Chris Broussard – in the December 29 issue of ESPN the Magazine – noted that a number of expensive NBA contracts signed last summer have not paid off this fall (the article – which I can’t find on-line — is on page 71 of the print edition). Included in Broussard’s list were the contracts paid to Elton Brand, Luol Deng, Baron Davis, Gilbert Arenas, and Monta Ellis. Each of these players is either below average in 2008-09 or hasn’t actually played. So at this point, one could have an issue with the nearly $400 million guaranteed to these five.
If Broussard would have confined his story to this quintet, then his article would have been on fairly solid ground (a bit quick with the assessment, though). But Broussard couldn’t stop himself. To these five he also decided to add the names of Andre Iguodala, Andrew Bogut, and Emeka Okafor.
According to Broussard, Iguodala is having the worst season of his career. Certainly Iguodala’s scoring numbers are down. But his rebounds and assists are up. And when we look at WP48 [Wins Produced per 48 minute] we see only a small decline. Last year Iguodala’s WP48 was 0.172 while this year it is 0.170. In sum, Iguodala is not responsible for the problems in Philadelphia this season.
Bogut and Okafor Score Less and Produce More
A similar story is seen with respect to Bogut and Okafor. Before I get to that story, I should note that I commented on each player’s deal last summer (see The Bogut Economy and Okafor Finally Signs). At the time I noted that both players were above average, but Okafor has consistently been better (at least on a per-minute basis).
When we look at the numbers from this year we again see that each player is above average. And Okafor is again offering more. But contrary to what Broussard contends, each player has actually improved.
Table One: Andrew Bogut and Emeka Okafor in 2008-09
As Table One indicates, both players – like Iguodala – are scoring less. Consequently it’s not surprising to see Broussard – who has a history of focusing on scoring — conclude that each player is actually less productive. When we look past scoring, though, we see each player is offering more. Bogut has improved with respect to shooting efficiency and rebounds. Okafor’s rebounding numbers are down, but his shooting efficiency is much better.
The Better Bucks
When we look at Wins Produced we can see what these players mean to their respective teams. Let’s start with Bogut.
Bogut – as Table Two indicates – is Milwaukee’s most productive player. Despite his new contract, though, he’s not the highest paid player on this team. Even after his new contract kicks in next year, his pay will still lag behind the compensation of Richard Jefferson and Michael Redd. Not surprisingly, Jefferson and Redd lead the Bucks in points scored per game. But both players are below average with respect to WP48 (and this was true last year as well).
Table Two: Milwaukee Bucks after 31 games in 2008-09
Before moving on to Okafor, let’s quickly note that the Bucks have improved in 2008-09. Given what the players have done in the past the Bucks should have expected to be on pace to win 26 games this year (what they won last year). The performance of this team this year, though, is consistent with a team that will win about half their games.
About one-third of this improvement can be tied to the play of Bogut. Although Bogut has taken a step forward, the biggest leap can be seen in the play of Luke Ridnour. Ridnour’s WP48 was 0.125 as a starter in Seattle in both 2004-05 and 2005-06. When Ridnour stopped starting on a regular basis, though, his performance fell into below average territory. This season he has been returned to the starting line-up and his WP48 has risen to 0.147.
In addition to Ridnour and Bogut, the Bucks are also getting an above average performance from Ramon Sessions and a nearly average performance from Luc Mbah a Moute. Each of these players was a recent second round selection. Their play suggests that someone in Milwaukee knows something about drafting players. Then again, these same people selected Joe Alexander and Yi Jianlian in the first round. Both of these players have been below average NBA players, so maybe Milwaukee still has some issues in its scouting department.
The Charlotte Duo
Okay, enough about Milwaukee. Let’s look at Charlotte.
Table Three: Charlotte Bobcats after 31 games in 2008-09
Although Charlotte has only won 11 of its first 31 games, the team’s efficiency differential (offensive efficiency minus defensive efficiency) of -2.33 is consistent with a team that will win 35 games this year. In other words, Charlotte is not far behind Milwaukee.
Like Milwaukee, Charlotte is led by the man in the middle. Okafor’s WP48 of 0.293 is similar to what he did in 2006-07 (a 0.290 mark) and a bit better than Bogut this season (Bogut has a 0.260 mark).
Currently Okafor is on pace to produce 17.4 wins, a career high mark. Gerald Wallace is also on pace to produce a career high in wins [though not a career high in WP48]. If these two players maintain their current performance, this duo will combine to produce 31.6 wins this season. Unfortunately the rest of the roster is only on pace to produce 3.9 wins. After Okafor and Wallace, no other player on this roster is above average. And once you get past Raymond Felton, D.J. Augustin, Boris Diaw, Ryan Hollins, Raja Bell, and Matt Carroll, every other player currently has a WP48 in the negative range.
For those who know Bobcat history, the story of the 2008-09 season is hardly new. From 2004-05 to 2007-08, the Okafor-Wallace duo have combined to produce 69.4 wins (35.2 by Okafor, 34.2 by Wallace). The remainder of the roster only produced 46.5 victories across these four seasons (or less than 12 wins per year). Of the players who played at least 2,000 minutes in a season, only Brevin Knight in 2005-06 and Jason Richardson in 2007-08 were above average players. In sum, Charlotte has been Okafor, Wallace, and not much else since this franchise was created.
The generally poor play of the Bobcats likely impacts the assessment of both Okafor and Wallace. Statistics, though, have the power to separate a player from his teammates. When we make this separation we see that Okafor has been a very good player. And like Bogut, he has so far been well worth the money.
– DJ
The WoW Journal Comments Policy
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.
Wages of Wins Journal 
Michael
December 28, 2008
Great article prof, looking at boxscores for the last few bobcats games Okafor and Wallace have been dominating.I guess its a shame they have Michael Jordan picking their team mates!
If you have a moment, I have been attempting to learn the method of calculating wins produced through PAWS. I have calculated for Tyson Chandler and Lebron James. For Tyson I got a WP48 of 0.111, and 1.52 wins produced so far, which is consistant with the drop I have seen in his box scores, if a little larger decline than I expected.
For Lebron I got a WP48 of 0.581 and 12.7 wins produced so far! Now I suspect this may be inflated becuase he has been playing some power forward, but I don’t know how to calculate how much, so I just used the small forward average. What do you think? Has Lebron been this amazing so far this year? Or do I still have a lot to learn?
Thanks a lot for your help!
dberri
December 28, 2008
Michael,
LeBron has been spending a fair amount of time at power forward. So that will deflate his overall numbers somewhat. Still, he is having a great year.
Chandler, though, as your numbers show, had declined.
Michael
December 28, 2008
So I am doing it right, great stuff thanks!
KMils
December 28, 2008
Interestingly enough, the article you linked to regarding Chris Broussard overemphasizing scoring is one that predicts a playoff berth for the Indiana Pacers. With Dunleavy injured, early returns seem to corroborate the predictions of the “experts”. I don’t suppose that any of them will remember that when it comes time to evaluate their predictions.
nerd
Dunleavy would probably take minutes away from Daniels and Rush, who are averaging .151 and .076 Win Score/min. Dunleavy averaged .211 WS/min last year – it’s very easy to see how the team is worse. If Dunleavy plays just 700 minutes so far (on pace for 900 less than last year) and plays at the same level as last year, the Pacers improve by .05 WS/minute. That’s assuming that those 700 minutes come from only Rush and Daniels (400 and 300 a piece). Giving some of Stephen Graham’s minutes to Daniels and Dunleavy makes the team even better.
/nerd
None of my calculations are Wins Produced, but there is significant evidence that the Pacers should be improved from last year to this year, a thought that will presumably be lost on the commentators and announcers at ESPN.
Lior
December 28, 2008
A point I have raised before: given the correlation between total team WP and wins, I don’t think WP48 figures can be said to have three significant figures. In other words, it should be said for both AI (comparing 07/08 to 09/09) and Okafor (comparing 06/07 to 08/09) that the player’s marginal productivity (as measured by WP48) was essentially unchanged.
It is simply not believable that WP48 measures marginal productivity to such an accuracy that a change from 0.172 to 0.170 can be reasonably characterized as a “decline” in productivity. AI’s WP48 has, in a literal sense, declined, but using language this way creates the wrong impression of accuracy.
Vince Gagliano
December 29, 2008
Lior:
You make a good point.
However, if we’re going to follow significant figures to the letter of the law, then we should probably be measuring WP48 to two decimal places. After all, the values used to calculate production are carried out to two places, while other stats like PAWSmin are carried out to three or more.
Besides, what WP48 does is measure a player’s overall productivity and parcel it out into nice and neat 48-minute chunks.
So even if Igoudala is producing a -.002 WP48, if he plays more than he did last season, then he is more productive, not less.
Going back to Kevin Love and O.J. Mayo, Love has an above-average WP48, but plays fewer than 23 minutes a night. O.J. Mayo has a below average WP48, but he is giving Memphis 35 to 40 minutes a night. All else equal, they are producing wins at approximately the same rate, again, because WP48 measures productivity in set intervals.
Vince Gagliano
December 29, 2008
Addendum: I meant to say
“If Igoudala is producting at a -.002 WP48 rate compared to last year.”
Michael
December 30, 2008
Vince, with your point on Igoudala being more productive with a (very slightly) reduced WP48, I wonder what your take on the following would be.
If you had player A, with a WP48 of 0.100, play 3000 minutes in a season, and player B, with a WP48 of 0.200, play 1500 minutes, who would you consider the most productive?
I personally would consider player B to be more productive than player A, regardless of the total wins produced. What do you think?
Vince G.
December 30, 2008
What positions do player A and player B play?
Vince G.
December 30, 2008
Actually, before I get that, here is my short answer:
All else being equal, I would usually take player B.
Long answer: It’s a little bit more complicated than that.
There are many factors that would go into a decision to take a player either way, including, but not limited to:
Position: Not all WP48’s, even identical ones, are created equal. Centers and forwards will usually give you more production than guards with comparable WP48’s. In that case, you’d usually want to go with the player who plays the “bigger” position, as he will give you more points and rebounds than a smaller player who will produce the same amount of wins.
Comparable players on the team: I already talked about Minnesota and their issues in the backcourt. Usually, picking the player who fills a position of need will give you a bigger change in wins than a player who already satisfies a position of strength.
Injuries: If player B can give you 30+ minutes a night, but only played 1500 because he was hurt much of the season, then, barring a long list of injuries (Here’s lookin’ atchoo, Grant Hill), take B. If he plays more minutes next year at the same level of consistency, it’s good for wins production.
Age: If B was a player who produced in the .300-.400 range in years past, you might not want that player to be your most productive player in the next few years.
Cap room: If A or B has an expiring contract, picking them would open up cap room for a player who produces a higher WP48 and/or more minutes.
Lack of experience: If A was LeBron James as a rookie (Note: he produced a .099 WP48 that year), or another promising rookie, by all means, take A.
Those are just a few factors that can impact selection. The moral of this story: B is not the best player in every situation.
Matt
December 30, 2008
All things being equal (basically everything cited by Vince G), you have to pick player B every time.
Both players produce 6.25 wins for the team. However, in player B’s case there remains an additional, unaccounted-for 1500 minutes to allocate to other players. If these players have a WP48 over .000, then Player B will always be better for the team.
For instance if a player (Player C) with a WP48 of .050 plays the 1500 minutes that Player B doesn’t play to make up for the minutes that the team would have had with Player A playing all 3000, the team would gain an additional 1.56 wins. Therefore with Players B and C playing, the team would get about 7.8 wins out of 3000 minutes whereas the team would get just 6.25 wins out of Player A playing all 3000 minutes.
So player B is the most productive both in terms of his per minute productivity and his overall net benefit to the team measured in wins.
Of course, barring injury, coach could just play Player B for 3000 minutes!
Vince G.
December 30, 2008
Matt makes a point, however, there is a caveat:
First off, the players *have* to be productive (better yet, not unproductive)
Second, if the players are not productive enough, then in certain cases, picking A guarantees that those minutes will have a slow, steady burn of productivity.
For example, if player C has a WP48 of .000, then playing A (who would have an average playing time of 36:35 per game in an NBA season) gives C less than one NBA quarter a game, while playing B will mean C will have to play at least one half of the game.
Furthermore, if C is your most productive bench player, then it’s better to play A than leave 18 minutes per game up for grabs amongst the unproductive benchees.
Michael
December 30, 2008
lol I really just meant that WP48 is a more accurate indicator of productivity than total wins produced. For example if you look at these two players from last season;
Manu Ginobili: 14.6 Wins Produced, 0.328 WP48
Kobe Bryant: 14.1 Wins Produced, 0.238 WP48
Both produced the same number of wins, but I would consider Manu Ginobili to be the most productive, due to his much higher wp48.
(figures from https://dberri.wordpress.com/2008/04/06/more-on-kg-and-the-m2p/ )
Houston
December 30, 2008
There is one thing that Andrew Bogut does quite well and is given no credit for – taking charges. This highlights a small problem in your scoring system: you don’t track how many times a player creates a turnover by taking a charge. Taking a charge should count as a steal and be credited to the win score.
dustin
December 30, 2008
If it isn’t in the boxscore, the system isn’t meant to count it. DB wanted something simple and based on the boxscore.
The day they start counting charges as a steal is the day this will be accounted for in WoW 1.0 .
Who knows, maybe WoW 2.0 will use some more advanced/extensive statistical measures.
Michael
December 30, 2008
Okay well I have just finished projecting the WP48 of every player/team based on the PAWS method. If anyone is interested the top 20 players so far this year in terms of wins produced are as follows;
Player WP48 WINS PROD.
01. LeBron James 0.429 9.76
02. Chris Paul 0.458 9.67
03. Dwight Howard 0.402 8.86
04. Jason Kidd 0.366 8.17
05. Andris Biedrins 0.369 8.03
06. Marcus Camby 0.436 7.67
07. Rajon Rondo 0.319 6.73
08. Dwyane Wade 0.296 6.61
09. Pau Gasol 0.308 6.51
10. Kevin Garnett 0.307 6.50
11. Emeka Okafor 0.279 6.26
12. Chris Bosh 0.244 6.10
13. Tim Duncan 0.269 5.97
14. David Lee 0.294 5.96
15. Jose Calderon 0.269 5.90
16. Brandon Roy 0.245 5.88
17. Kobe Bryant 0.264 5.82
18. Joel Przybilla 0.381 5.74
19. Gerald Wallace 0.270 5.72
20. Amare Stoudemire 0.237 5.46
I think the biggest suprise here is the play of Rajon Rondo as he is currently more productive than Kevin Garnett for the Celtics (at least with my numbers.)
Also just off this list is Mike Bibby at 21 with a WP48 of 2.48 and 5.29 wins produced. A major reason for the Hawks improvement this year (aside from improved play by Joe Johnson and Marvin Williams) is his return to form.
Also what a waste for Marcus Camby playing with the Clippers this year.
I should also mention that these numbers will not take into account todays games.
dberri
December 30, 2008
houston and dustin,
I thought we discussed this issue last time Bogut came up. At the time (if my memory is correct) I think someone found data on charges and this was added into Bogut’s calculation. And if I remember correctly, it didn’t make a huge difference. Then again, maybe my memory is incorrect.
All that being said, if you had data on charges you could easily include this in Wins Produced. It would count as a steal. So you don’t need a new formula. Just the data on charges.
Michael
December 30, 2008
Sorry, Bibby is 0.248, not 2.48. If he was 2.48 he would have produced 50 wins already and be the best player ever :-P
Vince G.
December 31, 2008
Yeah, but then again, Bibby has never played in an All-Star game.
Harold Almonte
January 7, 2009
Michael and Vince,
The A, B and replacement C players-problem, is a typical “usage-eff. tradeoff” puzzle coachs need to solve every night, and this is a sample why any metric don’t have a solution for this, neither at the lineup level nor at the individual level.
too many steves
January 7, 2009
Someone should point out the glaring error in this post: it seems to imply that Luke Ridnour is good at basketball. That’s a guy who has never helped an NBA team, ever. He might be the worst defensive point guard I’ve ever seen, and offensively, he’s average on a good night.