According to 111 members of the sports media (out of 12o voters), Derrick Rose was the very best rookie in 2009. Of the nine voters who differed from this conclusion, five preferred O.J. Mayo, two preferred Brook Lopez, and two preferred Russell Westbrook.
Of these choices, none was actually the most productive rookie. But before I get to that story, let me discuss why the Rookie of the Year in 2009 was merely average.
An Average Rose
Let’s first compare what Rose did this past season to what we see from a typical point guard.
Adjusted Field Goal Percentage: 47.4% avg. PG, 48.2% Rose
Free Throw Percentage: 78.8% avg. PG, 78.8% Rose
Points Scored per 48 minutes: 18.4 avg. PG, 21.8 Rose
Rebounds per 48 minutes: 4.7 avg. PG, 5.1 Rose
Steals per 48 minutes: 2.0 avg. PG, 1.1 Rose
Turnovers per 48 minutes: 3.4 avg. PG, 3.2 Rose
Blocked Shots per 48 minutes: 0.3 avg. PG, 0.3 Rose
Assists per 48 minutes: 8.6 avg. PG, 8.2 Rose
Personal Fouls per 48 minutes: 3.6 avg. PG, 2.0 Rose
Win Score per 48 minutes: 6.3 avg. PG, 6.6 Rose
From these numbers we see Rose is below average with respect to steals and assists. He is above average with respect to shooting efficiency (barely), points scored, rebounds, turnovers, and personal fouls. But except for scoring and personal fouls, the difference between Rose and the average point guard is quite small. And his advantage with respect to scoring is only due to his propensity to call his own number (i.e. he is above average with respect to field goal attempts). Consequently – when we put the whole picture together – we see there is little difference between the Win Score of Rose and the average point guard.
In sum, all these numbers tell us that Rose was essentially little better than an average player in 2008-09. And this is because he really didn’t do anything exceptionally well. Yes, Rose has his fans. But I think even his fans would be hard pressed to find any facet of the game where Rose currently excels.
I should emphasize that I am not talking about the future. At some point, Rose might develop into an above average player. But as a rookie, this didn’t happen.
So why was he named Rookie of the Year?
The key issue is point score per game. Rose had the second highest scoring average among rookies. And since the leading scorer – O.J. Mayo – played on a losing team (and was also drafted after Rose), we should not be surprised that most of the media focused on the point guard from Chicago.
The Most Productive Rookies
Now what happens if we look at all rookies via Wins Produced? To answer this question we turn to Table One.
Table One: Ranking the Rookies of 2008-09
In the past I noted that Kevin Love would probably lead all rookies in Wins Produced (see HERE and HERE). And as we see in Table One, that’s where Love finished in the rankings. When we focus on WP48 [Wins Produced per 48 minutes], we see that Love was the only rookie to surpass the 0.200 mark (if we ignore what Hamed Haddadi did in 120 minutes).
Love, though, was not the only above average freshman (average WP48 is 0.100). Of the rookies who logged 1,000 minutes, Greg Oden, Rudy Fernandez, Brook Lopez, Marc Gasol, Anthony Randolph, Nicolas Batum, Mario Chalmers, and Rose all surpassed the average mark (again, Rose just barely went past this mark). Typically rookies struggle to get past the 0.050 mark, so this many above average rookies – with three playing for one team (Portland) – is not the story we generally see.
The Top Rookies at Each Position
All of this tells us that the media had a few other choices beyond Rose. In fact, as both Table One and Table Two illustrate, if all they looked at was point guards, Rose may still not have been the best choice.
Table Two: Evaluating a Few Rookie Point Guards
Mario Chalmers did not take as many shots as Rose and he didn’t get as many rebounds or assists. But he was the more efficient scorer and he did a much better job of getting steals. Consequently, Chalmers was slightly more productive.
When we turn to the shooting guards, O.J. Mayo was considered the best by the writers. In fact, five sportswriters thought Mayo was the best rookie. And Mayo – as Table Two reports — was above average with respect to scoring and personal fouls. But he was below average with respect to every other aspect of the game. Hence Mayo’s overall production – which was below average — was eclipsed by Rudy Fernandez (who was well above average for the Blazers).
Table Three: Evaluating a Few Rookie Shooting Guards
When we turn to the forwards, the media selected Michael Beasley as the top forward. Again – as Table Four reveals – the media missed the mark. Beasley – like Mayo – was above average as a scorer. But – like Mayo — he was below average with respect to everything else. Consequently, Beasley’s overall per-minute production was eclipsed by Anthony Randolph and Nicolas Batum (another Blazer).
Table Four: Evaluating a Few Rookie Forwards
When we turn to the centers we see yet another center. Table Five includes Greg Oden, who the media completely ignored. On a per-minute basis, though, Oden (again, of the Blazers) was the second most productive rookie. His primary problem was personal fouls, which reduced his minutes (the play of Przybilla also kept him on the bench). But beyond this issue, Oden is already above average with respect to rebounds and shooting efficiency.
Table Five: Evaluating a Few Rookie Centers
What All This Means
Let me close by noting again what these numbers tell us. Last year Kevin Durant was not the most productive rookie. In fact, it wasn’t even close. This result made fans of Durant very angry and many of these feel vindicated by Durant’s sophomore performance. But as I noted a few weeks ago, regardless of what Durant does for the rest of his career, his below average rookie campaign will remain.
The same story can be told about Rose. Someday Rose might be one of the top point guards in the game (although he will probably never be as productive as Chris Paul). But that was not the case during Rose’s first season. And no matter what Rose does going forward, his future production will not change the fact he was just about average as a rookie.
– 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.
Peter
April 28, 2009
This is a tentative judgment, but so far, Rose’s performance per minute in the postseason from a productivity standpoint (about .124) has been below average compared with the benchmark for point guards (.132)
That’s not uncommon, given that most players tend to suffer a decline in performance from the regular season to the postseason, but consider a few key differences (this is written through the first four games of the Bulls-Celtics series):
* Rose’s scoring average in the postseason (19.5) is better than his regular-season average (16.8), which includes his 36 point effort in the debut. This, combined with the Bulls evening the Boston series, has led most media pundits to conclude that he has stepped up his game to another level.
*Rose only taken two three-point attempts in the four games, and hasn’t launched in ball in two of those four, compared to an average 3.6 attempts in a comparable regular-season sample. Having Kevin Garnett out of the lineup has enabled him to penetrate the interior more, resulting in much of the shot deficit.
* He has shot free throws better in the postseason (84.2%), than the regular season (78.8%), in large part because of a 12-12 effort from the line in the first game.
* The most notable disparity is in the assist-to-turnover ratio. He has gone from 2.53:1 in the regular season to 1.38:1 in the postseason. Not coincidentally, Rajon Rondo’s steals per game average in the postseason (3.5) is much better than his average in the regular season (1.9)
* Rose has recorded one steal and two blocks in the series. The blocks are above the regular season average, but the steals aren’t there.
Matt
April 28, 2009
Love this site and the provoking thoughts but… given the difficult adjustments faced by most rookie PG’s – wouldn’t we say “league average” isn’t a black mark on the future outlook for his career – or an indictment for selecting him as rookie of the year?
Or are we trying to tilt too hard the other against consensus?
Jacob Rosen
April 28, 2009
The Rondo-Rose rivalry has really been a huge storyline in this post-season so far. What makes Rondo the best point in the NBA right now (I am assuming that is the case) and Rose way far behind him?
Daniel Suhr
April 28, 2009
I would like to say that I love you.
We humans have such poor analytical brains. It really takes almost a contrarian mindset to get to the truth of any matter.
Whether analyzing something as relatively meaningless as a basketball game or the spread of swine flu through the population, with us you know the first opinion will be the wrong one.
Unless you are DJ Berri.
dberri
April 28, 2009
Peter,
Thanks. Good analysis. Wished I had looked at the playoffs before posting.
Matt,
Forgot to mention this in the column. Other rookie point guards who have been named ROY were actually well above average as rookies (see Chris Paul, Jason Kidd, Mark Jackson). So it is possible for a point guard to be well above average. Derrick Rose, though, was not.
Joe
April 28, 2009
Matt,
Read the whole story please.
Erich
April 28, 2009
Dave, this is excellent and quite interesting. I am wondering if you have any comment on these season results as compared to your rookie projections made in your post from October
https://dberri.wordpress.com/2008/10/26/the-nba-rookie-polls/
Peter
April 28, 2009
No. Usually, a player’s numbers tend to go up as he gets more experience in the league. While Rose might be about average from a league standpoint, he’s above average compared to all the rookies who have entered the system as of late. If anything, his future looks bright in the Windy City.
For example, Dave’s formula considered LeBron James’ rookie season as slightly below average when compared with a typical league player. But in later seasons, he improved his numbers across the board, and his productivity followed suit, particularly with respect to rebounds, assists, and field goal percentage.
Peter
April 28, 2009
Don’t worry about it, Dave. The playoffs are a small sample, anyway. As far as the regular season is concerned, it’s a more consistent sample of Rose’s work.
I should also mention that Joakim Noah is surprisingly underrated in the postseason as compared to the regular season.
With numbers of 11 points, 11.25 rebounds, and 3.0 blocks per, he’s morphed into a poor man’s version of Dikembe Mutombo as far as Boston is concerned, in large part due to increased minutes and KG’s injury.
If the Bulls pull an upset, however, don’t expect him to repeat that kind of line if Chicago faces Orlando.
Michael
April 28, 2009
…And now Lebron is presumably (except maybe CP3) the most productive player in the league!
I assume we are waiting to reveal that after he wins the official M.V.P award? (Or maybe to debunk the conventional wisdom by presenting the wp case for Chris Paul :-D )
Benny
April 28, 2009
DBerri,
While your analysis leaves no doubt that Rose, based on WP formulas, was average, it is a far stretch to conclude that Chalmers was a more productive point guard.
Chalmers played next to Dwyane Wade, who demands double-team attention and allows Chalmers to knock down wide open shots or to cut to the basket in an open lane.
Rose does not have such a teammate. How can you incorporate that into numbers?
simon
April 28, 2009
Benny I guess you’re new to this blog,
I recommend reading through dberri’s previous postings and explanations. That question had been asked and answered continuously for the past 3~4 years on this blog. The gist is this: while thought experiments will tell you otherwise, dberri asserts that based on his research from numbers, players’ WP numbers actually do not change from season to season even though their teammates change. Is there an effect? sure, but it’s not very big according to him.
You could bring individual examples, but to really say there’s a large change in WP with different teammates, you need to bring a real evidence based on a large set of data that show different teammates indeed affects productivity to a large degree.
ilikeflowers
April 28, 2009
Benny, the general effect on WP formulas is that having to share the ball with better players suppresses any given player’s numbers. And the productivity numbers are what they are, no stretching is required. Unless Rose or Chalmers’ circumstances are unusual there’s no reason to think that Chalmers’ numbers are inflated relative to Rose’s. Regardless, seasons two and three will serve to determine which player’s ‘true’ productivity is greater.
simulator
April 28, 2009
If you actually watch the game, Chicago would be in a pretty bad shape without Rose. He’s been the unstoppable go-to guy in the crunch time when they won. Would you rather replace him with a catch and shoot type of guy in Chalmers? I highly doubt that Chicago would be tied had Chalmers replaced Rose in their line up.
Statistical numbers like eff. differential (and individual assignments like WP) do not capture enough variables to predict particular match-up outcomes. Could you possibly have predicted Chris Paul, the best PG (according to WP) in the league was going to be that bad against Denver?
Watching the game sort of reveals that CP’s small stature, Denver’s defense and Chauncey’s bullying ways may have something to do with the huge drop-offs.
WP tells you that Brent Barry for his career has been almost as efficient as KB on a per minute basis, but it doesn’t tell you that Brent Barry has a very specific complimentary role while KB has to initiate a team’s entire offense. These are very important variables that you have to take into consideration when you build a team.
I’d also like to look at fluctuations of a player’s performance. Guys like KB, while on par on a per-minute basis with Brent Barry according to WP, are capable of having monster games (while stinking up really bad in some others). I believe guys like Brent Barry would be more consistent but wouldn’t have as many spikes in game-to-game basis.
Nick
April 28, 2009
How many other rookies are the primary ballhandler for their team, and a playoff team at that?
And how many other rookies could you imagine putting up 36-11 against the best defender at his position in the playoffs?
While Rose has adjustments he needs to make, it’s clear that he’s going to be a great player. If you watch him play it’s obvious.
Rob O'Malley
April 28, 2009
I haven’t looked directly at the numbers but from watching the NBA for a number of years and reading up on statistics for a while as well my guess is that Derrick Roses game is most similar to Tony Parker. I believe his ceiling statistically (WP specifically) will probably be around a rich mans Tony Parker.
Michael
April 28, 2009
Simulator/Nick bear in mind that the data presented here doesn’t necessarily show who is the ‘best’ or ‘most skillful’ players, or who has the most potential etc, just who are the ‘most productive’ according to the parameters of the model. I agree with you that Rose looks like he will be extremely good down the line, but thats irrelevant in these analyses.
Benny, if you look at the link Chalmers win score is like 0.1 higher than Rose (+0.2 adjusted.) Hardly enough to support a statement that one is more productive than the other!
People need to start interpreting this data for themselves rather than trying to argue with the prof! He isn’t going to alter his argument or change his approach!
Also simon says “based on his research from numbers, players’ WP numbers actually do not change from season to season even though their teammates change”
Are team mates not players?
brgulker
April 28, 2009
I should also mention that Joakim Noah is surprisingly underrated in the postseason as compared to the regular season.
I about pee’d myself when I saw that today on TNT, mostly because I gave one of my Chicago buddies so much grief when Chicago made the number 9 pick out of him.
If he keeps that up, though, it may have been a pretty good pick.
————————
Another great post, dberri.
One thing I am trying to figure out, though, is why there are so many productive big men? Is this just an isolated occurrence this season? Or, does this system as a whole tend to favor bigs?
If this has been discussed before, I’d be happy to do my own reading if someone can link me to it.
———————
@ Nick,
While Rose has adjustments he needs to make, it’s clear that he’s going to be a great player. If you watch him play it’s obvious.
With respect, if you’d take some time to read some of the intro posts to this blog (explaining win score and WP48), you would see that this metric has been developed largely to show that your sentiments are flat-out wrong — largely because you have equated scoring totals with greatness.
Do some searching on this blog for Allen Iverson, and be open-minded about it. If you are, it will be hard to make claims like you just did in the future.
simon
April 28, 2009
Michael,
“Are team mates not players?”
Could you clarify that sentence? I don’t understand what you mean by it…
Also do note that I did say, and dberri has mentioned many times, that the classic diminishing return has been observed with regards to WP and teammates albeit the effect is small.
.
April 28, 2009
and eric gordon gets no mention why? how?
brgulker
April 28, 2009
and eric gordon gets no mention why? how?
Did you look at the tables? Pretty self-explanatory, really.
Italian Stallion
April 28, 2009
“based on his research from numbers, players’ WP numbers actually do not change from season to season even though their teammates change”
The problem with statements like these is that even though they may be true in aggregate, they are often not true in very specific circumstances.
This is a lot like the Per 36 minutes debate.
In aggregate Per 36 minute stats do a very good job of predicting the productivity of players that eventually get more minutes.
However, you can always find a guy like Shaq whose minutes are declining at this late stage of his career because if he played more minutes consistently it would have a negative impact on his body and he would decline sharply or soon be out of action. He is also given the day off when the Suns have a back to back for the same reason.
You could even point to a laughable guy like Eddy Curry (and others with clear cut conditioning issues). A couple of years ago when he played a lot of minutes his raw stats DID improve as Per 36 predicted, but it was also the last time he was in any kind of shape. Does anyone think that he was in good enough shape the following season to play the same minutes without being carried off in a stretcher while being given oxygen?
You can also point to some specific role players that only get minutes when they match up very well and/or it’s garbage time.
To me, there is ZERO DOUBT that guys like Wade, Bryant, James, Jordan, Hakeem etc… draw a lot of double teams and get better looks for their teamates than they would get on another team.
Exactly how much it adds to each player would require tracking how often it actually happened for each and the FG% in those situations vs. normally etc… But on a team level, it clearly happens often enough to matter. One need only watch a handful of games to see that value.
Statistics are a lot like a woman in a bikini. They reveal a lot, but not everything. Sometimes you might want a closer look. ;-)
simon
April 28, 2009
IS, I understand your point but this is about statistical modeling and how the discipline of Economics as a social science works in general. Do keep in mind that dberri is an economics professor first and his approach is based on his profession.
A more similar example would be in the medical science where many people disparage the traditional Western science paradigm of being impersonal and not individual-based enough, and love to bringing out individual examples to back up whatever non-science cure they think that work.
Sure the individual cases might differ, but how do you know which individual cases work and which ones do not? What’s the point of having statistical model in the first place then? Bsically you’re back to trusting your own gut feeling and subjective judgment, which is precisely the things dberri tries to point out to be inefficient.
Nick
April 28, 2009
@brgulker
I have been reading this blog for several months, as well as the intro posts, and I fully understand that Allen Iverson is overrated.
Allen Iverson is a one-dimensional player who can’t do anything but score.
Derrick Rose clearly has the potential to be a great all around player, contributing across the board. Scoring is only one facet of his game. Instead of falling back on this single, all-encompassing metric, why don’t you point out specific aspects of his game that need improvement, after all, that’s what you expect from a 20 year old rookie. Going into his sophomore year, what does Rose need to do to improve his overall productiveness? If you can identify areas for improvement, then we can talk about whether or not he has the physical/mental ability to improve in those areas. My point is that by watching him play, one can see that he without a doubt has flashes of greatness, which show that he can indeed improve greatly, especially in consistency.
Surely you must admit that there are things immeasurable by WP48. For instance, it only uses turnovers and assists to judge passing skills and court vision (or at least I assume so because those are the only numbers that are kept on that). What about second assists? What about missed shots after good passes? And one thing that rookies most often improve on later in their careers is turnovers, certainly you’d expect that.
You make it sound that if Rose does indeed evolve into a superstar, it will be a coincidence, and you try to distance that possibility as much as you can from his rookie season. However, does that appear to be accurate with Kevin Durant? Or Lebron James for that matter?
WP48 is certainly an interesting and effective metric, however, it is far from perfect.
Italian Stallion
April 28, 2009
Do any of the loyal followers of this model (which I believe is excellent overall) ever even contemplate the possibility that while the general public and some other models overrate scorers, that this model underrates them?
IMHO (as often stated), this model over compensates for the general flaw of overvaluing inefficient high scorers by overvaluing low usuage efficient scorers and undervaluing reasonably efficient high scorers.
The current economics of the game motivate players to score, but if the economics were changed to motivate players to be lower usuage and more efficent players, many that are rated weakly here could easily improve their rating BUT WORSEN THEIR TEAM!
Almost all teams need a couple of high usage scorers. Those are the guys that allow the low usage highly efficient role players to have value.
Italian Stallion
April 28, 2009
Simon,
I understand your point, but at times it seems the evidence is so clear it’s almost foolish to ignore it. I think the real problem is that the type of analysis I am suggesting is only as good as the analyst himself. But IMHO, it’s certainly capable of superior results in the right hands.
Most people might be better off investing in indexes by dollar cost averaging, but guys like Warren Buffett are obviously better off looking at individual companies and circumstances.
Anon
April 28, 2009
@nick
as I understood the post, dberri simply said that Rose, compared to the average NBA pg, was average with respect to just about everything. He doesn’t do anything terribly, but he isn’t great at anything either. He undoubtedly can improve in a number of things across the board.
And I don’t really understand why it’s so hard to understand that dberri’s model says absolutely nothing about Rose’s potential to be an amazing player in 4 years or whatever once he has some more experience. The only claim was that RIGHT NOW he is an average NBA pg (which makes him above average for a rookie). I’d say it’s very likely that Rose improves, but that doesn’t mean that he wasn’t only average in his rookie year.
also, i’d guess that super minor things like “missed shots after good passes” are canceled out by equally trivial things such as “made shots after bad passes.” Things like that affect everyone just about the same, and there’s not evidence that I know about that shows that it affects Rose more than anyone else.
Noah Snyder
April 28, 2009
I remember several recent examples of players who have won awards “one year too early” by your analysis: Monta Ellis for most improved player and Kevin Durant for ROY. Is there a good model for predicting future performance for young players that captures some of these examples? Might Rose fit into the same pattern?
Matt
April 29, 2009
For a 20 year old rookie, playing at a “league average” level at point guard – is for better or worse – regarded as being a greater accomplishment then what Love does cleaning up the glass w/greater efficiency at the PF spot. I’m not sure how true or not true that may be – but isn’t Rose’s ROY award more a reflection of the consensus view that playing PG in the NBA is a more vital role than what Love does and doing so at a “league average” level is a greater accomplishment?
I think that is the true debate here vs. Rose being a better or worse player than Love or being league “average”.
I would agree that Love is a better player than Rose.
Michael
April 29, 2009
Sorry Simon I misinterpreted what you wrote! You were talking about players WP remaining constant when their team mates change, which I for some reason interpreted as the team mates WP’s changing rather than the actual team mates themselves :-0
As for this argument, the guy who brought up Lebron James’ rookie WP already killed it…Noah makes a good point also….Rose looks really good, even if his wp was in the minus he’d still look good going forward.
Tim
April 29, 2009
Love, Lopez, and Gasol all played for losing teams that missed the playoffs. Even so, Lopez at least got some consideration. Chalmers wasn’t that much better than Rose, and gets overshadowed by Wade. I’m not disputing that Rose was average, but I think the fact that he led his team to the playoffs without the help of a superstar made a big difference in the voting.
brgulker
April 29, 2009
Derrick Rose clearly has the potential to be a great all around player, contributing across the board. Scoring is only one facet of his game. Instead of falling back on this single, all-encompassing metric, why don’t you point out specific aspects of his game that need improvement, after all, that’s what you expect from a 20 year old rookie.
I suppose my disagreement with you, Nick, can be put fairly simply.
My point is this: if he is truly to become a “great” player — and not just a good player — then I don’t think it’s unreasonable to ask why he posted numbers that were just above average this season. In almost every category, he’s average.
You ask what parts of his game would need to improve, and if we’re going by the numbers, I think it’s only fair to say that every single facet of his game needs to be improved drastically before he can be considered to be a “great” player.
Furthermore, as best as I can tell, Rookie of the Year should be given to the best rookie — and according to Dr. Berri’s model, Rose is nowhere close to the best rookie. Again, he’s an average point guard by NBA standards; whereas, guys like Kevin Love or Brooke Lopez actually eclipse average by quite a bit.
IMO, this article demonstrates that when it comes to player evaluation, people favor scoring averages and totals above all else.
My point is not to say that Rose doesn’t have potential — he obviously does. But potential is just that. Yes, you can tell just by looking that he is talented — he’s athletic, he seems to understand the game fairly well, he finishes around the basket well, etc.
But that certainly doesn’t make him a great player now, and it doesn’t guarantee that he will become one in the future.
Maybe time will prove you right, and maybe it won’t. But, only time will tell.
I tend to be a bit skeptical about him ever becoming a “great” player, but I would think he will become a good player.
brgulker
April 29, 2009
@ Tim: Rose did not lead his team to the playoffs. That’s absurd. He’s not the best player on the team, and he’s hardly the team’s emotional leader. What evidence could you give that Rose actually led the playoff charge?
brgulker
April 29, 2009
Do any of the loyal followers of this model (which I believe is excellent overall) ever even contemplate the possibility that while the general public and some other models overrate scorers, that this model underrates them?
IMHO (as often stated), this model over compensates for the general flaw of overvaluing inefficient high scorers by overvaluing low usuage efficient scorers and undervaluing reasonably efficient high scorers.
The current economics of the game motivate players to score, but if the economics were changed to motivate players to be lower usuage and more efficent players, many that are rated weakly here could easily improve their rating BUT WORSEN THEIR TEAM!
Almost all teams need a couple of high usage scorers. Those are the guys that allow the low usage highly efficient role players to have value.
I completely disagree. If anything, this model values scoring as high as any model, but instead of valuing totals and averages, it values efficiency.
Let me give two examples.
First, let’s take the SA Spurs. You’ve got Tim Duncan, Tony Parker, and Manu Ginobili.
None of those three guys is ever among the scoring leaders in the league. But, they are all efficient offensively, and their team has been a perennial powerhouse. Do you think that’s coincidence?
And take the LA Lakers. I can’t seem to find the stat online at the moment, but I did see it posted on SportsCenter earlier this week.
Over the past two seasons, the Lakers are overwhelmingly a better team when KB takes 20 or fewer shots, and when he does take 20 or fewer shots, he usually shoots a high percentage. My question again is, do you think that’s coincidence?
Matt
April 29, 2009
His playing PG as a rookie and doing it in an “average” fashion was just as important as his scoring if you read the NBA papparazzi (sp?) I believe….
Michael
April 29, 2009
“IMO, this article demonstrates that when it comes to player evaluation, people favor scoring averages and totals above all else.”
Because OJ Mayo won rookie of the year?
“Over the past two seasons, the Lakers are overwhelmingly a better team when KB takes 20 or fewer shots, and when he does take 20 or fewer shots, he usually shoots a high percentage. My question again is, do you think that’s coincidence?”
So you are a believer in the effect of diminishing returns?
Anon
April 29, 2009
actually, as much as I think Kobe Bryant is really overrated, I’ve noticed that a lot of the time when he takes like 30 shots, it’s because no one else is doing particularly well, so he feels it’s his job to take a bunch of terrible shots.
so as opposed to him taking too many shots and making the lakers do bad, it’s just as likely that the lakers are doing bad and thus he takes a bunch of bad shots.
brgulker
April 30, 2009
So you are a believer in the effect of diminishing returns?
Jury’s out on that one. Certainly, there is some merit to the dimishing returns argument, albeit I haven’t been convinced one way or another as to how much.
But that wasn’t necessarily my point. My point was to say that Kobe Bryant is a very efficient scorer when he shoots 20 or fewer shots. When he does so, he shoots a very high percentage, and his team wins. Again, I’m citing ESPN, even though I can’t find the specifics on the web for some reason.
My point is that when KB is an efficient scorer, his team wins much, much more than when he loses.
And that point was meant to be set over against the claim that a team needs at least one high volume shot-taker and scorer in order to be effective in the NBA.
I think “need” is far too strong, and I used SA and LA as examples, and the LA example specifically because LA is at its best when the offense is spread around and not centered only on one guy.
As another example, you could take the Denver Nuggets this year. Melo’s scoring average and shot attempts are down this year, but the Nuggets are better than they have ever been. Coincidence? Personally, I don’t think so.
actually, as much as I think Kobe Bryant is really overrated, I’ve noticed that a lot of the time when he takes like 30 shots, it’s because no one else is doing particularly well, so he feels it’s his job to take a bunch of terrible shots.
so as opposed to him taking too many shots and making the lakers do bad, it’s just as likely that the lakers are doing bad and thus he takes a bunch of bad shots.
I would think there’s some merit to this, although I wouldn’t have a clue to go about tracking or measuring it.
Michael
April 30, 2009
“I haven’t been convinced one way or another as to how much…My point was to say that Kobe Bryant is a very efficient scorer when he shoots 20 or fewer shots. When he does so, he shoots a very high percentage, and his team wins.”
From Wiki “In economics, diminishing returns refers to how the marginal contribution of a factor of production usually decreases as more of the factor is used.”
So yes you do. I think that if you are going to completely disagree with Stallions point that usage can seemingly skew the verdict given by the wp metric either for or against a player, then you should bear in mind that being a proponent of diminishing returns (as it relates to shooting efficiency) contradicts that assertion.
Basically if you have a model which ignores diminishing returns/shot creation etc as a factor and focuses purely on the league/position average for efficiency then inevitably some players will be underrated because the effect of diminishing returns will lower their efficiency further towards and below the average level the further their production goes above it. Barring some idealised version of basketball in which skill is evenly distributed amongst all, there will always be players who are more skilled than others and therefore more able to shoulder a greater burden on offense (high usage scorers.) The production of these players allows for the efficiency of surrounding lower usage scorers to increase because they can maintain their contributions at the optimum level for their particular skill set. Therefore In my opinion Stallion’s point that wow ignores factors such as this, and thus rewards players who focus on production in the less risky areas disproportionately is a good one, that he has made well here many times previously. Also his statement in regards to the effect of installing a wowcentric salary system in the nba is spot on.
As far as the Nuggets go, firstly you should know that their improvements this year are defense more than offensive. Old Allen Iverson for old Chauncey Billups was really a steal in this regard, and in addition they have had solid production from Nene and Birdman as well. One thing to remember when you mention the Nuggets is that they lost Marcus Camby, who according to wp is one of the leagues very best players, and not only did they not get worse, but actually improved. Camby is the poster boy for the kind of player which can be overrated by this system, so instead of focusing on Melo’s shot attempts perhaps you should ask yourself if a team losing such a great player and actually becoming better is a just a coincidence, or do people like Stallion have a point when they argue that all may not be as it seems in the statistic?
Finally I don’t mean to should like I’m teasing but I can’t resist, you wrote; “My point is that when KB is an efficient scorer, his team wins much, much more than when he loses.”
Since Kobe plays for his team, I don’t think they ever win when he loses! :-D
mrparker
April 30, 2009
Michael,
So I guess you’re not going to give any credit to the addition of Balkman for 53 games, Nene for 71 games or the subtraction of Diawara(sp?) for 54 games any credit for Denver’s improvement on the defensive end?
Just some quick math but Camby was worth 6.4 defensive win shares last year and Balkman(1.3)+Nene(3.8-.5(16 games last year))+Anderson(2.9) added 7.5 from new guys this year.
I follow you on diminishing returns to a point. The argument seems to be about at which usage the law of diminishing returns takes affect. Noone who argues against wp48 has an answer for that. They usually just assert that basketball couldn’t exist without these players.
James
April 30, 2009
I think that Mr. Parker is right by pointing out that base on who the nuggets got (Anderson, Nene being healthy, Balkman, etc.) in fact eclipsed the defensive output of Camby. Camby is a great player by this model and since statistical +/- isn’t available to the public I couldn’t tell you how well he is rated by that system but PER and Win Shares think that Camby is at the very least quite a bit above average. What Dberri’s suggested through his model is that Billups would make this team better than they were with Allen Iverson. He did not state how they would get better, just that they would be better. WP explains what happened, but it does not explain how it happened very well.
Just an interesting tid bit (and this doesn’t really confirm anything, but is still kind’ve funny based on the gnashing of teeth over good defensive/bad offensive big men done by opponents of WP) Who was one of the highest rated players by statistical +/- during his prime? Ben Wallace is who.
Just found that a tiny bit funny.
Michael
April 30, 2009
Hi Mr Parker
Sorry I forgot about Balkman. I did mention Nene and Anderson though :-)
James F.Y.I Neil Payne from Basketball Reference has made quite a few posts on statistical +/- recently
http://www.basketball-reference.com/blog/?cat=22
SPM really shows how much Jordan absolutely dominated across his career. Among contemporary players it seems to really like Lebron James (no suprises there!)
Tom Mandel
May 1, 2009
Dave — I don’t understand your numbers on JaVale McGee. In 1143 minutes, I have him w/ a 10.86 Win Score, or .226/min. That’s just slightly better than an average NBA center (10.8 — took that number from a previous table of yours) and after calculation results in a WP of .106 — not .63 as you have it.
My formula for calculating WS is accurate (results usually match yours), and my data comes from the NBA’s own player page.
I’m a Wiz fan and so mostly interested in JaVale, but I wonder as well how your data gets loaded in and whether there’s an error somewhere.
Thanks!
Tom Mandel
May 1, 2009
I rechecked the data for JaVale — it’s the same at basketball reference as it is at the nba’s site (actually, BR has 1 fewer FTA for him).
I checked my WS formula for him — it’s correct. He comes out to 10.88 using BR data, which becomes .227 when I divide by 48.
I subtract your position adjustment of .225 for a center, (leaving .002) then use your formula ( 0.104 + (1.621*PAWSmin) ) to convert the resulting PAWSmin figure to WP48. The result, obviously, is .107 — but you have .065
This gives him something over 2.5 wins produced, rather than the @1.5 you have.
How do you load data?
dberri
May 1, 2009
Tom,
The data comes from basketball-reference. You need to remember that the link between PAWSmin and WP48 is an approximation. The actually calculaton of WP48 involves adjusting for team defense.