http://82games.com/nichols1.htm

What do you guys make of this?

DEFENSIVE COMPOSITE SCORE
By Jon Nichols

Although an individual player’s defensive talents remain one of the hardest things to quantify in basketball, there are already some pretty good statistics in this area. Here at 82games.com there is defensive plus-minus. Other people such as Dean Oliver have developed their own defensive statistics that do a pretty good job of explaining things. However, as the creators and record keepers of any of these stats will tell you, you can’t look at any of these individually and expect to know everything.

What I’ve decided to do is combine some of these stats (with a little input and manipulation of my own) and try to increase their universal appropriateness. Like the others, mine is not perfect and must be taken in context with other information to gain any insightful knowledge. .

I have developed a stat I call Defensive Composite Score, or DCS (if you’re a college football fan who hates the BCS, try to ignore the name similarities). It is three different statistics combined into one.

* The first system is made up of the raw box score stats you find in the newspaper every day.
* The second is the defensive on/off plus-minus statistics you can find here at 82games.com.
* The third and final system is Dean Oliver's Defensive rating. It is an estimate of how many points per 100 possessions a player allows. To calculate it requires a complicated formula, one that can be explained in detail in his book Basketball on Paper.

Each of the three parts is weighted equally.

Why did I choose to base DCS on these? Box score stats do a good job of measuring athleticism and activity on defense, and I like to think of them as a way of measuring not only what a player has already accomplished but how good they can potentially be in the future. Plus-minus does a great job of measuring a player’s defensive impact on the court, especially within a player’s role on the team. Defensive ratings reward good defenders on good teams.

The statistic I developed is based on the power of relativity (I’m no Einstein, and this theory is nothing like his). I rank every player in the league based on how they did in these different methods, and combine their rankings to come up with a final ranking. I then convert these rankings into a score based on a scale of 0 to 100 to make things easier to understand.

Player 		Team 	Pos 	Box	Def. 	+/-	Drtg 	Drtg	DCS Rating
				Rank	+/-	Rank		Rank
balkman,renaldo nyk 	SF 	10 	 -9.2 	2 	102 	36 	99
lee,david 	nyk 	PF 	171 	-2.6 	74 	105 	93 	74
collins,mardy 	nyk 	SG 	173 	-2.5 	76 	109 	214 	55
francis,steve 	nyk 	PG 	177 	-0.8 	124 	109 	214 	46
jeffries,jared 	nyk 	SF 	80 	+3.0 	268 	108 	170 	45
frye,channing 	nyk 	PF 	247 	-1.1 	114 	108 	170 	41
robinson,nate 	nyk 	PG 	206 	-2.4 	79 	110 	254 	40
rose,malik 	nyk 	PF 	233 	+1.3 	202 	107 	140 	34
richardson,quent nyk 	SG 	308 	+1.8 	225 	108 	170 	16
crawford,jamal 	nyk 	SG 	235 	+3.3 	277 	112 	300 	5
curry,eddy 	nyk 	C 	304 	+5.3 	306 	110 	254 	2
marbury,stephon nyk 	PG 	269 	+5.4 	308 	112 	300 	1

Too much is not explained here. How is "Box Rank" calculated? How is D-Rating calculated? Howe are each of these weighted? Are they equally weighted? Should they be? The lower the drating the better I'm assuming?

How much additional am I getting out of this than pure Defensive +/-. With the exception of Jeffries, Frye and Nate, they seem to go in order of the +/-. Jared's getting upgraded b/c of his box score rank, and Frye and Nate are getting downgraded. Again- how much does one characteristic matter more than the other, or are they equal?

looks like I didn't post the whole article.... following the URL for full explanation.

Overall, the results look pretty right on. I hope this can put to rest the idea that Eddy Curry and Stephon Marbury are 'average' or are doing 'enough' on defense and we can also give David Lee some credit for being decent on defense (certainly better than he's given credit for).

Add in Zach Randolf and we might not win a game all season if your start those four players

A legitimate critique of both defensive +/- and Drating is that they use team defensive performance to assign defensive credit to individual players. That would be fine if team defense were a linear combination of the quality of defense of each individual player. But it most probably is not. Put a bunch of above average individual defenders on a team and the team defense as a whole is likely better than the sum of the parts. Likewise, put a bunch of below average individual defenders on a team and the team defense as a whole is likely worse than the sum of the parts. That could explain in part why the Knicks have so many players who rank so extremely low. It also probably explains why Balkman comes out looking as good as he does. On a better overall team defense, the extremely low players (Marbury, Curry) wouldn't be as low as they are and the extremely high player (Balkman) wouldn't be so high. In that sense the ratings are probably distorted.

Interesting piece. But I find it hard to believe that Lee's defense is underrated. The forward position was by far one of the worst for the Knicks in points allowed per box scores. How does this author penalize a player or reward a player? For example since Curry is a center is he penalize for not providing sufficient help defense in the post? Is Lee rewarded for grabbing rewards but not penalize for allowing his opponent to outscore him. For example Frye who played both the PF/C positions was considered a better defender than Curry but looking at their individual stat lines during the season. But Frye only averaged better than Curry in two lines blocks .6 to Curry's .5 and steals .5 to Curry's .4. What are the real factors in this system because looking at the box scores the forward position was the weakest for the Knicks as far as opponents points against. Then I factor in how can either Frye or Lee be rated better than defenders than Richardson who averaged 7 rpg, .9 spg, .1 bpg. So was Richardson heavily penalized for his bpg.

Jon Nichols, I just spoke to Starbury and he wants your phone number.

Pretty bad attempt to tell a story by mixing already ridiculous numbers. Trenton Hassel is a 5 out of a 100 as a defender? Since when? Qrich a 16 out of 100? Raja Bell a 24 out of a 100? Brent Barry is twice the defender Finley is? Come on, lol. Deron Williams a 3 as a defender? Not according to his own coach.

Camby the defensive player of the year wouldnt even be listed top 30.
Camby is 4th on his own team!
LOL Illguskas is rated higher than Camby defensively.

Greg Buckner Dallas main defensive guy is rated last on their team
Malik Allen a better defender than Henrich?

Posted by Bonn1997:

Posted by Solace:

Overall, the results look pretty right on. I hope this can put to rest the idea that Eddy Curry and Stephon Marbury are 'average' or are doing 'enough' on defense and we can also give David Lee some credit for being decent on defense (certainly better than he's given credit for).

It's hard to imagine anyone thinking Lee and Curry are on the same level defensively. Lee moves his feet and hustles. Eddy just stands there, watching.

I was told by guys on this board (I won't name names) that Lee was just as bad as Curry and terrible in his own right. I read those posts and wondered if there was some alternate version of the Knicks that shows up on some peoples tv.

[Edited by - Solace on Jul 23 2007 2:12 PM]

LOL. So now David Lee is a better defender than Carlos Boozer.
boozer,carlos uta PF 231 +2.7 257 104 79 37

Etan Thomas?
thomas,etan was PF 88 +1.6 218 107 140 58

At least Rasheed ranks slightly better than Lee
wallace,rasheed det PF 21 +3.2 274 100 21 77

Donyell got them all beat.
marshall,donyell cle PF 78 -1.2 109 99 15 90