A couple of weeks ago Andre’ Snellings – of Yahoo! Sports and RotoWire.com- asked me to evaluate the top candidate for the MVP award. Snellings was primarily interested in where Wins Produced ranked Kevin Garnett. His column (Hoops Lab: MVP Talk IV – The Heart) covered far more than the Wins Produced numbers (and is hence a good read), but for here let me just note the Wages of Wins aspect of the Snellings story:
… let’s clear up a big misconception. Garnett does in fact have the numbers to compete directly with the other MVP candidates when you factor in defense, where he should be the NBA Defensive Player of the Year. According to Dave Berri from wagesofwins.com, Garnett entered this week behind Paul in Wins Produced per 48 minutes but ahead of both James and Bryant.
The M2P Leaders
As Snellings notes, KG is ahead of both LeBron and Kobe. In my discussion of Garnett posted a couple of days ago I intentionally avoided comparing KG to other players (besides Kobe). But perhaps I should report the numbers Snellings got to look at in writing his column.
Here are the Wins Produced and WP48 numbers for several of the top candidates for both M2P (Most Productive Player) and MVP (as of Friday, March 28):
Dwight Howard: 23.1 Wins Produced, 0.392 WP48
Chris Paul: 21.7 Wins Produced, 0.401 WP48
Marcus Camby: 19.4 Wins Produced, 0.383 WP48
LeBron James: 19.0 Wins Produced, 0.339 WP48
Jason Kidd: 17.9 Wins Produced, 0.336 WP48
Tim Duncan: 16.8 Wins Produced, 0.338 WP48
Kevin Garnett: 15.8 Wins Produced, 0.356 WP48
Manu Ginobili: 14.6 Wins Produced, 0.328 WP48
Kobe Bryant: 14.1 Wins Produced, 0.238 WP48
Paul Pierce: 12.6 Wins Produced, 0.227 WP48
One should note that this is not a comprehensive list. In other words, Kobe Bryant is not currently 9th in Wins Produced (I really only looked at the players that Snellings asked about).
With that being said, here are a few thoughts I had when I looked at this list:
- The current leader in Wins Produced is Dwight Howard. But Paul has a higher WP48. Plus, because of the “short supply of tall people”, it’s harder for a guard to lead the league in Wins Produced. In other words, what Paul has done is more impressive than what Howard has accomplished.
- The WP48 for LeBron James would be 0.378 if he only played small forward. LeBron, though, is not only the Cavaliers best small forward, he’s also their best power forward. So we can see why the Cavs use him at both forward spots. Still, if LeBron were strictly a small forward, his numbers would come close – although not quite equal – to what we see from Howard and Paul.
- Garnett is the best player on the best team. And he is having an amazing season. But he is not quite as productive as Paul or Howard.
- When we look at Wins Produced it is hard to conclude that Kobe Bryant should be MVP. Although Bryant is having a very good year, he’s not nearly as productive as the players at the top of the list. Plus, he is barely more productive than Paul Pierce and he falls short of Manu Ginobili.
So who should be MVP?
Because Howard has played more minutes than Paul, I think Howard will finish the season as the leader in Wins Produced. In other words, I think Howard will be the Most Productive Player or the M2P. Again, though, what Paul is doing is more impressive. And Paul does have the highest WP48.
I think you can also make a credible argument for LeBron. He has been very productive despite playing substantial minutes at a different position. KG should also get some consideration. Because he has great teammates, it’s possible that Garnett’s numbers are somewhat lower (i.e. diminishing returns).
I would eliminate Camby and Duncan, since I think both falls short of Howard. In other words, if you are going to give it to a big man, Howard is the most productive this year (that is if you can ignore the diminishing returns story I told about KG).
Likewise, if you are going with a small man, Paul is the obvious choice. This means Kidd, Kobe, Ginobili, and Pierce are not better choices than CP3.
In sum, I think this should be a race between Howard, Paul, James, and KG. If I had to vote – and I don’t – I would go with Paul. It’s harder for a little guy to distance himself from others at his position. Hence, the 0.401 mark of Paul (and I checked today and he still has this mark), is very impressive.
- DJ
Our research on the NBA was summarized HERE.
The Technical Notes at wagesofwins.com provides substantially more information on the published research behind Wins Produced and Win Score
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Simple Models of Player Performance
What Wins Produced Says and What It Does Not Say
Introducing PAWSmin — and a Defense of Box Score Statistics
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.
10 responses so far ↓
William // April 6, 2008 at 9:50 am
I’m struggling a little with how the “short supply of tall people” phenomenon implies that Paul’s feats are more impressive than Howard’s.
Here’s my understanding of the phenomenon:
Since there is a smaller population of potential NBA centers, the variance in their abilities will be greater than for point guards (let’s limit the discussion to these two positions).
Okay, but how does that translate into the variance of the sub-groups of NBA centers versus NBA point guards?
It seems to me that, since we’re selecting the best 90 (approximately) point guards and centers for all NBA rosters, the effect their different sized source populations will have is to ensure that the average NBA point guard is better than the average NBA center, relative to their respective population averages. I just don’t see what implication the phenomenon has on the sub-groups’ variances.
If you had first pick in the draft, with a team already composed of only exactly average players, and you could be assured that there was a point guard and a center who would each perform at the same value over average (for their respective positions) over their career, which would you take? You seem to have argued that you should take a center, due to his advantage over opposing centers, but isn’t it (for a given fixed value over position average) the point guard who is more rare?
dberri // April 6, 2008 at 10:00 am
William,
When the underlying population is smaller, the standard deviation in performance tends to be larger. Or… if the underlying population is larger, the standard deviation tends to be smaller.
This means that Chris Paul’s WP48 is probably more standard deviations above the mean at his position (relative to Howard).
This issue was discussed in The Wages of Wins in the discussion of Michael Jordan (without looking, I think it is in chapter eight).
William // April 6, 2008 at 11:06 am
“This means that Chris Paul’s WP48 is probably more standard deviations above the mean at his position (relative to Howard).”
Just double checking, you are talking about within the population of NBA point guards and NBA centers, right? I do understand that that statement can be made regarding Chris Paul vs. all ~6′2″ people and Howard vs. all ~6′11″ people.
If so, that’s a pretty strong statement. The average NBA fan would argue that we haven’t seen many big men of even Howard’s stature, while there is some argument as to whether Paul is the best point guard currently playing. (Not an argument I’m making, although my team’s point guard, Rondo, has embarrassed Paul this year.)
In terms of standard deviations above the (NBA position) mean, is Paul better than LeBron?
In a perfect world, where players had pure positions (no swingmen), and all draftees are the same age, would you take with the first pick the player the highest number of standard deviations above their (position) mean? I guess we’re getting into a knapsack problem there….
In any case, it will be interesting to watch them both (and LeBron… and Rondo) continue to mature.
antonio // April 6, 2008 at 11:14 am
dberri, how did you decide what position adjustment you used for lebron? and how many players do you apply this to, as lebron is not the only player to play mutliple positions
Andrew // April 6, 2008 at 11:22 am
Chris Paul is likely the best PG we’ll see since Magic. So there is really no argument that he isn’t, at the very least, the best PG today. What Dwight is doing is great, but it’s not like we haven’t seen similar production as recently as Shaq, Duncan, or KG.
pablasso // April 6, 2008 at 11:51 am
oh, i see.
So Bryant’s 81 were even more impressive than Chamberlain’s 100 right?
Andrew // April 6, 2008 at 4:18 pm
I guessed it.
http://dberri.wordpress.com/2008/04/03/missing-the-obvious-on-kg/#comment-56347
Animal // April 6, 2008 at 8:17 pm
Andy, very prescient guess. Nice job!
Todd // April 7, 2008 at 4:12 pm
It seems to me Chris Paul is the best choice for MVP by most any measure. His stats appear to be as good or better than just about everyone else’s based on almost any model. (first in WP48, second in win score, second in PER) He is on the best team in the West. He is even exciting to watch.
When you take into account the problems many of us see with over-weighing rebounding stats in the Wages of Wins, he appears even better.
Just looking at the list of players on the above list, it is clear that the failure to account for diminishing returns on rebounding is escalating a few of these top candidates. Howard, Camby, and Kidd all get 5 or 6 more boards than the average player at their position. In none of these cases would their teams lose 5 or 6 boards a game if they were replaced on the court with an average player.
Meanwhile, LBJ’s win score is hurt by the fact that he does not grab as many boards as an average power forward even though he is on the best rebounding team in the league. It would seem to me that it is no more logical to punish LBJ for not grabbing all of the boards in Cleveland than it would be to punish Camby for not scoring all the points in Denver. Because he is an inefficient scorer, Camby’s win score would actually go down if he took enough shots from his teammates to reach the scoring average of a typical center, and it should.
The reality is, there are a certain number of shots and rebounds to go around each game, and different teams distribute shot attempts and rebounding attempts differently. Denver is actually a poor rebounding team with a rebounding rate of .492. Perhaps Camby would be reduced to an average rebounder if he was asked to join a team with a .524 rebounding rate like Cleveland.
The EXTENSIVE MVP Rundown // April 15, 2008 at 11:52 pm
[...] Dberri [...]