In The Wages of Wins we make the observation that it might be best to evaluate scorers in the NBA relative to other scorers. Recently I updated the analysis seen in our book in this forum via an examination of the top scorers in 2005-06 – which one can see HERE. This analysis demonstrated that most scorers are very productive in terms of Wins Produced, although there are a few top scorers whose actual contribution is below average.

This analysis has led me to think about the value of scorers just a bit more formally. Young Hoon Lee and I recently completed an article where we present a production function for the NBA. Output in this function is wins. The inputs are the players, which we organized according to the position the players play.

The discussion of scorers, though, led me to think of a different formulation of the NBA production function. Instead of the inputs being point guards, centers, etc…, I thought it would be interesting to separate the players into just two groups: Scorers and Role Players.

Previously I defined a scorer as a player who produces at least one point every two minutes played (or 0.5 points per minute). Such a definition would be a problem for some teams since not every team has a player who scores at this rate. Consequently, I am relaxing the definition of a scorer to a player who scores four points every ten minutes played, or sixteen points every forty minutes played (or 0.4 points per minute).

Given these two inputs, how many wins do teams get from their scorers? And how many wins do they get from their non-scorers – who I will call the Role Players?

I looked at each team this past season in a fashion far less sophisticated then what Young Hoon Lee and I did for our article. Basically all I did was add together the wins produced by a team’s scorers and then did the same for the team’s role players (so I did not estimate a production function). Follow this link to a table that reports what I found.

On average each team’s scorers produced 59% of a team’s wins, although there was quite a bit of variation around the mean. The Wins Produced by Portland’s scorers was actually negative. After Portland we have Denver, whose scorers only produced 18.9% of the Nuggets total wins. At the other extreme there was Utah, whose scorers produced 99% of the team’s wins.

The top teams tended to be more balanced and can generally by found in the middle of the table. For example, the scorers for the Pistons produced 46.5% of their wins while the Spurs scorers contributed 64.6% of San Antonio’s Wins Produced.

What does this exercise indicate? Scorers tend to produce the majority of wins in the NBA. Still, a team cannot depend on scorers alone. The role players also matter. The top five teams in the league received on average 22.1 wins from their role players. In other words, the top teams manage to get contributions from the players whose primary job is to put the orange ball in the hoop. But these teams also receive contributions from the players charged with doing all the other tasks necessary to consistently win basketball games.

– DJ

*Basketball Stories*

Tom Mandel

July 23, 2006

DJ – it would be great if you’d implement search at this site. E.g., I’d like to find the Win Score piece, and there’s no effective way. I also seem to remember something about overrated players and underrated players – a table (but maybe this was in the book? in my mind?), which is impossible to find as well.

I’d like to measure how effective Jared Jeffries has been for the Wizards. Ditto Gooden for the Cavs.

dberri

July 23, 2006

Tom,

I think I added the search feature. You may have guessed that we don’t really know what we are doing when it comes to this blog and the website. So any suggestions are welcomed.

The over-rated vs. under-rated players is both in the book and in one of the posts Malcolm Gladwell put up at his website. Gladwell’s post has this analysis updated for the 2005-06 season. Perhaps I should put this list here and offer a few comments.

As for Jeffries and Gooden. Jeffries produced 5 wins and had a 0.122 WP48 while playing small forward for the Wizards. Gooden produced 9.7 wins and had a WP48 of 0.214 for the Cavs. For the Cavs Gooden played power forward. Between the two, Gooden is the more productive player.

Travis

July 23, 2006

DJ,

When our USA Basketball team starts internatioal play on Auguest 19th it will be interesting to find out which players continue to be “Scorers” and which players become “Role Players.” Team USA’s roster can be found here: http://www.nba.com/usabasketball/. This team, unlike past teams, does have some shooters on it and looks to have more then just a collection of scorers.

Using your Wins Produced model how many wins would you forecast these players to produce. For example if the Pistons traded their roster for this one how many wins would you expect them to accumulate?

Keep up the good work,

-Travis W

dberri

July 23, 2006

Travis,

I know there was some effort to put a few role players on this team. But without looking at the numbers, my guess is that most of these players are scorers. This likely means that some players will have to take on other roles when they take the court.

And that might be the problem with USA Basketball. Players are chosen because of their scoring ability. But when they play many have to do something else.

I will try and look at the roster in terms of Wins Produced. It might take me a few days to get this posted.

Travis

July 23, 2006

DJ,

Your correct in that this year’s team consists primarily of scorers like LBJ, Kobie,Carmelo, and D-Wayde to name a few. Coach K though said that he wants this years team to play more “like a team” then teams of the recent past and specifically left A.I off the team (which is probably a good move). I’m curious to see which rotation he uses to start the game(s) with. I wonder if he just put all the superstars over there together or if he will have some of the younger and more inexperienced players playing together. No matter what Coach K decides to do he can’t do much worse then our last team did.

Thanks again,

-Travis W

Jason

July 23, 2006

There’s something more than a little artificial with any division between “scorer” and “role player” since the points that a role player score count just as much in the final score as anyone else’s. I’m curious though about the division of stats. It seems to me that it should be possible to do an analysis on how different categories of “wins produced” break down. Are balanced attacks (many players dividing the scoring load) different from attacks where a few people do all the scoring? Is this true of rebounders too? And does this in any way covary with the error for the Wins Produced vs. actual wins? At first glance, I don’t see anything in your table that indicates any such obvious patterns, but it seems it’s worth a closer look.

Jason

July 23, 2006

It’s interesting that your breakdown of 4 points per minute means that 57% of the points scored in the league last year were scored by players (n=160) who meet that criterium.

If you subtract out those who score at such a rate without playing significant minutes (I arbitrarily broke it down at averaging less than 12 mpg but still scored more than .4ppm and exclude them [n=20]and their points as “garbage time” scorer–I’m assuming that these players likely got into games late and likely didn’t factor much into the outcome of the games much) then 59% of the non-garbage time points were scored by “scorers.” That’s pretty much the average you got in terms of “win score” produced by scorers as well. Not sure if the coincidence means much, but it’s interesting that the simple measure of a scorer (points) and your wins produced by them break down so similarly. Would indicate to me that there’s not a problem *as a group* in over or undervaluing anyone’s contributions.

Tom Mandel

July 24, 2006

DJ – search seems to be working, thanks (and I’ll gladly offer suggestions when I think of them).

The Wizards love Jeffries, and he does have some great skills. He and Gooden are in similar contract situations with their teams, and of course the reason for my query was that I’d love to see the Wiz trade Jeffries for Gooden, whom the Cavs don’t seem to properly appreciate.

Do you have figures for Songaila, btw? And, did you run Gooden/Jeffries in response to my query, or do you have a great big table. If the former, many thanks. If the latter, please publish that whole table somewhere; how great it would be to have it.

Harold Almonte

July 24, 2006

I can see in that table, teams which top scorers are “near the basket” players and/or efficient shooters, get more wins from them. And, these wich top scorers are inefficient shooters, produce less wins than role players. I´m correct?

Like Jason, I would like to see similar tables based on another stat (Rebound Rate, Turnover Rate, Steal Rate, etc.).

I think an updated second edition of your book, will be awesome.

Harold Almonte

July 24, 2006

Maybe Detroit was a different model. You can´t say it´s an inefficient offensive, but Ben Wallace is the ultimate role player.

dberri

July 25, 2006

Tom,

Songaila in his career has produced 2.7 wins. That’s it. So he is not that productive.

I do have data on every NBA player back to 1990-91, and I can with a bit of effort analyze the productivity of a player back into the 1980s.

I guess at some point I should put all this information on-line for people to sort through. I haven’t had the time to do this, though.

Here is what I am supposed to be working on this summer. I have been asked to explain in separate papers the models for the NBA and the NFL. Each of those papers will be published, but first I have to get these written. That is a big part of my summer plans. I also have four other academic articles I promised various co-authors I would finish. These projects are my priority. Basically, when you see me posting material I am actually shirking my responsibilities.

Of course, shirking is fun. So one thing I would like to do is gather all the tables I linked to in various posts in one location. That would be a nice resource to add to the homepage. And at some point, posting the evaluation of all NBA players would be good to add also. So much to do, so little time.