player understood the pecking order, it's possible that player might choose to go elsewhere because he could read the writing on the wall. For example, if Alex Murphy had stayed he'd have been a 2.5, tied with Marshall and Semi for 3rd interior player and tied with Matt and Semi for 5th perimeter player. Under that scenario, the best he could hope for would be around 10 mpg, and it could have been less. His decision to transfer could have been less about unrealistic expectations ("I should be playing more") and more about entirely realistic expectations ("I'm not going to play so much and I'd like to"), although which of those views is consistent with his transferring to Florida (where he might not play so much anyway) is admittedly up for debate.

[COYS]

Just incredible work, Kedsy.

This should become a sticky that anyone who wants to post about minutes/playing time must read, first.

What's really impressive is that even the seasons that contain exceptions aren't all that extreme and only involve the last guy in the rotation. It also shows what an interesting case Tyler Thornton was, especially since he was rated outside of the RSCI 100. I know you haven't awarded an extra point for an unranked recruit, but if you throw an extra point onto Tyler's ranking, he becomes an even bigger exception, which might more accurately portray his time at Duke. I'd be willing to bet that we are never going to see another unranked recruit get as much time in the rotation over four seasons as Tyler Thornton.

The other really interesting point is that there aren't really that many outliers. The big recruits were stars, instantly. The lowest recruits needed to wait their time to crack the rotation. Similarly, very few of the recruits significantly underperformed. Casey Sanders was the notable one you already mentioned. Andre and Josh were probably close to falling in that category, too, but they never fell all the way out of the rotation.

I guess the question is, does this mean the staff does a really good job evaluating players before they enroll at Duke or does it mean that Duke has either been unlucky or unable to find a diamond in the rough? I'd wager that the staff must do a truly incredible job evaluating the talent level of recruits in high school. Casey Sanders was one notable miss, at least based on rankings. However, the top 10 guys all became stars. The top 20 guys all became either stars or major contributors, eventually, and the rest of the group fit in as role players if they stuck around for a few years (or happened to be freshman in a year when they were needed on the perimeter).

Jon Scheyer looks like a bit of an exception to that, since he became a star and held his own even in seasons when Gerald and Nolan looked to gain a bigger share of the minutes. I would argue, however, that this was due to Rivals inexplicably ranking him #76 in the nation while everyone else had him 18-22. His RSCI would probably have been top 20, eliminating a point in your system and ending up predicting his emergence as a star.

[COYS]

An interesting question. If you'd equate the above analysis with reasonable expectations, then none of the players who transferred from Duke in the period should have reasonably expected to play more than they did, with the possible exception of Michael Gbinije, IF he realistically thought he was more deserving of playing time than *both* Quinn Cook and Tyler Thornton.

On the other hand, if the analysis is accurate and a player understood the pecking order, it's possible that player might choose to go elsewhere because he could read the writing on the wall. For example, if Alex Murphy had stayed he'd have been a 2.5, tied with Marshall and Semi for 3rd interior player and tied with Matt and Semi for 5th perimeter player. Under that scenario, the best he could hope for would be around 10 mpg, and it could have been less. His decision to transfer could have been less about unrealistic expectations ("I should be playing more") and more about entirely realistic expectations ("I'm not going to play so much and I'd like to"), although which of those views is consistent with his transferring to Florida (where he might not play so much anyway) is admittedly up for debate.

As much as fans like to see the lower ranked guys excel and demand playing time, I think your analysis on transfers is spot on. Unless they leapfrog people early in their career (Tyler Thornton) or there are a lot of minutes at their position to go around either immediately or in the very next season, Duke recruits at such a high level that a lower-rated recruit's position is likely to be filled by someone in the 1.0 or 2.0 point range at some point in their career OR the recruits ahead of them stick around and stay ahead of the lower rated guys.

Alex Murphy would have been a 2.5 this coming season had he stuck around, but that doesn't do him much good in terms of getting playing time, as he'd still be behind Justise on the perimeter based on these predictions (It didn't seem as if Alex were ever really considered among the post players, even though he had the size to play the four at Duke). I don't blame the guy for transferring.

[Kedsy]

...or does it mean that Duke has either been unlucky or unable to find a diamond in the rough?

Well, interestingly (and predictably, since why otherwise would we accept a transfer?), all of our incoming transfers in the period would (in my opinion) qualify as diamonds in the rough (which is why I added an extra half point for incoming transfer's redshirt year):

Dahntay Jones, #98
Seth Curry, unranked
Rodney Hood, #27
Sean Obi, unranked

So, maybe we're just using the rest of college basketball as our farm team. Or, more seriously, it's probably a lot easier to find "proven" diamonds in the rough through the transfer process than it is to find them in high school.

[Li_Duke]

Duke has either been unlucky or unable to find a diamond in the rough?

I'd argue that we just don't have many opportunities to find "diamonds in the rough". We don't recruit many players outside the top 35, and those we do target almost always shoot up the ranking shortly after.

[COYS]

I'd argue that we just don't have many opportunities to find "diamonds in the rough". We don't recruit many players outside the top 35, and those we do target almost always shoot up the ranking shortly after.

This is a good point. Without going into an analysis comparable to what Kedsy did, Duke recruits do seem to jump up the list, although that often happens before they even commit. I remember being disappointed back in 2009 when it looked like interest had cooled a bit with number 1 point guard Brandon Knight (remember when getting a point guard was the ONLY THING THAT MATTERED on the board? See: John Wall Recruiting thread), but that's because Duke had zeroed in on , at the time, top 20 point guard Kyrie Irving. One could make the argument that there are some instances where Duke does find a diamond in the rough, it's just that the staff finds them in high school before everyone else does.

[tommy]

Fantastic work, as usual, Kedsy. Major kudos.

One thing it made me think about is that maybe the recruiting analysts deserve a little more credit than they receive in some places. Coach K (and others) of course aren't making playing time decisions based on where a kid is ranked in his class. He's doing it based on how the kid performs on the practice floor (and in games) once he's at Duke and competing at that level. But if the apportioning of playing time is able to be correlated, with some of the variables you identify, in some way to high school ranking, well then it stands to reason that those rankings usually have a fairly high degree of legitimacy. Casey Sanders appears to be the most glaring example of a Duke guy who was way overrated by all or almost all the analysts, but most of the rest seem to have had their PT at Duke roughly correlated with their high school ranking, when experience of the given player and the roster as a whole are considered. Yes, there is the fact that kids who get recruited, especially successfully recruited, by Duke (and other elite programs) often see their class rankings rise, but still, I think your analysis, in the main, says some good things about the recruiting analysts whose ratings have gone into the RSCI over the years.

[Des Esseintes]

Fantastic work, as usual, Kedsy. Major kudos.

One thing it made me think about is that maybe the recruiting analysts deserve a little more credit than they receive in some places. Coach K (and others) of course aren't making playing time decisions based on where a kid is ranked in his class. He's doing it based on how the kid performs on the practice floor (and in games) once he's at Duke and competing at that level. But if the apportioning of playing time is able to be correlated, with some of the variables you identify, in some way to high school ranking, well then it stands to reason that those rankings usually have a fairly high degree of legitimacy. Casey Sanders appears to be the most glaring example of a Duke guy who was way overrated by all or almost all the analysts, but most of the rest seem to have had their PT at Duke roughly correlated with their high school ranking, when experience of the given player and the roster as a whole are considered. Yes, there is the fact that kids who get recruited, especially successfully recruited, by Duke (and other elite programs) often see their class rankings rise, but still, I think your analysis, in the main, says some good things about the recruiting analysts whose ratings have gone into the RSCI over the years.

Great point. Everyone rags on the rankings. We remember the busts, and we remember the guys who came from nowhere, while conveniently forgetting the vast majority who were rated pretty accurately. As you say, it's not the be-all end-all, but of course no recruiting expert would ever claim otherwise. It's merely the very best starting point we have for team evaluation.