Putting things together

I’m not 100% happy with this one but when am I ever? Let’s get it out there.

And that’s kinda why I got off track. And it was an awesome detour. And we’ll get back to it. And I’m going to stop with the gimmick now.

But first,Combined WP48 take 5,6, eleventy? Let’s get on with it.

Just lie back and picture your favorite MVP candidate.

Because we also only stop for kittens (and the odd inspiration)

Ok, explanations first, fun and profit(?) second.If you’re new, welcome to my funhouse, feel free to peruse  the Basics for background on the metrics used  and remember that the numbers are made possible through the kind donations of Nerdnumbers .

What were doing here is looking at opponent adjusted Wins Produced. The goal is to get as close as possible to adding an individual opponent adjustment to the Wins Produced model. The driver for this has been the fact that currently Wins Produced divides defense up at the team level for all stats that are not in the boxscore and I have been trying for a while to get at this. Why? I don’t know call it a pet peeve.

The goal then  is WP48 at the player level adjust for what the player does and what his opponent does.In layman’s terms instead of being compared to the average team, we compare you to the player on the other side. Sounds good but it’s a little difficult to achieve with publicly available data sources.

But that has never stopped me before.

What did I actually do? I worked out:

  • Classic WP48 for each player.
  • Opponent WP48 for each player by game based on time by position (this is Wins Produced by team based on deviation of the opponent from average) .
  • Used opponent WP48 to generate losses Produced by opponent.
  • Added them up into a number I’m calling Combined WP48 (by the simple expedient of taking the average)

Here’s the result:

More than a few fun conclusions come up. Top twenty in opponent Losses produced per 48:


Rank Player Current Team
1 Rajon Rondo BOS
2 Dwight Howard ORL
3 Kobe Bryant LAL
4 James Jones MIA
5 Kevin Durant OKC
6 Gary Neal SAS
7 Jodie Meeks PHI
8 Richard Jefferson SAS
9 Andre Iguodala PHI
10 George Hill SAS
11 Ron Artest LAL
12 Ronnie Brewer CHI
13 Marquis Daniels
14 Shannon Brown LAL
15 Kyle Korver CHI
16 Kevin Martin HOU
17 Manu Ginobili SAS
18 LeBron James MIA
19 Andre Miller POR
20 Andrew Bynum LAL

Rondo is numbers #1 (which is very probably more than slightly helped by the Celts killer D). Howard is no surprise at number 2 (and Durant at 5 isn’t either , he keeps showing up when I do these things) . Kobe and James Jones are a surprise.


As for the MVP? I guess, I get to agree with Hollinger and call out the Big Man in Orlando.


Superman and the Weight of the World



Maybe I was just tired of having Love reign over me.


Oh what the hell, Before I go  I’ll give you one Usage Curve:

Dirk Nowitski how you doing?



Dwight Howard
Kevin Durant
Kobe Bryant
LeBron James
Rajon Rondo
Luol Deng
Derrick Rose
Paul Pierce
Andre Miller
Jason Kidd
Richard Jefferson
Andre Iguodala
Ray Allen
Kevin Martin
Manu Ginobili
Jason Terry
Chris Bosh
Marc Gasol
Ron Artest
Pau Gasol


  1. I think Hollinger has been right on the money in his MVP columns. I have to give him a lot of credit. His numbers don’t make sense but his logic is generally sound!

  2. Evan

    Some odd results on here. Some players known for defense have below average numbers (Ben Wallace for one). Other players who are bad defenders seem to out perform his teammates on defense (Bargnani).

    It also seem that players get a better defense rating when playing on a team with good defense overall, which seems to remove individual impact (Bulls players).

    • Evan,
      Still doing by position so it’s to be expected. The more minutes you play the more accurate it’ll be. Interesting effects as well with player on above par D’s .

      Next step is play by play (and I may have to pay :-( )

  3. Some Dude

    Think you have to switch Ron and Kobe. Ron often covers the SG. He defends the better wing.

    I still think you need to look at WP48 opp vs that opponent’s avg WP48.

    I agree on Dwight. He’s the MVP.

    To the Bargnani comment, he’s actually not a terrible man on man defender. He’s bad, but not terrible. He’s a horrific help defender, though. And bigs need good help defense. That’s why it seems like his teammates are outperforming him, but they’re not. Ben Wallace, on the other hand, has superb help D. But also he’s old now and Detroit sucks. Good defenders will appear to suck on poor defensive teams since they’re wasted.

    Unfortunately, it’s very hard to incorporate help defense, which is probably the majority of a big man’s defense in today’s NBA, into the stats. I’m not sure how Arturo could tackle that without using differentials of some kind.

  4. Some Dude

    Just a thought. Maybe what you do is this.

    WP48 counterpart opponent.

    WP48 opponent team average (when player on court)

    Then you do this. Say for Love his counterpart is .100. And say the team is .200. What you do is assign help d like this: Take the difference of the average opp team and subtract is by the counterpart and divide based on position.

    Say, C = .3, PF = .3, SF = .15, SG = .15 and PG = .1

    Then add it to the total.

    So for Love, it would be .100 for counterpart plus+.03 for help defense (.3 x .1), so .130 is his Defensive WP48 measure.

    I don’t know what values to put for each position. And maybe you only take the average WP of the other 4 players, not all 5, to determine opponent average WP48.

    I feel like I’m on to something here.

  5. bedsidetrash

    This is a cool idea. I wonder if loses produced would remain relatively constant (like WP) for a player that changed teams. And I’d guess that quantifying help defense would lead to +/- adjusted type problems.

  6. Crow


    I’d like to review this data further but find it somewhat hard with a picture this wide and don’t know any way to convert to excel directly. Any chance, as a general option, you could offer excel file links as well? Or for a wide picture like this, perhaps put the player name on the right and maybe in the middle too to aid reading across and knowing more easily which player’s line you are on?


    I think I follow your overall concept but I think you’d have to give the counterpart rating a % of the total and then help defense the other part to get th scores of all to sum properly. Also having a fixed set of shares for distributing help defense credit and blame (as a simplifying assumption) is an option that Evan and I discussed on his site. Maybe it helps get closer on average but individuals with vary from their position average responsibility based on team scheme and how an offense attacks it and that ideally would be something to try to specify from the data instead of assuming. Perhaps the Synergy data on player involvement in plays with switches and joint shot defense could give the data for a pretty good statistical estimate. Perhaps this is what Arturo hinted at paying for.

    • Some Dude

      yeah, i definitely agree. I was thinking through a possible process that might get a lightbulb to go off for arturo or someone else that would be better.

      You’re right that help D should play into your counterpart too. Didn’t think of that one.

  7. Crow

    Frequency of plays where a player didn’t help could be inferred and the impact of those failures to be involved could be estimated using certain simplifying assumptions.

    I don’t think Synergy offers an evaluation of the defensive effort, at least in the public version. Some teams do that themselves for each shot defense. Both efforts and outcomes are worth knowing. Is not always 1 to 1. Would need clear standards. Was the close-out a good one or just going thru the motions?

  8. 4/2/2011

    The issue w/ James Jones came up on the Miami Heat podcast last week. http://miami-heat-index.blogspot.com/2011/03/podcast-best-heat-options-at-point.html

    Long story, short – Jones has played most of his minutes paired w/ LeBron, Bosh or Wade. He’s also played the majority of his minutes at SF. Playing the same position as LeBron & Wade MAY make you look good defensively thru the boxscore lens but it also reduces your marginal productivity because no one else at the position can be more productive than them (see http://miami-heat-index.blogspot.com/2011/01/heat-check-benching-john-hollinger.html).

Leave a Reply

Your email address will not be published. Required fields are marked *