Charting Duke vs. Georgia Tech (I)

[Jumbo]

Not much to say after the loss at Georgia Tech. Mason Plumlee's strong performance coincides with his plus/minus rating, while for all the heat Kyle Singler has taken, Duke actually outscored Tech by a point with him on the floor. Meanwhile, a number of people have mentioned that Lance Thomas could have neutralized Gani Lawal had he not fouled out so early, but Duke didn't play particularly well when he was in the game.

Anyway, on to the numbers ...

Individuals
Mason Plumlee 51-46 (+5)
Kyle Singler 67-66 (+1)
Brian Zoubek 33-32 (+1)
Miles Plumlee 29-30 (-1)
Jon Scheyer 65-69 (-4)
Andre Dawkins 22-26 (-4)
Ryan Kelly 0-4 (-4)
Nolan Smith 52-57 (-5)
Lance Thomas 16-25 (-9)

Per 40 Minutes
Mason Plumlee +7.7
Brian Zoubek +2
Kyle Singler +1.1
Miles Plumlee -2.1
Jon Scheyer -4.1
Nolan Smith -6.1
Andre Dawkins -17.8
Lance Thomas -25.7
Ryan Kelly -80

Lineups

Scheyer-Smith-Singler-Mason-Zoubek (x6) 21-18 (+3)
Scheyer-Smith-Singler-Thomas-Miles (x4) 12-9 (+3)
Scheyer-Dawkins-Singler-Mason-Zoubek (x2) 8-5 (+3)
Scheyer-Dawkins-Singler-Mason-Miles 5-2 (+3)
Scheyer-Smith-Singler-Mason-Miles (x5) 10-8 (+2)
Smith-Dawkins-Singler-Mason-Zoubek 2-2 (+0)
Smith-Dawkins-Singler-Thomas-Miles 0-0 (+0)
Scheyer-Smith-Dawkins-Singler-Thomas 0-0 (+0)
Scheyer-Dawkins-Singler-Kelly-Mason 0-1 (-1)
Scheyer-Smith-Dawkins-Mason-Miles 0-2 (-2)
Scheyer-Smith-Dawkins-Singler-Mason (x3) 5-8 (-3)
Scheyer-Smith-Thomas-Kelly-Miles 0-3 (-3)
Scheyer-Dawkins-Singler-Thomas-Miles (x2) 2-6 (-4)
Scheyer-Smith-Singler-Thomas-Zoubek (x2) 2-7 (-5)

[ice-9]

Not much to say after the loss at Georgia Tech. Mason Plumlee's strong performance coincides with his plus/minus rating, while for all the heat Kyle Singler has taken, Duke actually outscored Tech by a point with him on the floor.

As did the Big Z!

[airowe]

I wonder how much of LT's -9 has to do with his 5 fouls? If the players he fouled were shooting free throws, and made a good percentage of them (GT only missed 5 all game so probably a good chance that happened) that certainly would have impacted his +/- numbers. I remember Lance being taken out immediately after fouling a couple times in the second half as well, which wouldn't have given him a chance to get his +/- back into the positive since he missed an offensive possession.

This is a situation where the numbers and the eye test tell two different stories, because it sure appeared to me that we were a whole lot better at stopping GT (and almost every team we've played) when Lance is on the floor.

[superdave]

Interesting that Mason and Zoubek seem to be such a good frontcourt together. I guess Z allows our 4 to do a little more because his presence keeps opponents out of the lane some.

As Pedro would say "I'd like to see more of that."

[CDu]

I wonder how much of LT's -9 has to do with his 5 fouls? If the players he fouled were shooting free throws, and made a good percentage of them (GT only missed 5 all game so probably a good chance that happened) that certainly would have impacted his +/- numbers. I remember Lance being taken out immediately after fouling a couple times in the second half as well, which wouldn't have given him a chance to get his +/- back into the positive since he missed an offensive possession.

This is a situation where the numbers and the eye test tell two different stories, because it sure appeared to me that we were a whole lot better at stopping GT (and almost every team we've played) when Lance is on the floor.

Hard to say whether appearances were deceiving or not. But in the 14 minutes Thomas played, we scored 16 and gave up 25. That equates to per-40 numbers of 45.7 points and 71.4 points, respectively. So it looks like the defense gave up points at roughly the same pace with and without Thomas, but scored a lot less.

Now, without having information on number of possessions both offensively and defensively, it's hard to tell if there were a lot more defensive possessions for Thomas than offensive possessions or not. Either he really got hosed out of several offensive possessions, or we just weren't very effective offensively when he was on the floor.

[Wander]

I see you fast forwarded through the part when Duke didn't have the ball.

Much like not turning the ball over is generally an under-appreciated part of playing good offense (why Scheyer and Wisconsin are underrated so often), not fouling and getting defensive rebounds are often overlooked as parts of playing good defense. Lance fouled really often and didn't get any defensive rebounds. He did a decent job on Favors, but I think we might want to see how Favors performs in ACC play before we put the Duke-GT game in the "Lance is a great defender" file.

[airowe]

Much like not turning the ball over is generally an under-appreciated part of playing good offense (why Scheyer and Wisconsin are underrated so often), not fouling and getting defensive rebounds are often overlooked as parts of playing good defense. Lance fouled really often and didn't get any defensive rebounds. He did a decent job on Favors, but I think we might want to see how Favors performs in ACC play before we put the Duke-GT game in the "Lance is a great defender" file.

It would be pretty foolish of me to use one game to prove a point. ;)

On another note, I wonder how much having to guard Lawal had an impact on Kyle's play in this game?

[CDu]

It would be pretty foolish of me to use one game to prove a point. ;)

On another note, I wonder how much having to guard Lawal had an impact on Kyle's play in this game?

How much did Singler defend Lawal? I didn't notice him doing this during the game. If he did, which of the wings did our second big (the one not guarding Favors/Peacock) defend when Singler was guarding Lawal? That would seem like an odd decision putting Singler on their center and putting our center on a GT wing.

[airowe]

How much did Singler defend Lawal? I didn't notice him doing this during the game. If he did, which of the wings did our second big (the one not guarding Favors/Peacock) defend when Singler was guarding Lawal? That would seem like an odd decision putting Singler on their center and putting our center on a GT wing.

I think I was wrong about that one.

[CDu]

I think I was wrong about that one.

Ah, sorry. Couldn't tell if there was sarcasm or if I had missed something significant/strange. Nevermind. Carry on.

[jv001]

The plus-minus seems to point out that Nolan's game was below average for him. Looking at the lineup with Andre in for him, the results show a plus. I guess Andre's defense is improving. Let's hope he gets his shot back. Go Duke!

[Saratoga2]

Scheyer played the majority of the game and got what the team got for the overall score. Since he played well, does this call into question the value of the +/- scores in general? I think it provides more information for a guy that plays for shorter periods so the game results while he is in can be compared to the time when he is not in, ala Mason Plumley.

[bird]

Scheyer played the majority of the game and got what the team got for the overall score. Since he played well, does this call into question the value of the +/- scores in general? I think it provides more information for a guy that plays for shorter periods so the game results while he is in can be compared to the time when he is not in, ala Mason Plumley.

Maybe you can reach into the lineups to get a sense for Scheyer.

Compare:

Smith-Dawkins-Singler-Mason-Zoubek 2-2 (+0)

with

Scheyer-Dawkins-Singler-Mason-Zoubek (x2) 8-5 (+3)


and

Smith-Dawkins-Singler-Thomas-Miles 0-0 (+0)

with


Scheyer-Smith-Singler-Thomas-Miles (x4) 12-9 (+3)


Scheyere was out twice. Both times the comparable lineup with Scheyer in the game was more productive than with Scheyer out of the lineup in favor of Smith/Dawkins. The sample size here is small, but I think you can say that Scheyer was one of Duke's two best guards in this particular game. What a startling conclusion.

[CDu]

Scheyer played the majority of the game and got what the team got for the overall score. Since he played well, does this call into question the value of the +/- scores in general? I think it provides more information for a guy that plays for shorter periods so the game results while he is in can be compared to the time when he is not in, ala Mason Plumley.

The +/- has many limitations which have been discussed ad nauseum. The biggest problems include:
1) basketball scoring is random enough that on a small sample size (such as a single game) it is possible for the numbers to vary widely from the player's actual performance value. In other words, single game +/- stats are frequently not a very good reflection of a player's value in a particular game.
2) raw +/- stats in a losing effort will tend to make guys who play more look worse. for example, as a rookie in the NBA, Kevin Durant had the worst +/- in all of basketball
3) and this relates to (2): you shouldn't compare +/- stats across teams, because it undervalues good players stuck on bad teams (and overvalues lesser players on good teams).

The key to +/- is understanding how to interpret +/-. On a large enough sample size, within-team comparisons can be meaningful. But reading too much into a single-game +/- can lead to incorrect conclusions. This is especially true for players who play shorter periods of time, because the sample size (i.e., playing time) is smaller and more at risk of the effects of random variation.

[-jk]

Or perhaps we could create a "corrected" "per 40" that takes into account the actual game margin. If Joe has +20 in a 15 point win, he has a "corrected" +5. Similarly, if Joe has +10 in that game, his corrected per 40 is -5.

What do you think, Jumbo? I think it might balance a specific weakness of plus/minus.

-jk

[Acymetric]

Or perhaps we could create a "corrected" "per 40" that takes into account the actual game margin. If Joe has +20 in a 15 point win, he has a "corrected" +5. Similarly, if Joe has +10 in that game, his corrected per 40 is -5.

What do you think, Jumbo? I think it might balance a specific weakness of plus/minus.

-jk

When someone mentioned using average scoring margin to compare the season plus/minus stats I believe they were getting at the same thing (in the stickied thread) and I think its a great idea.

[Jumbo]

The +/- has many limitations which have been discussed ad nauseum. The biggest problems include:
1) basketball scoring is random enough that on a small sample size (such as a single game) it is possible for the numbers to vary widely from the player's actual performance value. In other words, single game +/- stats are frequently not a very good reflection of a player's value in a particular game.
2) raw +/- stats in a losing effort will tend to make guys who play more look worse. for example, as a rookie in the NBA, Kevin Durant had the worst +/- in all of basketball
3) and this relates to (2): you shouldn't compare +/- stats across teams, because it undervalues good players stuck on bad teams (and overvalues lesser players on good teams).

The key to +/- is understanding how to interpret +/-. On a large enough sample size, within-team comparisons can be meaningful. But reading too much into a single-game +/- can lead to incorrect conclusions. This is especially true for players who play shorter periods of time, because the sample size (i.e., playing time) is smaller and more at risk of the effects of random variation.

Plus, what I'm doing is some pretty low-level stuff. Adjusted plus/minus, for instance, accounts for the quality of the opponent, too. I can't track who is in for the other team, or what that player's rating is. So some guys are penalized because they spend more time playing against the other team's starters. The season-long plus/minus figures are handy, but should also be taken with a grain of salt. The single-game plus/minus stats are fun, but should be taken with several grains.

[Jumbo]

Or perhaps we could create a "corrected" "per 40" that takes into account the actual game margin. If Joe has +20 in a 15 point win, he has a "corrected" +5. Similarly, if Joe has +10 in that game, his corrected per 40 is -5.

What do you think, Jumbo? I think it might balance a specific weakness of plus/minus.

-jk

I feel like it's almost implicit, no?

[Wander]

Or perhaps we could create a "corrected" "per 40" that takes into account the actual game margin. If Joe has +20 in a 15 point win, he has a "corrected" +5. Similarly, if Joe has +10 in that game, his corrected per 40 is -5.

What do you think, Jumbo? I think it might balance a specific weakness of plus/minus.

-jk

Isn't that what the net plus/minus is?