Cumulative Plus/Minus

[mr. synellinden]

As we're past the midway point in the season with a statistical sample of 21 games, it's interesting to me that our core group of players are all bunched in between 18.3 and 19.7 in terms of plus/minus on a 40 minute basis. Of course, McClure is the exception.

But I think that's somewhat to be expected, yet also impressive and it shows how deep and interchangeable we are. I think that's why we keep admiring how from one game to the next somebody different steps up and makes the big plays, scores the most points, etc. You only have to look at the MOTM diversity to see it. How many games this season would someone other than Hansboins;ocza be UNC's HOTM (Hole of the Match)?

[CDu]

As we're past the midway point in the season with a statistical sample of 21 games, it's interesting to me that our core group of players are all bunched in between 18.3 and 19.7 in terms of plus/minus on a 40 minute basis. Of course, McClure is the exception.

But I think that's somewhat to be expected, yet also impressive and it shows how deep and interchangeable we are. I think that's why we keep admiring how from one game to the next somebody different steps up and makes the big plays, scores the most points, etc. You only have to look at the MOTM diversity to see it. How many games this season would someone other than Hansboins;ocza be UNC's HOTM (Hole of the Match)?

The team as a whole has an average margin of victory of 19.4 ppg. So what we're seeing is that the guys who play the most minutes are regressing to the mean team average margin of victory. It makes sense: over a large enough sample, if you're playing most of the game, it's reasonable to expect your plus/minus to be fairly close to the team's overall plus/minus (i.e., margin of victory).

The interesting part is that we're basically seeing that the team performs similarly well with all of the regular guys out there. King and McClure, on average, probably get right around +19 per 40 minutes between them. Sort of saying the same thing you're saying I guess: that we're very well balanced.

[Jumbo]

Now updated through the B.C. game...

[ACCBBallFan]

Finally got a chance to rank-order by stat category and then sum as I had been doing a while back.

Ttl MP FG 3% FT RP AP BP SP PP Range Player
73 10 08 07 06 09 09 05 09 10 5-10 DeMarcus Nelson (2,-0))
69 09 07 06 07 10 05 09 07 09 5-10 Kyle Singler (1, -0))
63 08 03 8+ 10 07 08 04 08 07 3-10 Jon Scheyer (1,-0))
59 06 06 04 05 08 06 10 06 08 4-10 Gerald Henderson (1,-0)
58 07 02 10 09 03 10 01 10 06 1-10 Greg Paulus (3,-1)
40 04 05 05 08 01 07 2+ 04 04 1-08 Nolan Smith (0,-1)
39 05 09 02 03 05 01 6+ 05 03 1-09 Lance Thomas (0,-2)
38 03 01 8+ 04 04 3+ 6+ 03 05 1-8+ Taylor King (0,-1)
37 02 10 02 02 06 3+ 08 02 02 2-10 Brian Zoubek (1,-1)
16 01 04 02 01 02 02 2.5 01 01 1-04 David McClure (0,-5)

Very spread out as Paulus leads in 3 categories, Nelson in 2 and several others each have 1.

As far as least effective, that wold be McClure in 5 categories but his value may not show up in metrics alone. Among the rest, only Lance with 2 had more than 1 occurrence of worst on team.

Nelson and Singler were never rated lower than 5 out of 10 and only McClure with a personal high of 4 in FG%, never achieved at least an 8 out of 10 in some category. Gerald was never worse than 4 and Jon never worse than 3 of 10. Everybody else was loss leader in something.

Very close whether Gerald Henderson or Greg Paulus is fourth most effective after Demarcus Nelson, Kyle Singler and Jon Scheyer, and just as close among four guys for sixth most effective across Nolan Smith, Lance Thomas, Taylor King and Brian Zoubek.

Kind of a statistical representation of why it is hard to pick your poison when playing Duke.

Had to trim the column titles to get chart to align

Total MPG FG% 3FG% FT% RPG APG BPG SPG PPG Range Player

Total of All rankings across the 9 categories, so max score would be 90

MPG - Minutes per game
RPG - Rebounds per game
APG - Assists per gaem
BPG - Blocks per game
SPG - Steals per game
PPG - Points per game

Range - loswest to highest individual rank order across the 9 categories

[ACCBBallFan]

Jumbo, like others, I really appreciate the time you must spend doing those +/- and breakouts by combinations.

I tried to look at which players were part of any +20 or better combo and which were part of any -6 or worse combo.

From a cumulative perspective,

Singler was +6 -3 = +3
Lance was +3 - 1 = +2
Scheyer at +5 - 4 = +1
Paulus was +5 -5 = zero
Nelson was -4 - 4 = zero
Nolan was +1 - 1 = zero

So these would appear to be a combo to use when trying to hold a lead, even though it is not the best combo, still a very positive one.

Gerald was +4 -6 = -2 higher risk, higher reward than Scheyer
McClure @ zero - 2 = -2 18 games player vs 22
Zoubek @ zero - 1 = -1 only 13 games played, so less data
King was +2 - 3 = -1 Offensive specialist who can be high reward when hot

Another interesting note I forgot to mention in last post is 9 guys over 10 minutes per game and Dave McClure rounds to 10 at 9.7 MPG.

Five Duke guys average double figures with King now @ 7.5 since he has been cold lately but can score in bunches when he gets hot again, and he will.

Nolan @ 6.5. Lance has been more productive lately but too many games played to drastically change his 4.7 PPG.

Zoubek can probably raise his 3.7 PPG with more minutes.

http://goduke.statsgeek.com/basketba...season=2007-08

GP GS MPG FG% 3FG% FT% RPG APG BPG SPG PPG Player
22 22 31.3 0.513 0.403 0.632 5.9 3.2 0.4 1.6 15.5 DeMarcus Nelson (2,-0))
22 22 28.2 0.486 0.364 0.772 6.1 1.3 0.9 1.0 13.8 Kyle Singler (1, -0))
22 00 27.8 0.463 0.411 0.855 4.2 2.2 0.3 1.1 10.8 Jon Scheyer (1,-0))
22 22 25.8 0.482 0.278 0.616 4.9 1.6 1.0 0.9 13.2 Gerald Henderson (1,-0)
22 21 26.6 0.449 0.438 0.824 2.0 3.5 0.1 1.7 10.5 Greg Paulus (3,-1)
22 01 16.1 0.481 0.342 0.784 1.7 1.7 0.2 0.6 06.5 Nolan Smith (0,-1)
20 16 18.6 0.523 0.000 0.556 3.2 0.3 0.5 0.8 04.7 Lance Thomas (0,-2)
22 00 11.9 0.444 0.411 0.591 2.4 0.5 0.5 0.5 07.5 Taylor King (0,-1)
13 02 10.9 0.579 0.000 0.412 3.4 0.5 0.8 0.4 03.9 Brian Zoubek (1,-1)
18 04 09.7 0.471 0.000 0.286 1.8 0.4 0.2 0.3 01.0 David McClure (0,-5)

[Jumbo]

Now updated through the second Maryland game...

[-jk]

Looking a little closer (and if my math is right) we've averaged +18.5 as a team for the season. That puts our top 6 rotation within half a point of our team average in per 40 minutes

Kyle Singler +19.0 (28.6 mpg)
Greg Paulus +18.6 (27.0 mpg)
Jon Scheyer +18.5 (27.7 mpg)
Lance Thomas +18.3 (18.8 mpg)
DeMarcus Nelson +18.3 (31.4 mpg)
Gerald Henderson +18.2 (25.9 mpg)

and

Nolan Smith +17.4 (15.7 mpg)

not far off, either. It's not terribly surprising that everyone is closing in on the team average; it's where the math takes us. Equally, a single game is too small a set for any meaningful analysis.

Jumbo, is your game by game data still in a form where it would be easy to do the math on, say, just the ACC games or just the last five, to see any recent trends?

-jk

[Jumbo]

Jumbo, is your game by game data still in a form where it would be easy to do the math on, say, just the ACC games or just the last five, to see any recent trends?

-jk

I've got all the info, just not in one place. Should be easy, though, and will post it in the next day or two.

[ACCBBallFan]

Not as scientific as what Jumbo does, just rank order players by metric and sum, updated.through Miami game.

Tot Mpg fg% 3% ft% rpg apg bpg spg ppg Player
76.0 10.0 08.0 09.0 05.0 10.0 09.0 05.0 10.0 10.0 DeMarcus Nelson
67.5 09.0 07.0 06.0 07.0 09.0 05.0 09.0 06.5 09.0 Kyle Singler
63.0 08.0 03.0 08.0 10.0 07.0 08.0 04.0 08.0 07.0 Jon Scheyer
60.5 06.0 05.0 04.0 06.0 08.0 07.0 10.0 06.5 08.0 Gerald Henderson
56.5 07.0 02.0 10.0 09.0 02.5 10.0 01.0 09.0 06.0 Greg Paulus
40.5 04.0 06.0 05.0 08.0 01.0 06.0 02.5 04.0 04.0 Nolan Smith
40.5 05.0 09.0 02.0 03.0 06.0 01.0 06.5 05.0 03.0 Lance Thomas
37.0 03.0 01.0 07.0 04.0 04.0 04.0 06.5 02.5 05.0 Taylor King
36.0 02.0 10.0 02.0 02.0 05.0 02.5 08.0 02.5 02.0 Brian Zoubek
17.5 01.0 04.0 02.0 01.0 02.5 02.5 02.5 01.0 01.0 David McClure
495.0 55.0 55.0 55.0 55.0 55.0 55.0 55.0 55.0 55.0

[ACCBBallFan]

										
Tot	Mpg	fg%	3%	ft%	rpg	apg	bpg	spg	ppg	Player
76.0	10.0	8.0	9.0	5.0	10.0	9.0	5.0	10.0	10.0	DeMarcus Nelson 
67.5	9.0	7.0	6.0	7.0	9.0	5.0	9.0	6.5	9.0	Kyle Singler 
63.0	8.0	3.0	8.0	10.0	7.0	8.0	4.0	8.0	7.0	Jon Scheyer 
60.5	6.0	5.0	4.0	6.0	8.0	7.0	10.0	6.5	8.0	Gerald Henderson 
56.5	7.0	2.0	10.0	9.0	2.5	10.0	1.0	9.0	6.0	Greg Paulus 
40.5	4.0	6.0	5.0	8.0	1.0	6.0	2.5	4.0	4.0	Nolan Smith 
40.5	5.0	9.0	2.0	3.0	6.0	1.0	6.5	5.0	3.0	Lance Thomas 
37.0	3.0	1.0	7.0	4.0	4.0	4.0	6.5	2.5	5.0	Taylor King 
36.0	2.0	10.0	2.0	2.0	5.0	2.5	8.0	2.5	2.0	Brian Zoubek 
17.5	1.0	4.0	2.0	1.0	2.5	2.5	2.5	1.0	1.0	David McClure 
495.0	55.0	55.0	55.0	55.0	55.0	55.0	55.0	55.0	55.0

[Jumbo]

I'm going to update the cumulative numbers in the next couple of days. I'd prefer to keep this thread to plus/minus.

[Jumbo]

Finally added the Wake Forest and Miami games. All stats are current.

[Jumbo]

Now updated through the St. John's game...

[Jumbo]

Now updated through the Georgia Tech game...

[Jumbo]

Now updated through the second NC State game...

[Neals384]

I was curious if there are players who complement each other particularly well in the Duke lineup. Using Jumbo’s data thru the second NC State game, I totaled the points for/against for each five-man lineup that includes a particular two-man combo. For example, Zoubek & Paulus appear together in 17 different five-player lineups, totaling 164 for, 122 against, +42. Because some two-man combos occur much less often than others, I used percent of total points scored, rather than a simple +/-.

The most effective two-man combos:

Pts Pts
For Ag +/- Pct
King – Smith 421, 299, +122, .585
King – Scheyer 573, 409, +164, .584
King – Singler 222, 161, +61, .580
King – Zoubek 142, 106, +36, .573
Paulus – Zoubek 164, 122, +42, .573
Singler – Zoubek 155, 116, +39, .572
Henderson – Zoubek 178, 134, +44, .571

And the least effective:

McClure – Zoubek 52, 63, -11, .452
McClure – King 87, 88, -1, .497
McClure – Smith 166, 166, 0, .500
McClure – Henderson 187, 184, +3, .504
Smith – Thomas 227, 220, +7, .508
McClure – Paulus 237, 228, +9, .510
Nelson – Zoubek 211, 198, +13, .516
McClure – Scheyer 269, 247, +22, .521

As expected, McClure is included in many of the worst two-man combos. But McClure – Nelson (291, 230, +61, .559) is above the team average of .553.

All of the best and worst two-man combos include at least one of the four role players. All combos of the starters plus Scheyer are very close to the team average. The best of these is Henderson – Singler (1170, 901, +269, .565).

Conclusions: If one assumes that the data are predictive (big if), then:

1) When Zoubek comes into the game, Nelson should sit.
2) When McClure comes into the game, Nelson should play.
3) When Smith comes into the game, Thomas should sit.

[mus074]

I may disagree with your conclusions as overly broad, but it does illuminate some trends and tendencies. Well done and thanks!

BTW, where did Paulus-Singler come in as compared to all other intra-combos of the top 6 rotation guys? Just curious, because they seem to compliment each other's game well when either is hitting 3s well.

[Jumbo]

I was curious if there are players who complement each other particularly well in the Duke lineup. Using Jumbo’s data thru the second NC State game, I totaled the points for/against for each five-man lineup that includes a particular two-man combo. For example, Zoubek & Paulus appear together in 17 different five-player lineups, totaling 164 for, 122 against, +42. Because some two-man combos occur much less often than others, I used percent of total points scored, rather than a simple +/-.

The most effective two-man combos:

Pts Pts
For Ag +/- Pct
King – Smith 421, 299, +122, .585
King – Scheyer 573, 409, +164, .584
King – Singler 222, 161, +61, .580
King – Zoubek 142, 106, +36, .573
Paulus – Zoubek 164, 122, +42, .573
Singler – Zoubek 155, 116, +39, .572
Henderson – Zoubek 178, 134, +44, .571

And the least effective:

McClure – Zoubek 52, 63, -11, .452
McClure – King 87, 88, -1, .497
McClure – Smith 166, 166, 0, .500
McClure – Henderson 187, 184, +3, .504
Smith – Thomas 227, 220, +7, .508
McClure – Paulus 237, 228, +9, .510
Nelson – Zoubek 211, 198, +13, .516
McClure – Scheyer 269, 247, +22, .521

As expected, McClure is included in many of the worst two-man combos. But McClure – Nelson (291, 230, +61, .559) is above the team average of .553.

All of the best and worst two-man combos include at least one of the four role players. All combos of the starters plus Scheyer are very close to the team average. The best of these is Henderson – Singler (1170, 901, +269, .565).

Conclusions: If one assumes that the data are predictive (big if), then:

1) When Zoubek comes into the game, Nelson should sit.
2) When McClure comes into the game, Nelson should play.
3) When Smith comes into the game, Thomas should sit.

Awesome! Very, very cool stats. I'll need some time to mull them over.

[Neals384]

BTW, where did Paulus-Singler come in as compared to all other intra-combos of the top 6 rotation guys? Just curious, because they seem to compliment each other's game well when either is hitting 3s well.

Paulus - Singler combo is 1277 points for, 997 points against, +280, .562
So, yes, the stats show that they do complement each other nicely.

[mus074]

Paulus - Singler combo is 1277 points for, 997 points against, +280, .562
So, yes, the stats show that they do complement each other nicely.

They also say nice things about each other's game.