3029 Individual Defense (359 views)

[eric]

As pioneered by 20s Navidad , I took every team's box scores and tallied them up by opponent by position. I then calculated points per true shot attempt, free throw attempts per possession, and turnovers per possession for each opponent-position combo, then I calculated averages and standard deviations for each stat-position combo, and finally I summed the distance from the average in terms of standard deviations for each opponent-position combo (z score). Positive numbers mean good, negative mean bad. Here are the results, first in combined tenths of z score, second as overall rank.

C PF SF SG PG team
18 12 31 22 -16 76ers
16 25 12 29 33 Bobcats
-12 -25 -11 -32 -19 Bucks
36 24 29 41 14 Bullets
-12 -23 -22 -31 2 Bulls
56 38 34 16 17 Cavaliers
-14 5 9 -9 -10 Celtics
-4 -21 14 7 12 Clippers
-15 -6 -28 -22 -22 Grizzlies
-21 -12 4 9 -16 Hawks
-43 -22 -38 1 8 Heat
-8 29 -6 -3 -8 Hornets
15 10 27 16 9 Jazz
-10 26 1 9 -10 Kings
14 10 1 10 6 Knicks
34 11 14 0 15 Lakers
22 45 -7 38 -19 Magic
-2 22 13 44 22 Nets
8 0 20 -3 27 Nuggets
-6 3 15 -48 5 Pacers
-16 -59 -61 -41 -47 Raptors
-47 -47 -21 -18 -28 Rockets
10 -21 3 7 -20 Spurs
22 23 10 20 27 Suns
-19 -1 -23 -36 -9 SuperSonics
6 5 -27 -17 10 Thunder
-4 -16 -2 -2 44 Timberwolves
-9 -12 -7 -8 -24 Trailblazers
-13 -15 7 1 1 Warriors

C PF SF SG PG team
6 9 2 5 21 76ers
7 5 10 4 2 Bobcats
21 27 22 26 23 Bucks
2 6 3 2 8 Bullets
20 26 24 25 15 Bulls
1 2 1 7 6 Cavaliers
23 14 12 21 20 Celtics
14 24 8 13 9 Clippers
24 18 27 24 26 Grizzlies
27 20 14 11 22 Hawks
28 25 28 15 12 Heat
17 3 19 18 17 Hornets
8 11 4 8 11 Jazz
19 4 17 10 19 Kings
9 12 16 9 13 Knicks
3 10 7 16 7 Lakers
4 1 20 3 24 Magic
13 8 9 1 5 Nets
11 16 5 19 4 Nuggets
16 15 6 29 14 Pacers
25 29 29 28 29 Raptors
29 28 23 23 28 Rockets
10 23 15 12 25 Spurs
5 7 11 6 3 Suns
26 17 25 27 18 SuperSonics
12 13 26 22 10 Thunder
15 22 18 17 1 Timberwolves
18 19 21 20 27 Trailblazers
22 21 13 14 16 Warriors
As a check of the method, I summed the combined z score for each team and graphed it against points per possession allowed, and got this:



So I'm pretty confident this is good stuff. Here's the All-Defense Teams, recognizing that starters who were injured or otherwise played less minutes would disproportionately gain (or lose) based on their backups:

All Defensive First Team
C_ Jahlil Okafor - Cavaliers
PF Brook Lopez - Magic
SF Thabo Sefolosha - Cavaliers
SG Stanley Johnson - Nets
PG Joe Silvestri - Timberwolves

All Defensive Second Team
C_ Hasheem Thabeet - Bullets
PF Tristan Thompson - Cavaliers
SF Shareef Abdur-Rahim - 76ers
SG Jesus Shuttlesworth - Bullets
PG Ty Lawson - Grizzlies
Thabeet's name is very interesting because he fared so poorly in 20x4's original analysis. He gets his mark here by being okay at defending the shot, okay at forcing turnovers, and excellent at not fouling.

Shareef gets a big asterisk because he missed so much time, but he played 50 games at SF to Chitwood's 32 so I gave it to him, plus KILL WHITEY

It's really interesting to me how some teams allegedly had good defenders all the way across (e.g. Bullets, Cavs) while some had massive albatrosses (e.g. 76ers PG, Magic PG and SF). I think this is a good sign, it shows the analysis isn't being clouded by team success. Sorry everyone on the Raptors and Rockets, you probably actually are that bad.

I'm definitely open to ideas on how to present this data or other calculations you might find interesting. It's something I'm going to keep doing in the future for sure.

[20s Navidad]

So Bismack was really good defensively, Brooks/Porzingis were good, and everyone else was meh.

[Lazy Pete]

Mar 10, 2016 16:28:27 GMT -6 eric said:

Sorry everyone on the Raptors and Rockets, you probably actually are that bad.

[eric]

Another factor I considered was true shot attempts, with the idea that a better defender could deny his man the ball and/or discourage him from shooting at all once he did have it. To test whether that existed in the software I ran two seasons: all 50s, then all 50s but starting centers have 100 shot blocking. When I did this the possessions used per minute by starting centers decreased by about 0.15 possessions per 36 minutes, which was a) not statistically significant and b) even it if was it's waaaay too tiny to actually measure on the level of individual teams.

I'm also leaning towards taking opponent turnover rate out of the equation. In the software there are four basic kinds of turnovers:
1. traveling
2. primary or trapping defender steals the ball
3. ball thrown away
4. the intended recipient's defender steps in the way of the pass

Given that #2 exists even for teams that "never" trap, I feel like a significant portion (perhaps the majority) of turnovers cannot be credited to the primary defender.

[Bankz]

Curious to how much Mystic helped my PF score... Had Pettit running at PF for the majority of the year.

Super shocked Butler was a negative defensive player considering he's one of the better defenders in the NBA