free NBA basketball picks
Home | Teams/Players | Commentary | Premium Services | Contact Us

Interested in one of the top master’s in Economics programs in the country or
a new, innovative microeconomics-focused Ph.D. program in Economics?

Defense is All about Keeping the other Team from Scoring

NBA Statistical Analyst Dan Rosenbaum

by Dan T. Rosenbaum, August 2005



Defense is all about keeping the other team from scoring. A player can be a good defender by getting a steal, a block, or a defensive rebound, but those stats offer only a snapshot of how effective a player is on defense. A player can also be an effective defender by denying the ball to an efficient scorer or by letting an inefficient scorer shoot more. Or by providing help defense in a way that does not expose other good scoring opportunities.

We would like to have more and better data to measure some of those “unobservable?aspects of defense. And one direction is to do just that by collecting better defensive statistics, an effort that is being spearheaded by Roland Beech at nbatop.com

Another approach is to use plus/minus statistics to measure how a team defends when a player is in the game. If, for example, the Nets consistently defend better when Jason Collins is on the floor, regardless of who is on the floor with him, then we would have to conclude that Collins was a good defender. And it doesn’t matter if he ever got a steal, block, or defensive rebound. Or what his reputation as a defender is.

Defensive statistics, votes for the All-Defensive team, and accolades in the press are nothing more than indicators that a player may be a good defender. Plus/minus statistics are not some subjective opinion or some vague statistic that may or may not be correlated with defensive impact. Plus/minus statistics are a record of which players actually have an impact on the defensive end.

We can debate about what plus/minus statistics tell us about future defensive effectiveness (especially if a player changes roles), but there can be no debate about what they tell us about past defensive effectiveness. Like a free throw percentage statistic that records who was the most effective free throw shooter, plus/minus statistics can be used to record who was the most effective defender. The calculation is more complicated, especially when we account for the other players on the floor, but the concept is the same.

But there is a difference between free throw percentage statistics and plus/minus statistics ?in addition to the need to account for the other players on the floor.? In most cases we only need a few dozen games to get a meaure of free throw percentage that would be a good predictor of future free throw percentage. With plus/minus statistics, the data demands are much greater. We need a full season or more to get a rating that is a reasonable predictor.

To account for the other players in the game, I compute adjusted plus/minus ratings, which I break into their offensive and defensive components. Playing with Ben and Rasheed Wallace can make defenders look more effective; for example, Tayshaun Prince looks great before accounting for the Wallaces and not-so-great afterwards.

To deal with the greater data demands, I compute statistical plus/minus ratings that measure the average adjusted plus/minus ratings of players similar to a given player.? This statistical plus/minus rating helps get around the severe data demands of plus/minus statistics by using the argument that players with similar statistics are likely to have similar defensive ratings.

In “Measuring How NBA Players Help their Teams Win?/u> I describe the gory details of how these adjusted plus/minus ratings can be computed. (I have made a few changes since then, along with adding another year of data, but it is a useful piece for those who want more detail.) Here are some highlights of the methodology that I am currently using.

(For more comments about this methodology and these results from some of the top basketball statistics experts, as well as lots of other interesting discussions about basketball statistics, see the APBRmetrics message board

  • I use data from 2002-03 through 2004-05 with 2004-05 being weighted more heavily.
  • I weight “clutch?time more heavily and “garbage?time less heavily (or in some cases not at all).
  • I assume that players improve and then decline in effectiveness as the age and gain more experience. Well technically, I don’t assume this; I use the data to estimate this relationship. The ratings below are projected for players at the end of the 2005-06 season. The ages given are for the end of the 2005-06 season.
  • The statistical plus/minus ratings only use data from 2004-05.
  • The overall ratings are 60 percent based upon adjusted plus/minus ratings and 40 percent based upon statistical plus/minus ratings. (This differs from my blog where I gave the statistical plus/minus ratings more weight.)
  • These lists are limited to players playing 1,000 or more minutes in 2004-05.
  • I got the positions from Doug Steele’s site and I know that a lot of them are wrong. I have fixed a few dozen, but get a life if that is your biggest complaint about these ratings.
Before getting to the lists, let me talk about how to look at these data.
  • Nobody, including this author, believes that these lists are the be all and end all.
  • Age is as of the end of the 2005-06 season. Minutes are for the 2004-05 season.
  • The overall rating tells us, for example, that if we replace an average defender with Ben Wallace, defense should improve by about 5.8 points per 40 minutes.
  • Bigger standard errors indicate that the rating is more prone to error. (I have incorporated differences between the statistical and adjusted plus/minus rating into these standard errors. For that reason they are probably a little bigger than they should be.)
  • The percentiles tell us what percentage of players at the same position have worse adjusted (or statistical) plus/minus ratings.
  • Remember that in the last three columns, adjusted plus/minus ratings will vary more when players have not played a lot of minutes. In cases where players have played only a few minutes, we can get some extreme results.
  • Don’t look only at the ranking or only at the rating. Look at the standard error. See how big it is. See if the adjusted and statistical plus/minus ratings are about the same. See if the adjusted plus/minus ratings vary a lot from year to year (beware of rookie ratings because we can’t see how they vary from year to year). Think about all of these things before coming to a conclusion about how much to trust any one given number.
  • For those of you who think these ratings have no basis in reality, ask yourself this question ?courtesy of Kevin Broom. If your favorite player isn’t rated highly enough, why are his defensive contributions not showing up on the scoreboard? And why isn’t the scoreboard what really matters?

The Best and Worst Defensive Adjusted Plus/Minus Ratings:
CENTERS

Overall
Percentiles
Adjusted +/-
Player
Age
04-05
Minutes
Rating
Std
Error
Adj
+/-
Stat
+/-
02-
03
03-
04
04-
05
1
 Ben  Wallace
31
2,671
5.8
1.4
97
95
7.0
7.1
5.9
2
 Dikembe  Mutombo
39
1,212
5.3
1.8
95
90
1.7
1.3
10.0
3
 Theo  Ratliff
33
1,731
5.2
1.3
88
91
7.4
5.0
5.8
4
 Jason  Collins
27
2,542
4.8
1.4
91
77
6.1
5.8
5.3
5
 Jeff  Foster
29
1,594
4.4
1.7
90
55
3.1
5.4
6.9
6
 Brendan  Haywood
26
1,865
3.7
1.5
82
57
2.3
4.7
5.3
7
 Erick  Dampier
31
1,609
3.6
1.5
85
40
5.5
4.6
4.8
8
 Joel  Przybilla
26
1,858
3.6
1.5
59
93
4.8
1.3
2.7
9
 Kelvin  Cato
31
1,525
3.6
1.3
71
86
3.0
4.1
3.0
10
 Rasho  Nesterovic
29
1,785
3.5
1.1
73
82
3.3
3.2
4.3
 ..
34
 Robert  Traylor
29
1,326
0.6
1.5
13
42
1.0
-1.9
0.2
35
 Chris  Kaman
24
1,632
0.5
1.5
23
22
0.0
-0.8
0.4
36
 Mikki  Moore
30
1,178
0.4
1.5
32
13
-12.0
2.4
0.9
37
 Nazr  Mohammed
28
1,933
0.4
1.7
11
43
-2.4
0.7
-0.7
38
 Eddy  Curry
23
1,815
0.2
1.4
28
10
0.4
-0.9
0.5
39
 Mark  Blount
30
2,130
0.0
1.2
26
7
-0.8
0.6
0.1
40
 Predrag  Drobnjak
30
1,435
-0.3
1.4
25
3
0.3
-2.9
1.2
41
 Raef  LaFrentz
29
2,195
-0.4
1.6
7
35
1.3
0.3
-2.8
42
 Marc  Jackson
31
1,976
-0.7
1.4
21
0
0.8
-1.1
-0.1
43
 Primoz  Brezec
26
2,276
-1.5
1.6
3
5
2.7
-3.6
-2.1

Here are some points to consider.

  • Defensive ratings, on average, are highest for centers, then power forwards, then small forwards, then shooting guards, and then point guards.?Offensive ratings go the other way.?This squares with the conventional wisdom that defense is anchored by big guys.
  • Ben Wallace is rated highly on both the adjusted and statistical plus/minus ratings.?His contributions are picked up in these numbers.
  • Dikembe Mutumbo had a phenomenal year last season on the defensive end, but his ratings bounce around a good deal from season to season, so we should be skeptical of exactly how good of a defender he will be next season.
  • Jason Collins is a consistently great defender; I don’t remember ever hearing him discussed as an elite defender in this league, but only a select few players have been more effective defensively than he has been over the past three seasons.
  • Jeff Foster is a very effective defender despite not having great defensive statistics.? Joel Pryzbilla is an effective defender who may be a bit overrated by his defensive statistics.
  • There is a big dropoff after the top five defenders.?The difference between #5 (Foster) and #6 (Brendan Haywood) is the same as the difference between #6 (Haywood) and #15 (Shaquille O’Neal).?Thus, at the very least rankings #6-#15 should be considered to be about the same.
  • The big drop in defensive effectiveness between Nazr Mohammed and Rasho Nesterovic is interesting.

The Best and Worst Defensive Adjusted Plus/Minus Ratings:
POWER FORWARDS

Overall
Percentiles
Adjusted +/-
Player
Age
04-05
Minutes
Rating
Std
Error
Adj
+/-
Stat
+/-
02-
03
03-
04
04-
05
1
 Tim  Duncan
30
2,203
5.4
1.7
91
96
5.2
5.4
7.1
2
 Kevin  Garnett
29
3,120
5.2
1.8
95
94
7.5
6.7
4.3
3
 Nick  Collison
25
1,389
4.8
1.8
98
79
0.0
0.0
5.4
4
 Nene  ?/td>
23
1,317
4.7
1.9
99
68
5.6
7.2
4.7
5
 Rasheed  Wallace
31
2,687
4.5
1.6
93
73
6.2
5.1
6.0
6
 Tyson  Chandler
23
2,182
3.2
1.9
58
98
1.2
1.5
1.7
7
 Kenyon  Martin
28
2,272
3.2
1.5
88
64
3.1
4.4
3.6
8
 Robert  Horry
35
1,396
3.0
1.5
87
72
4.9
4.8
3.5
9
 Reggie  Evans
25
1,887
2.9
1.3
83
88
2.9
2.8
2.4
10
 Dirk  Nowitzki
27
3,020
2.7
1.3
78
80
4.1
1.4
2.2
 ..
48
 Donyell  Marshall
32
1,645
-0.3
1.6
13
41
1.3
-0.7
-2.3
49
 Shareef  Abdur-Rahim
29
1,867
-0.4
1.3
22
5
-0.3
-0.4
-0.5
50
 Troy  Murphy
26
2,375
-0.5
1.6
14
27
-0.4
-1.0
-1.7
51
 Al  Jefferson
21
1,051
-0.7
2.6
2
85
0.0
0.0
-4.4
52
 Antawn  Jamison
29
2,605
-0.8
1.2
18
0
-0.3
-0.9
-0.4
53
 Juwan  Howard
33
1,624
-0.8
1.2
17
3
-1.6
-0.9
0.4
54
 Austin  Croshere
31
1,827
-1.0
1.5
11
11
-0.8
1.5
-1.8
55
 Antoine  Walker
29
2,955
-1.0
1.4
7
23
0.7
-2.8
-2.3
56
 Clifford  Robinson
39
1,688
-1.4
1.6
6
15
0.3
-0.4
-4.2
57
 Matt  Bonner
26
1,552
-2.4
1.8
0
26
0.0
0.0
-5.0

Here are some points to consider.

  • Again there is a big gap between the top five and everyone else.
  • Remember rookies like Nick Collison, Al Jefferson, and Matt Bonner are more prone to extremes given the scarcity of data on them.? I would expect them to gravitate more toward the middle in the future.?That said, Collison subjectively seems like a very good defender to me, so this is not a big surprise.
  • Players similar to Tyson Chandler have great adjusted plus/minus ratings.?Chandler’s ratings are so-so.?That to me is a red flag about exactly how effective his defense is.
  • Interestingly, the list of least effective defenders has a number of high-priced players.? Championships may be won with defense, but it is offense that pays the bills.

The Best and Worst Defensive Adjusted Plus/Minus Ratings:
SMALL FORWARDS

Overall
Percentiles
Adjusted +/-
Player
Age
04-05
Minutes
Rating
Std
Error
Adj
+/-
Stat
+/-
02-
03
03-
04
04-
05
1
 Shane  Battier
27
2,516
3.5
1.6
96
85
4.4
1.1
5.6
2
 Andrei  Kirilenko
25
1,349
3.3
2.0
83
99
1.4
3.2
2.0
3
 Darius  Miles
24
1,699
2.8
1.6
94
88
4.3
2.0
3.7
4
 Trevor  Ariza
20
1,382
2.5
1.7
92
90
0.0
0.0
2.1
5
 Paul  Pierce
28
2,959
2.2
1.5
88
74
4.2
2.8
2.2
6
 James  Jones
25
1,330
2.0
1.7
91
57
0.0
-3.6
3.3
7
 Shandon  Anderson
32
1,171
1.9
1.4
77
95
3.3
-1.0
2.5
8
 Bruce  Bowen
34
2,627
1.8
1.5
81
82
3.7
3.5
2.2
9
 Josh  Smith
20
2,050
1.8
2.1
61
97
0.0
0.0
0.0
10
 Vince  Carter
29
2,828
1.7
1.6
85
58
3.7
2.5
2.0
 ..
36
 Tim  Thomas
29
1,940
-0.7
1.2
31
19
0.4
-0.8
-1.0
37
 Jarvis  Hayes
24
1,560
-0.8
1.4
25
33
0.0
-1.1
-1.5
38
 Caron  Butler
26
2,746
-0.9
1.3
18
42
-2.0
-0.6
-2.1
39
 Rashard  Lewis
26
2,697
-1.1
1.4
13
39
-1.5
-2.4
-1.4
40
 Wally  Szczerbiak
29
2,558
-1.4
1.2
21
4
0.0
4.2
-2.6
41
 Josh  Childress
22
2,376
-1.5
2.0
8
68
0.0
0.0
-3.9
42
 Lee  Nailon
31
2,017
-1.5
1.3
16
2
-2.4
-6.9
-0.7
43
 Andres  Nocioni
26
1,895
-1.5
1.8
2
93
0.0
0.0
-4.5
44
 Peja  Stojakovic
28
2,534
-2.0
1.3
10
9
-1.0
-3.9
-2.4
45
 Matt  Harpring
29
2,584
-2.2
1.7
4
25
-2.4
-2.5
-3.5

Here are some points to consider.

  • For those of you who are going to complain about Bruce Bowen, note that he has the highest adjusted plus/minus rating in 2003-04 among those in the top 10.?He is an elite defender and these numbers show that.
  • Paul Pierce and Vince Carter are often criticized for not giving full effort on the defensive end, but their teams have played better defense when they are in the game.?Notice that both are brought up by the adjusted plus/minus ratings.?On the defensive end, they do not appear to be stat whores.
  • Andres Nocioni is a curious case ?great defensive stats but horrible defensive adjusted plus/minus ratings.?I’d like to see him for another year before coming to any conclusions.
  • Wally Szczerbiak has been very inconsistent defensively.?Matt Harpring has been very consistent ?in a bad way.

The Best and Worst Defensive Adjusted Plus/Minus Ratings:
SHOOTING GUARDS

Overall
Percentiles
Adjusted +/-
Player
Age
04-05
Minutes
Rating
Std
Error
Adj
+/-
Stat
+/-
02-
03
03-
04
04-
05
1
 Tony  Allen
24
1,263
4.7
2.1
99
93
0.0
0.0
5.5
2
 Ben  Gordon
23
2,003
3.4
2.1
97
52
0.0
0.0
5.1
3
 Andre  Iguodala
22
2,686
2.4
1.7
87
98
0.0
0.0
0.3
4
 Gerald  Wallace
23
2,147
1.8
1.4
91
96
0.2
-2.8
2.0
5
 Aaron  McKie
33
1,118
1.5
1.7
72
97
1.4
1.2
-0.4
6
 DerMarr  Johnson
26
1,232
1.4
2.2
96
51
0.0
-6.2
3.0
7
 Manu  Ginobili
28
2,193
1.3
1.4
85
87
1.1
0.1
2.1
8
 Eddie  Jones
34
2,839
1.2
1.2
83
84
1.0
0.9
1.9
9
 Trenton  Hassell
27
2,068
1.2
1.3
93
67
2.8
2.5
0.3
10
 Quinton  Ross
25
1,659
1.1
1.5
70
94
0.0
0.0
0.0
 ..
50
 Devin  Brown
33
1,238
-1.8
2.0
8
65
1.1
-2.2
-2.5
51
 Jerry  Stackhouse
31
1,617
-1.8
1.4
18
19
-1.6
-4.9
-1.8
52
 Ricky  Davis
26
2,696
-1.8
1.2
21
13
-1.6
-3.7
-2.3
53
 Rodney  Buford
28
1,314
-1.9
1.8
9
49
0.0
0.7
-3.0
54
 Raja  Bell
29
1,790
-2.2
1.4
7
37
0.7
-2.0
-3.8
55
 Jalen  Rose
33
2,710
-2.2
1.2
15
1
-0.6
-1.4
-2.5
56
 J.R.  Smith
20
1,859
-2.2
1.9
4
47
0.0
0.0
-4.3
57
 DeShawn  Stevenson
25
1,089
-2.4
1.3
11
0
-3.6
-1.9
-3.0
58
 Keith  Bogans
25
1,791
-2.4
1.8
6
29
0.0
-3.7
-3.2
59
 Michael  Redd
26
2,848
-4.2
1.7
0
10
-5.6
-6.8
-6.0

Here are some points to consider.

  • Tony Allen, Ben Gordon, and DerMarr Johnson all have very large standard errors.? Take their ratings with an even bigger grain of salt.?In fact, of the top six players, only Gerald Wallace has a reasonably small standard error.
  • Gordon and Tyson Chandler played a lot of minutes together.?Looking at 82games lineup data for Gordon and Chandler, it appears that the Bulls played very good defense in the few times when Gordon played without Chandler. When Chandler played without Gordon, the Bulls played pretty poor defense.? Statistically, this implies that it is Gordon who is the better defender and thus he is getting the lion’s share of the credit for the good defense played when both were in the game.?This does not square with anyone’s perception of who should get credit, but it will be interesting to see if this pattern continues into next season.  
  • In my blog, I erroneously said the following: “Interestingly, [Raja] Bell who is being signed by the Suns in order to shore up their defense rates as a bad defender. What is remarkable about Bell's results is that he has played for several teams over the past three seasons and yet his defensive ratings have been consistently bad.?/u> This is wrong. Well, that might be a little harsh. It is not quite right. ?/span>Bell has been an ineffective defender the past two seasons for Utah, but was effective for Dallas in 2002-03.?Thus, perhaps in a new environment in Phoenix he will be effective once again.
  • Michael Redd is in a league of his own on defense at the shooting guard position.?The difference between Redd and the next lowest-rated shooting guard is three times as large as the difference between the 2nd and 10th lowest-rated shooting guards.?And Redd has been very consistently ineffective over the years.?Again, defense may win championships, but it does not pay the bills.

The Best and Worst Defensive Adjusted Plus/Minus Ratings:
POINT GUARDS

Overall
Percentiles
Adjusted +/-
Player
Age
04-05
Minutes
Rating
Std
Error
Adj
+/-
Stat
+/-
02-
03
03-
04
04-
05
1
 Chris  Duhon
23
2,177
3.7
1.6
98
98
0.0
0.0
3.8
2
 Marcus  Banks
24
1,145
2.3
1.7
95
91
0.0
-2.1
4.8
3
 Earl  Watson
26
1,808
2.0
2.0
96
62
4.2
2.7
3.3
4
 Jason  Kidd
33
2,436
1.2
1.6
88
95
0.6
2.5
1.3
5
 Eric  Snow
33
1,844
1.1
1.2
86
93
2.1
0.9
1.7
6
 Beno  Udrih
23
1,149
1.0
1.9
91
76
0.0
0.0
1.1
7
 Steve  Francis
28
2,978
0.9
1.7
92
68
1.9
1.9
1.3
8
 Jason  Hart
28
1,887
0.8
2.0
85
82
0.0
5.6
1.0
9
 Baron  Davis
27
1,581
0.7
1.6
90
70
4.5
1.3
-0.2
10
 Speedy  Claxton
28
1,866
0.3
1.3
78
80
-0.2
1.0
0.1
 ..
51
 Chucky  Atkins
31
2,903
-2.8
1.3
17
8
-0.1
-1.4
-4.0
52
 Nick  Van Exel
34
1,620
-2.8
1.3
19
3
-3.1
-4.7
-2.2
53
 Tony  Delk
32
1,340
-2.8
1.5
9
16
-0.8
-6.0
-4.1
54
 Howard  Eisley
33
1,428
-2.9
1.4
8
15
-1.8
-4.6
-3.7
55
 Leandro  Barbosa
23
1,088
-3.0
1.6
7
22
0.0
-3.4
-4.3
56
 Tierre  Brown
26
1,066
-3.2
1.8
10
1
-7.2
-28.0
-3.4
57
 Carlos  Arroyo
26
1,448
-3.2
1.7
3
23
4.3
-2.6
-5.3
58
 Damon  Stoudamire
32
2,762
-3.2
1.4
5
17
-0.7
-4.2
-4.5
59
 Tyronn  Lue
29
2,007
-3.6
1.4
2
6
-3.1
-4.3
-4.9
60
 Troy  Hudson
30
1,729
-6.0
2.0
0
0
-7.4
-8.9
-9.2

Here are some points to consider.

  • Chris Duhon defensively is the highest-rated point guard both with the statistical and adjusted plus/minus ratings.?It would be good to see another year of data on him, but he appears to be an elite defender (and one who has gotten very little interest in the free agent market).
  • Earl Watson also has consistently been a very effective defender, despite not having great defensive stats.?He also is a testament to the defense wins championships, but does not pay the bills theme of this piece.
  • Note that because of injury or indifference Baron Davis?defensive effectiveness has fallen rapidly over the past three seasons.
  • Steve Francis has consistently been an effective defender.?He is a lightning rod for criticism but on two different teams he has consistently helped his team on the defensive end.
  • Troy Hudson probably gets the award for the being the worst defender in the league.? He is dead last among point guards in both the statistical and adjusted plus/minus ratings and his adjusted plus/minus ratings are consistently horrible.?He is playing a game on the defensive end that is not remotely like anyone else’s in the league.  ?


Dan T. Rosenbaum is an economics professor at the University of North Carolina at Greensboro. Besides this statistical work, Rosenbaum has been cited in numerous publications for his expertise on issues related to the NBA collective bargaining agreement and especially the luxury tax. He is thankful to the many remarkable individuals who have helped him tremendously in better understanding the NBA.?See his blog for more of his writings on statistical and financial issues related to the NBA.

 

Also see Dan's previous articles at 82games:
- Measuring How NBA Players Help Their Teams Win
- Picking the Difference-Makers for the All-NBA Teams


Rate this Feature
Poor   Fair   Good   Excellent

Enter your comments in the box

Email (optional)

We want your feedback! Tell us your thoughts

Copyright ?2005 by www.nbatop.com, All Rights Reserved