Defensive Breakthrough Article

[antiks]

Got this from Nbalive.org (an nba live 2005 site)

take it for what it's worth.....

Defensive Breakthrough?


benji - December 24th, 2004

For years it's been easy to measure how well a player produces on offense because nearly all the stats the NBA records are offensive statistics. In the old naive days for me, I considered blocks+steals+rebounds to be a good judge of defensive prowness for players, and points per game a good measure for teams. As we know now both these are almost idiotic measures of defense. Danny Forston grabs a bundle of rebounds, and there are plenty of players who get lots of steals and blocks but can't guard their man.

For teams, plenty of people have realized using points per possessions works just dandy. I personally use John Hollinger's method of determining points per 100 possessions. There's some quibble over offensive rebounds by the statheads on the interweb, and the value of free throws, but in most cases the different methods come out with the same rankings. I use Hollinger's because it's what I started using a few years ago, and I despise change. To get this we do:

(PTS*100)/(FGA+(.44*FTA)-OR+TO)

Recently, I've been taking that and putting it under the league average (which was 100.9 at my last measure) to get a "Defensive Rating" in which the league average = 100. Here's the top five and bottom five teams so far this year (I've included their defensive rating as well as the regular points per 100 possessions):
RANK TEAM DEF RTG DEF
1. SAS 109.7 92.0
2. DET 104.8 96.2
3. HOU 103.5 97.5
4. DAL 102.6 98.4
5. CLE 102.5 98.4
26. CHA 97.2 103.8
27. LAL 97.0 104.0
28. MIL 96.6 104.5
29. ATL 95.1 106.1
30. UTA 94.3 107.0

That's all well and good for teams, as it gets rid of the pace differences. But what about players? Hollinger created a defensive PER in his last book, but I don't have time to go through every boxscore or pbp. Well, I do, but I'm lazy. Knickerblogger was using the opponents PER that 82games.com has on their player pages, and I started using that as well. Then I realized that it's measuring all sorts of things (defensive rebounds, blocks, steals, fouls) that are irrelevant if you're trying to measure defense. I'm sure the difference is really marginal, but I'd like to narrow it down.

Points is really the only thing that matters in basketball. Everything leads to points. You have the points you make and the points you allow. Hollinger has found the value, in terms of points, for nearly every stat (blocks, turnovers, fouls, etc.) so I took those values and using the 82games data created a bundle of defensive ratings. (An O infront of a stat in further equations, indicates it's the opponents stats.)

First I created my "DEF" rating (I give things three or four letter designations so they don't take up space in Excel, deal with it.) which I purported to try and measure the points the player allows per 10 shot attempts. I could've gone all the way and used possessions but doing possessions for players weights too heavily on point guards in my opinion and this looked well balanced.

((OPTS-(.72*BLK)-OTO+(.35*PFLS))/(OFGA+(.44*OFTA)))*10

With this data I also did a DPSA. I use Hollinger's points per shot attempt (along with eFG%) on players offensively, so I thought why not do it for defense?

(OPTS)/(OFGA+(.44*OFTA))

Finally, I also have the points allowed per 48 minutes, which I don't like as much as it's dependent on how much the opposing player gets the ball, but I included since 82games provided the data per 48 minutes.

(OPTS-(.72*BLK)-OTO+(.35*PFLS))

So now that I've plugged these all in, as well as gotten the team at each position, let's look at some tables, shall we. First I added the five positions together and divided by 5 to get a team man defensive average.
RANK TEAM DEF
1. SAS 8.28
2. PHO 8.61
3. DAL 8.81
4. POR 8.85
5. DET 8.88
26. NOR 9.77
27. SEA 9.90
28. MIL 9.97
29. UTA 10.10
30. ATL 10.20

Now we can see who has defended the Center position the best (and worst) so far this season. (League average for the Position: 9.34)
RANK TEAM DEF
1. SAS 7.56
2. DEN 8.07
3. DET 8.26
4. MIA 8.27
5. POR 8.47
26. DAL 10.23
27. MIL 10.36
28. NOR 10.70
29. ATL 11.61
30. SEA 12.13

Power Forward. (LA: 9.30)
RANK TEAM DEF
1. SAS 8.10
2. LAC 8.30
3. DAL 8.42
4. DET 8.45
5. WAS 8.52
26. NYK 9.97
27. LAL 10.08
28. PHI 10.16
29. IND 10.30
30. MIL 10.59

Small Forward. (LA: 9.57)
RANK TEAM DEF
1. PHO 8.41
2. ORL 8.63
3. HOU 8.74
4. IND 8.80
5. DAL 8.82
26. NYK 10.37
27. LAL 10.59
28. LAC 10.74
29. CHA 10.74
30. GSW 11.12

Shooting Guard. (LA: 9.47)
RANK TEAM DEF
1. LAL 8.21
2. SAS 8.30
3. POR 8.36
4. DET 8.66
5. MEM 8.69
26. CHA 10.26
27. TOR 10.28
28. NYK 10.46
29. DEN 10.84
30. UTA 11.13

Point Guard. (LA: 9.08)
RANK TEAM DEF
1. DAL 7.84
2. BOS 7.85
3. DEN 7.97
4. PHO 8.08
5. IND 8.08
26. ORL 10.00
27. CHI 10.05
28. NJN 10.05
29. MIL 10.40
30. UTA 10.83

Then of course, we have to take a look at all the players in the league, first with the 12.5% minute cut-off. (LA: 9.38)
RANK Player TEAM DEF
1. D. Mutombo HOU 5.51
2. A. Kirilenko UTA 5.72
3. M. Banks BOS 6.55
4. M. Palacio TOR 6.57
5. R. Horry SAS 6.65
318. M. Pietrus GSW 12.56
319. J.R. Smith NOR 13.15
320. M. Fizer MIL 13.16
321. D. Fortson SEA 13.60
322. K. Snyder UTA 13.82

You may notice Pietrus is so low, well, last year he was in the stopper range, as was Tayshaun Prince, both have been horrible on defense this year. In his latest book, for his Defensive PER, Hollinger rose his cut-off to a third of minutes (33.3%) so we'll do the same. (LA: 9.29)

RANK Player TEAM DEF
1. A. Kirilenko UTA 5.72
2. M. Palacio TOR 6.57
3. S. Dalembert PHI 6.68
4. E. Gill IND 6.77
5. R. Nesterovic SAS 7.04
208. L. Nailon NOR 11.68
209. W. Green PHI 11.70
210. S. Smith CHA 11.78
211. P. Drobnjak ATL 12.21
212. D. Fortson SEA 13.60

I'll do one last cut, this time guys who have played at least 50% of minutes, since they would be having the largest defensive impact on their teams. (LA: 9.21)
RANK Player TEAM DEF
1. R. Nesterovic SAS 7.04
2. S. O'Neal MIA 7.16
3. A. Iverson PHI 7.38
4. T. Duncan SAS 7.44
5. T. Ratliff POR 7.46
127. M. Harpring UTA 10.87
128. C. Butler LAL 10.89
129. J. Crawford NYK 10.95
130. B. Simmons LAC 11.06
131. R. Bell UTA 11.24

A few more notes before I go. Hollinger has noted, and if you look at the numbers it's obvious, that there is a "halo" effect on defense. It's likely because we don't have exactly who guarded who at all times, so switches could have a small effect. For example, Duncan and Nesterovic don't always defend the man they're across from position wise, they often take the nearest big man, so Duncan has an effect on Rasho and vice versa. On the perimeter the same thing happens between the two wing spots and sometimes onto the point as well. Teams also tend to have an effect great defensive teams seem to lower all the positions, though it could be the other way around actually, because they're great at all positions their defense is great. But by looking at the players and knowing their reputations there appears to be an effect (if small) by the other four players if they're great on that one lousy defender in the lineup.

One more thing to note, is that this is for the first 25 games or so this season, I plan to go back and insert the 2003-04 numbers in the near future. Due to this small number of games, some players just aren't in their vibe, such as Pietrus and Tayshaun Prince. Prince last year posted a 7.18 DEF and a 0.86 DPSA, and was the best perimeter defender in the league. This year he's got 9.96 DEF and 1.13 DPSA. To illustrate the difference in the DPSA, last year Ray Allen shot well enough to get a 1.13 PSA (offensively), while Ira Newble shot poor enough to get a 0.86 PSA (offensively). Meaning Prince has gone from holding people to Ira Newble's level on offense, to allowing people to reach Ray Allen's level.

By no means are we anywhere near done measuring defense, but for me, this is a step in the right direction. You can definately expect to see more of my defensive stats in future columns.

[bloodyhell]

Rasho has been a very good defender. But I find it odd we're onlyl 26th at the center position. This says a lot about dampier. At times he's good but others, ugh.

[Pongo]

i thougt raja's D-rating would be excellent...
nice to be third on PF ("There is no D in Irk...) and first on Pg