Why Carmelo Anthony Is the Ultimate Team Player (and What ‘Advanced’ Stats Miss About Him).
The ‘advanced’ stat in this article from Nate Silver, is Wins Produced. And here is what Silver argues Wins Produced is missing.
What is missing from formulas like Berri’s is an account of what Anthony does to the rest of the Nuggets. Because he is able to score from anywhere in the court, Anthony draws attention and defenders away from his teammates, sometimes leaving them with wide-open shots. He also allows them to be more selective about the shots that they choose to take, since they know that Anthony can usually get a respectable shot off before the 24-second clock expires if needed.
Silver goes on to provide evidence – detailed in the following table — supporting his contention.
And then Silver argued…
The effect of a player who improves the rest of his team’s TS% by 3.8 points is extremely substantial: it is works out to their scoring about 5 or 5.5 additional points per game solely on the basis of this efficiency gain. That, in turn, translates into about 15 additional wins per season for an average team, according to a commonly-used formula. This is how Anthony creates most of his value — not in the shots he takes himself, but in the ones he creates for his teammates – and some of the “advanced” formulas completely miss it.
Wins Produced argues that Carmelo Anthony has produced 33.5 wins across his eight seasons. And that means Anthony is hardly an elite player. But Silver argues that Anthony –because he increases the shooting efficiency of his teammates — is worth 15 additional wins per season. So Melo – according to Silver’s analysis – is clearly elite.
Silver’s argument certainly reflects conventional wisdom and it appears supported by some empirical evidence. Unfortunately – as people have noted since Silver’s story appeared – there are some problems with this analysis.
Before I get started, let me first apologize for the length of this post. Explaining the issues with this analysis takes more than a few words, so this post may take up some of your time (that is, if you wish to read all of it).
And before I get to the problems, let me also note two basic issues one needs to think about in considering such a study. The first is statistical significance. Or more simply, can we confidently declare the estimated relationship actually exists (i.e. is different from zero)? The second issue is economic significance. Or (again) more simply, what is the size of the estimated effect?
If we determine that the estimated relationship probably isn’t real (i.e. isn’t statistically significant), than the second issue isn’t important anymore. And as I will note, I do not think this relationship actually exists. Nevertheless, I want to start by noting that even if one insisted that Silver’s simple approach was ‘best’ (as I will note, this approach is not the ‘best’), his calculation of the size of the effect failed to take into account two obvious differences in the players included in the sample.
Calculating an Average
The reason I want to start with the size of the effect is that I think Silver does something ‘odd’ in his calculation. The table above presents the change in TS% for sixteen players. Some of these players – like Marcus Camby, Nene Hilario, and Kenyon Martin – have played more than 10,000 minutes with Carmelo Anthony. Others – like DeMarr Johnson, Voshon Lenard, and Aaron Afflalo – played less than 3,000. Given these difference in time spent with Melo, one might expect the analysis to take this into account. But from what I can tell, all Silver did was calculate the simple average change in true shooting percentage. So whether they played 10,140 minutes (like Kenyon Martin, whose TS% appears to improve 1.8% because of Melo) or 3,276 minutes (like Greg Buckner, whose TS% appears to improve 8.9% because of Melo), the impact each player’s change has on the overall Melo Effect (the 3.8% impact estimated above) is the same. One would expect, though, that the analysis should at least present a weighted average. And if we weight these numbers by minutes played, the reported effect does fall to 3.5%.
This is not much of a decline. Then again, weighting by minutes is not the best approach. This issue here is shot attempts, so a better weighting scheme is to adjust by how many field goals a player is taking. I didn’t feel like gathering all this data, but I would note that the three of the four players with the largest Melo Effect – Anthony Carter, Greg Buckner, and Chris Andersen – are not known for taking many shots. Specifically, in the eight seasons where these three players appeared in at least 50 games, none of these players ever averaged more than 6.8 field goals attempted per game for a season [6.8 is the mark Carter posted in 2007-08]. And in five of these seasons, the field goals attempted per game was 4.3 or less.
Again, according to Silver, this trio has three of the four largest Melo Effects. But even if we could argue that increased shooting efficiency we observe for these players is entirely about Carmelo Anthony (and again, I will note in the moment that this is unlikely), if these players don’t really shoot much then the change in shooting efficiency noted can’t matter much. Given this observation, one might expect a simply adjustment for the number of shots each player takes. But again, all we have is a simple average.
Is it All About Melo?
Unfortuntely, even if the weighting of the average was correct, there is a much bigger issue to consider. As a number of people noted, Silver’s analysis doesn’t consider any other factors. He argues that the changes we observe in each player’s TS% is entirely about Carmelo Anthony. But player performance could change for other reasons. And because other factors could matter, the analysis of the Melo Effect is incomplete – and quite misleading – if no effort is made to control for the other factors.
To illustrate this point, let’s briefly talk about the study I published on NBA coaching (with Michael Leeds, Eva Markova Leeds, and Michael Mondello). The purpose of this study (discussed in Stumbling on Wins) was to examine how coaching impacted player performance. At the onset of the study we first report how player performance changes when the player comes to each coach in our sample. This analysis did not initially consider any controls. And the coach that we report having the largest effect was Dan Issel. Of the fifteen players who came to Issel, twelve posted higher per-minute performance.
If we were following Silver’s example, we would have stopped at this point and declared Issel the greatest NBA coach [across our sample from 1977-78 to 2007-08]. As we note in the paper, though, other stuff matters. And when you control for past performance, age, injury, etc…) the impact of Issel vanishes (i.e. Issel’s impact was not statistically significant) and the top coach – according to our analysis – is Phil Jackson.
Of these ‘other factors’, age appears to be one factor Silver should have considered (and people noted this issue as well). After Buckner, Carter, and Andersen, the top seven players in the Melo Effect rankings includes J.R. Smith, Nene Hilario, DerMarr Johnson, and Aaron Afflalo. Here is how old each player was when he first became Melo’s teammate.
Smith: 21 years
Hilario: 21 years
Johnson: 24 years
Afflalo: 24 years
Player performance in the NBA – as reported in Stumbling on Wins – tends to peak in the mid-20s. So each of these players was at an age when improvement in performance was still likely to occur. To estimate the size of the Melo Effect, the impact of age needed to be considered.
And that is what I attempted to do. Utilizing the same data set employed to study coaching [i.e. data on players from 1977-78 to 2007-08], I looked at the factors that explained a player’s TS%. The explanatory factors I considered included past TS%, age, game played (to capture injury), etc…. In addition, I considered a dummy variable, equal to one if a player was in his first year as Carmelo Anthony’s teammate. If the estimated coefficient for this dummy variable is statistically significant (and positive), then we can conclude that Silver is on to something. When the model was estimated, though, the Melo dummy variable was clearly insignificant. In sum, it doesn’t appear that a player’s TS% — when we consider a number of factors that impact player performance – is impacted by joining at team with Carmelo Anthony.
One should note that even if the estimated coefficient was significant we still wouldn’t have been able to conclude that there is a Melo Effect. Again – as people noted – Melo is not the only factor unique to Denver. The Melo Effect – if it existed – could have been the George Karl Effect. Or it could have been the Dean Oliver Effect (Oliver is the author of Basketball on Paper and he does statistical analysis for the Nuggets). Or it could have been the altitude in Denver, or any other factor unique to Denver.
Although this exercise failed to uncover evidence of a Melo Effect, it does serve to highlight an important point about statistical analysis. If we wish to understand how one factor impacts another, an effort must be made to control for other explanatory factors. Silver’s analysis didn’t control for anything. As a consequence, his estimate for the existence and size of the Melo Effect appears to be incorrect.
Quoting from Others
As noted, I was not the only one to note problems with this analysis. So let me close by noting some of the other points people have made (some of this echoes what I said above).
Let’s begin this tour with links to the words of Andres (Dre) Alvarez (from Nerd Numbers) and Arturo Galletti (from Arturo’s Brilliant Stats).
Dre — Silver and Gold: Prospecting Melo’s Past — looks at Melo’s history in Denver. And Arturo — in Fanservice: Followup notes on Melo, Rookies and A simple response to Mr. Silver – looks at how TS% for players in Denver changes with and without Melo. For this interested in more on this topic, these are excellent reads. Dre’s point that Denver’s success with Melo is not all about Melo is especially important.
Beyond these posts, let me also reports some of what I have seen in the comment section (and in the interest of space, these are all just partial quotes; please read the comment section for all that people had to say):
from ilikeflowers
- Don’t you need to look at a bigger population of scorers who might make their teammates better before coming to this conclusion? If you examine n guys who fit the Melo profile and the results in general argue against this teammate effect then isn’t this effect likely to be happenstance (or unpredictable)? And then what about the same effect from lesser (presumably more numerous and cheaper) versions of Melo, so that we can determine the marginal value? What about the effects of Melo’s backups with Denver over all these years? This analysis has all the drawbacks of plus-minus. Of course if the results from a statistically significant sample size then confirm this effect then it’s certainly a valid statement.
- I’m stunned at this particular statement [“In taking all of those shots, however, Anthony has also done something else: he’s made his teammates much more efficient offensive players”]. How can he make a statement on causality based upon the evidence that he presents? It would have been so much better for him to just say: ‘When Anthony is on the floor his teammates are much more efficient offensive players’ and then let his audience make of that what they will.
from John Giagnorio
- Why does Anthony get credit for improving his teammates even when he is not on the court? Was it that difficult to break down the data further?
- Why use TS% instead of eFG%? Does Anthony deserve credit for his teammates becoming better free throw shooters? Look at Iverson’s age 22 and age 32 seasons. The eFG% is identical, but he’d become a much better foul shooter.
- Take a look at Nene’s career on basketball-reference. He’s played all of 1 year without Anthony, yet his TS% didn’t really improve until 08-09.
from Italian Stallion
- High usage scorers that often get doubled should be in a position to get a lot of assists, but Melo doesn’t.
- The article gives all the credit for the improvement in the TS% of his teammates to Melo when it could easily be partially be Billups (an underrated PG), better coaching, a combination of players, or random.
- IMO he should not have compared a players lifetime TS% to his TS% with Melo because the one thing most good players do over time is improve their shot selection and shot. So most good players improve their TS% also as they develop. He is probably giving Melo credit for the natural improvement of the players.
From Philip
- TS% is only a part of what determines wins, acquiring the ball and preventing your opponent from doing so also are important, or so I’ve read in one of the basketball blogs that I frequent. So even if had demonstrated that Anthony improves his teammates’ scoring efficiency, it’s a narrow view; what if Melo’s teammates are rebounding less and turning the ball over more?
- While I agree that it’s pretty laughable that Melo could impact his teammates’ FT% (what, does he give better high-fives between shots?), he could conceivably improve their FT rate by passing them in a situation where they are more likely to be fouled while shooting. This would improve their TS%, though not their EFG%. However, Silver has failed to show any correlation between playing with Anthony and an increase in FT rate, much less a causative link.
from Peter (commenting at Nerd Numbers)
- Even with the changes in true shooting percentage, as a stat major, there is also the concept of statistical significance. In a nutshell, yes, most of the players that played with Anthony improved their true shooting percentage. But it is possible that, at least for some players, it is highly likely that their improvement is not “significant”, that is, they could have had that performance with or without Anthony based upon measuring their previous performance. And if they could have had those performances with or without Anthony, then Anthony was not the reason why they shot so well.
- Expounding on the previous point, two of the players with the biggest gains, Nene and J.R. Smith, played some of the fewest minutes of the group before Anthony arrived. Even with their great gains, at least some of their improvements may have been age-related, not Anthony-related.
- The author only addresses shooting. Winning basketball games entails great shooting, obviously, but it also requires rebounding, assists, etc. When you look at Anthony’s non-scoring statistics, for example, he is below average with respect to the average shooting guard in turning the ball over and fouling, which are activities that do not help the Nuggets win, let alone help his teammates shoot. The author even admits that Anthony’s assists are below average with respect to other scorers such as Bryant and James. Besides, the reasoning behind Berri’s metric is that scoring *isn’t* all that there is to winning games, and as such, he tries to reward players who contribute in ways other than scoring, which hurts Anthony.
When we look at these comments we certainly see some similarities. A number of people have noted that Silver failed to show causality (so he overstated his case), failed to show statistical significance, and failed to control for other factors (like age).
Let me close by repeating something others have noted. During the past two elections I have enjoyed reading Silver’s analysis (and the analysis from other people at FiveThirtyEight). And I want to emphasize that what ever you think of Silver’s analysis of Carmelo Anthony, the analysis of the Melo Effect doesn’t tell us anything about the quality of analysis offered on other topics. In other words, it is incorrect to argue that because Silver may not have gotten this story right, all the other stories he tells also have problems. Such an approach would be drawing an inference from a sample of one. And yes, a sample of one isn’t statistically significant as well.
– DJ
P.S. Again, sorry for the length of this post. It is more than 2,700 words and if you got to the end… well, I am not sure this was the best way to spend your time. For my next post I will try and say less (and I hope not to use the word “Melo” at all).
welbilt bread machine
January 17, 2011
I just don’t buy it. It’s the same argument as the Allen Iverson effect. He’s such a great scorer that it make’s it easier for everyone else. Yes, but that’s only one part of his game. What about defense? To be a “superstar” you have to impact offense and defense, not just one side of the ball. -alfredowelbilt
marc
January 17, 2011
What’s the effect the presence/absense of nene or billups on his teammates?
Adam C. Morrison
January 17, 2011
“It is more than 2,700 words and if you got to the end… well, I am not sure this was the best way to spend your time. ”
Do people really read that slowly?
Ben
January 17, 2011
Great analysis, thanks Dave.
Would you mind putting up the dataset that you used. I’d like to take a look at the numbers myself.
Also, any chance that you could comment on why Phoenix has gotten worse if it isn’t Amare Stoudemire? It seems like the same issue.
Sam Cohen
January 17, 2011
I’m assuming it would be fairly time intensive, but it would be really interesting if this analysis (or a similar analysis of WP48 rather than TS%) could be done for every player in the NBA. Just like a few coaches do actually impact the performance of players, there might be a few players who impact the performance of their teammates. (And if there are, it’d be interesting to see if those players have any similarities in common– i.e., are they the high-volume scorers or are they proficient rebounders, etc.)
G Wolf
January 17, 2011
So basically:
“I mention several times how long this post is, and obviously spent several hours on it, but didn’t feel like taking the additional 5-10 minutes to look up their actual shot attempts from a well-know, publicly-available source. Oh wait, I ended up doing it for only a few players to prove my specific point, but still didn’t feel like spending the additional 5 minutes doing it for the rest of the players.”
And we’re supposed to take you a face value when you claim you are a serious academic researcher and above the riff-raff of “a small on-line community who call themselves APBRmetricians”? Is this serious academic research…insisting you can’t be bothered, merely because of your caprice, to actually substantiate a claim you’re making, but then cherry-picking through some of it anyway?
dberri
January 17, 2011
Farkas,
Were you this upset about what Silver did?
nerdnumbers
January 17, 2011
G Wolf,
I like it! You quote a point in the article. Don’t address the point. Then point out there is additional data, without using the additional data. You then bring up the credentials game (which unfortunately does not make an argument more or less valid). In short you spent 3 paragraphs “arguing” without really arguing. Impressive.
Btw, DJ linked a few other analysts that may have some of the data you want. Also to use a common Adam Morrison line “Citation Needed” on the claims DJ has made on the APBR community. I’d love to read those if you have links.
Italian Stallion
January 17, 2011
I enjoyed this one a lot because it brought up a couple of points I didn’t consider earlier (though I have convinced myself I would have thought of them eventually). :-)
I think there is some empirical and visual evidence of this effect for several elite NBA players, but that article simply did not prove it or measure it accurately in the case of Melo. It would not shock me if he does have some smaller positive impact though.
Cheech Cohen
January 17, 2011
Great post, Dr. Berri. Exactly what I hoped for.
arturogalletti
January 17, 2011
IS,
I found some evidence of a very small effect but Melo’s inefficiency ate the entire thing up. I need to do something similar for Lebron in Cleveland.
ilikeflowers
January 17, 2011
…it is incorrect to argue that because Silver may not have gotten this story right, all the other stories he tells also have problems.
This may be true, but I would argue that if the range of Silver’s story quality was between High and Poor before this one, that now the range of quality is between High and Abysmal. This means that prior to this story we would have expected a higher average quality going forward than we do now. The significance of the change depends upon the sample size but certainly the direction of the change is negative. Silver may end up being more Kevin Love than Corey Brewer, but you can only have so many Lousy Games before the fans stop paying attention.
CHris ROck
January 17, 2011
Excellent Post, Sir.
Mike
January 17, 2011
@Ben have a look at this:
http://nerdnumbers.com/automated-wins-produced
In 2010, the suns had 7 players above average and two others, Frye and Lopez, close to average. In 2011, the figure is only 5, and one of those, Childress, never plays.
Losing Amar’e hurt, for sure, but the biggest difference is that most everyone got worse, and the Suns traded Amar’e and Amundson, 2 above average PF/Cs, for Warrick (below average) and Hedo (was awful in Phoenix). That, and everyone on Phoenix is a small forward or shooting guard :)
reservoirgod
January 18, 2011
In a surprise move, Kevin Pelton comes to the defense of his old boss . But at least he included TOs to evaluate the change in offensive efficiency of Melo’s teammates. And of course he ignored defensive efficiency. But hey, WoW Journal readers already know scoring dominates NBA analysis. http://www.basketballprospectus.com/article.php?articleid=1412
Italian Stallion
January 18, 2011
Mike,
“Losing Amar’e hurt, for sure, but the biggest difference is that most everyone got worse, ”
This is part of the same debate.
Wilson Chandler’s and Raymond Felton both have higher efficiency this year and both have directly credited Amare presence for getting them better shots.
DSMok1
January 18, 2011
DBerri,
I rarely post over here, but I would appreciate it if you would investigate the empirical points that Kevin Pelton brought up, and perhaps formulate a response!
Thanks!
Daniel
dberri
January 18, 2011
DSMok1,
It is amazing how polite you are in this forum. And yet in the APBR chatroom your group seems to rely so heavily on insults. Why the difference in behavior?
As for Pelton’s analysis… I wonder about the minutes — and circumstances — each player actually plays without Anthony. There are also the problems with plus-minus analysis (and adjusted plus-minus) that have I have detailed many times before.
One also wonders why such an effect doesn’t show up in the aggregate data. As I have noted in the past… I have looked at the usage argument several times. The evidence for this seems very weak. Certainly the strength of the belief people have in this idea seems inconsistent with the actual evidence.
DSMok1
January 18, 2011
DBerri:
Thanks for the reply! Please consider each person at APBRmetrics separate from the other. I personally do not insult others. I do agree that discussions in most internet forums (APBR sometimes, your comment threads here sometimes) tend to devolve into unpleasant ad hominem attacks at times. A few people posting inappropriate things may inaccurately reflect on the character of the rest. I find APBRmetrics a valuable source of highly analytical commentary on and discussion of my own research.
I fully agree that there are significant collinearity issues with +/- data; nevertheless, it is, when properly handled, an extremely valuable independent (from box-score statistics) cross-validation tool. There are not many ways to completely independently validate box-score metrics.
It is reasonable to wonder about the minutes and circumstances on such on-court/off-court player pair assessments. Nevertheless, don’t you think the consistency of such effects in this investigation indicates the likelihood some underlying signal? If I look at this subject as a Bayesian updating problem, my prior ideas on the matter probably still out-weigh this new look, but it still bears further research.
Personally, I currently believe that there is some beneficial effect from high-usage players, but also that it likely varies from team to team (the dreaded contextual effects for us stat guys!) and from player to player.
mystic
January 18, 2011
@Italian Stallion
On the Suns it was Steve Nash who gave Stoudemire better looks and it shows up. Stoudemire right now has 57.1 ts% and a 12.2 turnover rate. During the time he was on the Suns he had 56.8 ts% and a 13.9 turnover rate in games without Nash from 2004/05 to 2009/10. In games with Nash he was 63.3 ts% and 11.1 turnover rate player. Looks like a huge difference here. While Stoudemire is less efficient on offense, Nash kept his level. It is also interesting to see that the Suns with Nash on the court have still a 116.6 ORtg, rather similar to previous years (117.8 in 2008, 116.4 in 2009, 117.5 in 2010). The problems starts when Nash isn’t on the court anymore.
Anyway, I agree with Berri here that Silver’s analysis isn’t good at all. The weird average is something I stumble over a lot of times somehow. He is right that using the exact total numbers would be the best way to do. basketball-reference.com offers a tool to collect data from different seasons for a specific team together, that would have been a good starting point. Kevin Pelton’s analysis is obviously the best in terms of that. Looking at the available data while those players where on the court together and comparing them when they where not together on the court makes a lot of sense. Unfortunately those numbers aren’t adjusted for the strength of their opponents.
I looked up how Anthony effected the ORtg and DRtg in Denver over the years. With him on the court the Nuggets were a 110.1 ORtg team, that is +3.7 in comparison to the Nuggets without him. On defense the Nuggets got worse by 2.3 points per 100 possessions. It is still a Net gain of +1.4 per 100 possessions. Well, using the pythagorean expectation we get that the Nuggets with Anthony on the court were a 0.607% winning team while without him a 0.556% winning. That doesn’t look like an elite improvement, but nontheless it is an improvement. We also should take into account that those numbers weren’t adjusted for the strength of the opponents. But for the sake of comparison: During the same timespan the Mavericks with Nowitzki on the court were a 113.9 to 105.8 team per 100 possession (or 0.771 win%) while they decreased to a 103.5 to 106.4 (or 0.338 win%) without him. That is a huge difference and more the kind of difference you can expect from an elite player.
So, well, the overall conclusion is pretty much the same here, Anthony isn’t an elite player in terms of impacting the game.
EvanZ
January 18, 2011
I think Pelton’s article is important, because it points out the effect of limiting teammate’s turnovers, the second most important of the “four factors”.
mystic
January 18, 2011
@EvanZ
Right, that is a very important point. If we look at the data from 1973/74 to 2009/10 we get a linear correlation coefficient for the relationship between ORtg and ts% of 0.869, for turnover ratio it is -0.571. Putting both things together (I call that off% = ts%*(1-(to-r/100))) we get a correlation coefficient of 0.944. That basically means that ts% and turnover ratio can explain the offensive efficiency of a team. On the other side the offensive rebounding rate of a team has a 0.081 correlation coefficient to the ORtg during that time span. There isn’t a big effect by offensive rebounding regarding the offensive efficiency (at least not a high correlation).
Italian Stallion
January 18, 2011
mystic,
Thanks for the data on Amare with the Suns.
I was referring to Amare’s potential positive impact on some of the other Knicks. His usage rate has risen (hence turnovers and assists have risen) and he often gets doubled teamed and game planned against. A couple of Knicks have stated they are getting better looks than they used to because of that. So it’s possible that’s why there own TS% has risen.
Italian Stallion
January 18, 2011
DSMok1,
“Personally, I currently believe that there is some beneficial effect from high-usage players, but also that it likely varies from team to team (the dreaded contextual effects for us stat guys!) and from player to player.”
This is the position I have taken from a non stats point of view.
IMO the impact usage is going to have (if any) is probably going to be dependent on makeup of the team and whether the player(s) are currently being used to the full extent of their ability or already stretched to their limit.
mystic
January 18, 2011
Italian Stallion,
that sounds reasonable. Stoudemire getting more attention on the offensive end than before should open up some space for his teammates. Not quite sure that this is especially true for Felton, because he had a higher eFG% last season than in this season. And nobody can actually tell me that Stoudemire helped him with his free throws (he increased his ft% a lot). Chandler has indeed a higher eFG%, thus it sounds very reasonable that he is getting more open looks. Overall the offense of the Knicks gets way better with Stoudemire on the court. They go from a 102.6 ORtg to 112.4. it seems evident that Stoudemire has a huge positive impact on the offense of the Knicks. Well, we don’t want to look at the other side of the court, because Stoudemire has also a huge negative impact on defense. Overall there is still a Net gain of +3.0 per 100 possessions. Similar to Anthony in the end. Both are not really elite impact players.
Italian Stallion
January 18, 2011
evanz & mystic,
Are you saying that higher usage leads to higher turnovers and some of those turnovers are taken away from the player’s teammates because their usage naturally declines?
Therefore, if a high usage scorer gets little or no extra credit for his extra scoring and penalized for his higher turnovers, that would tend to underrate his contribution.
EvanZ
January 18, 2011
@mystic,
Interesting. At least to a first approximation, those figures look similar to an analysis I did of the four factors recently (and compared to DeanO’s):
Factor DeanO Mine
Shooting 40% 54%
Turnovers 25% 22%
Rebounding 20% 15%
Foul Rate 15% 10%
Italian Stallion
January 18, 2011
mystic,
Thanks again.
Felton’s recent play is a big topic of discussion among Knicks fans right now. He started off the season shooting much better than in the past (some credit given to Amare) but has been dreadful lately. The question is whether it’s a short term blip, reversion to the mean, nagging injuries, shot selection etc…
DSMok1
January 18, 2011
One other note:
You said “Utilizing the same data set employed to study coaching [i.e. data on players from 1977-78 to 2007-08], I looked at the factors that explained a player’s TS%. The explanatory factors I considered included past TS%, age, game played (to capture injury), etc…. In addition, I considered a dummy variable, equal to one if a player was in his first year as Carmelo Anthony’s teammate. If the estimated coefficient for this dummy variable is statistically significant (and positive), then we can conclude that Silver is on to something. When the model was estimated, though, the Melo dummy variable was clearly insignificant. In sum, it doesn’t appear that a player’s TS% — when we consider a number of factors that impact player performance – is impacted by joining at team with Carmelo Anthony.”
Would you mind reporting the coefficients and standard errors on this regression? It sounds like a well-posed regression; I’d like to see all of the results to it.
EvanZ
January 18, 2011
IS, I think the main point is that a “good” ball handler relieves his teammates of those duties, which means fewer turnovers for the team as a whole.
Here are the USG% and TOV% of the 5 Nuggets with the most minutes this season:
Player USG% TOV%
Afflalo 14.2% 10.5%
Anthony 31.8% 11.4%
Billups 21.6% 17.5%
Nene 18.3% 14.3%
Lawson 20.0% 14.7%
The league average TOV% is about 13.1%. Anthony has a very low TOV%, second only to Afflalo. If Anthony’s usage went down to say 20%, Billups, Nene, and Lawson would have to handle the ball more. This would likely result in 1-2 more turnovers per game. Doesn’t sound like much, but remember that each turnover is worth about 1 point on average, which is about 2.5 wins over the course of a season.
ilikeflowers
January 18, 2011
Abbott has a nice post up about this kerfuffle. His post makes me wonder how easy it would be for Melo to become an elite producer. It would seem that all he needs to do is spend less time at three point line and grab some more boards (each of which should reinforce the other). I don’t think that anyone questions his offensive skillset, but highly skilled + poor strategic use of said skills = good but not great contributor to wins. Unless you sign Melo with very specific aims in mind (using perception to lure in more quality players, increasing attendance – only temporarily if you’re not winning, or changing his playing style) then it’s almost a recklessly risky signing unless you’ve done enough quality analysis to conclude that The Carmelo Effect is likely both significant and real in this specific case. Even if you have a good plan, it’s still a really risky max contract to absorb unless you have the aforementioned solid evidence to the contrary.
Daniel
January 18, 2011
The other thing to note here is that Silver only talks about 16 players. Carmelo Anthony has had exactly 62 teammates in his career. With such a small population, you HAVE TO USE THE ENTIRE DATA SET or your data is meaningless.
marparker
January 18, 2011
Finally someone brings up high usage and low turnover as opposed to high usage and difficult shots.
Melo has an offensive rating of 107 which is about league average over his career. So to some up Melo he is a guy who gets his teammates into better offensive opportunities while offering only average opportunities from himself. As opposed to a truly elite player who by definition offers great opportunities for himself while getting lesser though better than average opportunies for his teammates.
A chart showing team offensive rating vs. player offensive rating would have sufficed for this excercise.
mystic
January 18, 2011
Italian Stallion,
higher usage by a player who has a lower turnover ratio than his average teammates should decrease the overall amount of turnovers. Thus the offense of the team gets more efficient. Pretty simple stuff.
We can check that and I will again use Anthony and Nowitzki for that. While Anthony was on the court since 2003/04 the Nuggets have an average of 0.5 less turnovers per 48 minutes. It obviously decreased the amount of overall turnovers. Checking those numbers for Nowitzki we get a decrease of 2.3 per 48 minutes.
Another factor would be more open looks for the teammates. During the time with Anthony the Nuggets had a 1.0 higher eFG% (went from 49 to 50). For the Mavericks the increase is again higher 47.1 to 51.2.
Nothing really surprising here. Nowitzki increases the eFG% and lowers the amount of turnovers whenever he is on the court. Nowitzki had a slightly less eFg% for that timespan (50.9 eFG%), but Anthony had a way lower eFg% with 47.6. Somehow he gets so much defensive attention that his teammates get more open looks and can convert their shots more efficient.
That is the impact of a player beyond his boxscore numbers.
@ilikeflowers
You can’t just say that Anthony should operate less at the 3pt line, because being out there also provides the necessary spacing for the offense. If he goes more inside, someone else should be more on the perimeter. And it is obviously that the Nuggets for many years didn’t have a stretch 4 who can do that, but their PF and C were inside players. Thus it had to be Anthony who went more outside. Now the Nuggets have a stretch 4 with Harrington and Anthony is actually taking more shots closer to the basket. He is below his career average in 3Pa per 36 minutes. And the Nuggets right now are playing faster than in average during Anthony’s career. He and Karl know that he is more efficient while working closer to the basket, but that doesn’t make the whole team offense more efficient, if there is not enough room to operate in that area. A phenomena they are facing when the opponents defense just collapse (which happened a couple of times in this season already).
Italian Stallion
January 18, 2011
mystic,
I understand what you are saying about turnovers, but I believe I am bringing up another related issue.
Let’s say we have a “simple model” of a team of perfectly equal ball handlers, with equal usage, equal turnovers, and changes of usage don’t impact the scoring efficiency of any of them.
If one increases his usage, the usage of his teammates must fall as a group.
If that happens, the player that increased his usage will have more turnovers and his teammates will have fewer by an equal amount. The team total will be the same.
So in this simply model, the higher usage player will bare the burden of higher turnovers in the boxscore. In order for him to be considered an equal player to before, he must be getting enough credit for his higher scoring usage to compensate him for the extra turnovers. If the model does not compensate him enough for higher usage scoring, he’s going to look like an inferior player and his teammates like superior ones even though the team results will be identical.
I hope that is clear.
I am not talking about the positive impact of increasing the usage of a BETTER ball handler, just the fact that players that increase their usage will tend to accumulate more turnovers.
ilikeflowers
January 18, 2011
mystic,
I don’t think that Melo’s game is going to decline if he shoots fewer threes. As for team strategy, that’s why you have teammates. If your scenario is really a significant issue for a particular team then that’s a team construction issue and has very little bearing on maximizing Melo’s contributions to winning.
mystic
January 18, 2011
Italian Stallion,
For a linear model the player should benefit also by getting more points, assists or whatsoever. In the same fashion the numbers should decline for his teammates. Overall the player gets already that compensation, no need to increase that further.
If a player gets more touches and his turnover ratio is the same, he has more overall turnover. That should be clear. But the important thing in terms of efficiency is his turnover ratio, not the real amount of turnovers after all.
ilikeflowers
January 18, 2011
mystic,
Disregard my previous post, I’m in a hurry and it didn’t come out well. Let me restate what I was trying to get at. What you describe is primarily a team construction (strategic) issue and secondarily a marginal value issue. I believe that Anthony can increase his personal productivity by shooting fewer threes and grabbing more rebounds. Whether or not his team is properly constructed has little to do with him and his general value to a typical team. I’m sure that all players are more or less valuable to specific teams based on a host of other issues. The significance of that variance may or may not be high relative to wp48 (I’m inclined to think that is low).
mystic
January 18, 2011
ilikeflowers,
It is easy to say that this is a “team construction issue”, because indeed it is. But most times the teams have to deal with the players they have and they can’t get better players who also fit their system (because those players aren’t available for the right price). The coach has to use the players on the roster and get them to play at the best level together. He can’t just focus on maximizing one players contribution. That is not how that works.
We saw a couple of times teams are overpaying to get such players. The Mavericks paid Erick Dampier a lot of money, because they needed a defensive minded center, the Orlando Magic paid Rashard Lewis a lot, because they needed his size and his outside shooting. That is part of the strategy to get such players. The Nuggets gave Al Harrington a 3 yr MLE contract (with some “insurance” for two years) to fill a team need. And sometimes a team has to wait for awhile to get the needed player. That’s how it works.
Nima Ghamsari
January 19, 2011
Someone with a good stats background please help me understand this concept. It sounds like Berri is saying that for each player the sample size is too small, so let’s just throw it all out.
But why doesn’t he combine all of the smaller sample sizes into a larger scoring-efficiency-over-historical for Carmelo’s teammates and see if that’s a large enough sample size to be statistically significant? It seems silly to just disregard such a huge amount of overall data because it’s broken into subgroups that in themselves are too small to be analyzed.
Eyal
January 19, 2011
Stoudemire’s Knicks career so far has 3 issues that seem to indicate a lack in WOW.
1. When they were 3:8 he came out very strongly, saying a number of times that they have to change the culture in NY to a winning one, and he sounded very confident, and secure in his ability to do just that. Shortly after they started winning. It’s called leadership, and maybe, just maybe it worked.
2. 4th quarter scoring. I know WOW assumes every point (reb, assist, TO, etc) is equal but I’m pretty sure that’s debatable.
3. Stoudemire’s WP48 is calculated for a center while he’s actually a natural power forward. I’m curious how does WOW account for that. In other words, how does WOW take into account coaceas, like D’Antoni, playing people out of position. (Fields is a similar story, btw., because his great rebounding would be a tad less impressive had he played SF.)
DSMok1
January 19, 2011
@Nima
Interesting observation–and I believe that is a relevant point here.
ilikeflowers
January 19, 2011
mystic,
No one will disagree with anything in your last post, but what does it have to do with trying to measure a player’s individual productivity range and predicting how said player’s productivity can be maximized (of use when evaluating a potential signing from another team, when trying to determine the best way to utilize your current players, and when trying to determine what additional players you need to add)?
Philip
January 19, 2011
dberri,
I enjoyed this article. Thanks for letting me contribute.
Italian Stallion
January 20, 2011
mystic,
>For a linear model the player should benefit also by getting more points, assists or whatsoever. In the same fashion the numbers should decline for his teammates. Overall the player gets already that compensation, no need to increase that further.<
"IF" the model values scoring properly. That is a massive "IF".
Mike G
January 25, 2011
This was a good rebuttal.