Friday, August 10, 2007

Economists, Engineers and the Effects of Steroids

Economists and engineers have one thing in common--they know everything; just ask any economist or engineer and he will heartily confirm that, never mind a few faulty projections, collapsing bridges or exploding space shuttles. Generally, economists and engineers are far too busy to comment on sports but now some economists have ventured over into the playground and decided to set the rest of us straight about what really is happening in the world of fun and games. For instance, apparently we were too dumb to realize it but Barry Bonds has not really gotten any bigger, performance-enhancing drugs do not actually work and Bonds' late career performance should not raise any suspicions. Yes, that was a mouthful and no, I am not making it up--someone actually wrote all of that.

The first claim is based on a statement by Mark Silva, who is erroneously referred to by Wages of Wins as a "doctor" (Silva is actually an orthotist). Silva designs and builds the custom-made arm guard that Bonds famously wears over his elbow. Silva says that he uses calipers to measure Bonds' arms and that "there’s been no significant change in the size of his arms" in the 15 years that Silva has measured the slugger. That is interesting, to say the least, since Bonds' listed weight has increased 53 pounds during that time, from 185 to 238, and some people believe that Bonds is in fact even heavier than 238. I'm no medical doctor--but then neither is Silva or the Wages of Wins author--but I will go out on a limb and say that it is physiologically impossible to gain 53 pounds without your arms increasing in size; this is even more true if the person in question is an elite athlete who quite obviously gained much of this new mass in his upper body. Maybe "significant" means one thing to Silva and something else to the rest of us. Maybe the arm guard is adjustable, maybe there is some other explanation but I'm not buying the idea that Bonds' arms are the same size that they were 15 years ago--but this kind of thinking fits in perfectly with the very credo that Wages of Wins espouses in all of its analysis: don't believe your eyes, because your eyes lie; only go by the numbers (though in this case Wages of Wins chooses to ignore the fairly obvious conclusion that one would draw about the arms of someone who has gained over 50 pounds). Their writers will earnestly say that they can crunch a few numbers and achieve a better understanding of sports than general managers and coaches who make their living in the field. I'm all for using statistical analysis as a tool to better understand sports (and anything else)--but nothing is better than the trained eye and nothing hurts the cause of statistical analysis more than the attitude of some of its adherents that they know everything there is to know about everything.

It's only one step from Mr. Silva's remarks to this quote, made by a Wages of Wins reader and cited in the aforementioned Wages of Wins article: "...please don’t show me a picture of the 1986 Bonds compared to today. We all looked at leaner than when we were 21. Bonds began bulking up well before he faced any steroid suspicions. Most players do." The wording of the quote is a bit garbled but as I understand it the writer is asserting that people tend to look leaner at 21 than they do when they get older. I'm not sure how scientific that is, how we would go about proving that or what exactly it has to do with Bonds and his usage of performance-enhancing drugs. The writer then cites Silva's arm measurements of Bonds and concludes, "Even if his head and feet have grown, who cares! Anyone who has bothered to look at the scientific literature knows that HGH has no performance-enhancing effects. This is the consensus opinion of the exercise physiology profession."

So there you have it--Bonds has not actually gotten bigger, except for possibly his head and feet, and performance-enhancing drugs don't work, so what's the big deal? But wait--there's more. This author also wants us to believe that, save for Bonds' 73 home run explosion in 2001, his "aging curve is actually worse than Aaron’s in many ways (as far as the ability to hit home runs is concerned)." This statement is demonstrably untrue. Bonds has hit a total of 73 home runs (and counting) after the age of 40, a major league record; Aaron had 42 home runs after the age of 40. Bonds also has far more most home runs after the age of 35 than any other member of the 500 Home Run Club, with his 312 (and counting) placing him well ahead of Aaron's 245. Bonds has hit more than 40% of his career home runs after the age of 35. The next closest member of the 500 Home Run Club in that category is Rafael Palmeiro (37%), who wagged his finger at Congress while denying that he cheated only to be disgraced when it was revealed that he tested positive for steroids.

Let's take a deep breath for a moment; sometimes it is hard to know how to respond to something that is so obviously incorrect from beginning to end. One thing I know going in is that there is no way to convince this writer that he is wrong because he "knows" without question that he is 100% correct. So the issue is how to discuss these subjects intelligently so that other people are not led astray by his confident certainty. Let's start with some facts about Bonds' size. Bonds was listed at 6-1, 185 in 1986 when he was a 21 year old rookie. He was listed at those exact same numbers until 1991, when his weight changed to 190. In 1992 his weight went back to 185. From 1993-96 his weight was listed at 190. So from ages 21-31 Bonds gained exactly five pounds. In 1997, Bonds was listed at 6-2, 206. I don't know how he grew an inch but that is a story for another day. He remained at 206 in 1998 and in 1999-2000 his weight increased to 210. In 2001, his weight jumped to 228, where it remained until this season, when he is now listed at 238. It should be noted that, if anything, the post 1999 numbers likely understate Bonds' size. In any case, Bonds stayed essentially the same size for more than a decade and then transformed himself fairly quickly from a lean, fast player to a huge, power-hitting machine. This change is not at all similar to how people's bodies gradually evolve from age 21 to 40+; Bonds' body actually stayed about the same size for more than half of his career.

Mark Fainaru-Wada and Lance Williams, the authors of Game of Shadows, based their account of Bonds' use of steroids "on more than a thousand pages of documents and interviews with more than 200 people, many of whom we spoke to repeatedly. In our reporting on the BALCO story for the San Francisco Chronicle, we obtained transcripts of the secret grand jury testimony of Barry Bonds and seven other prominent professional athletes." One is free to disregard all of this information, just like one was free to not believe the Dowd Report that linked Pete Rose to gambling--but make no mistake that you are then choosing to disregard a lot of evidence. Their research indicates that Bonds began using steroids prior to the 1999 season. These drugs not only helped to transform his body but they helped to transform all of his numbers, not just his home run totals. Four of Bonds' five best offensive seasons came after 1999--in other words, after the age of 35. Fainaru-Wada and Williams go on to note that baseball researcher Lee Sinins, using the "runs created" formula devised by Bill James, has listed the best offensive seasons in baseball history. Bonds has compiled three of these seasons, more than any player other than Babe Ruth--and all of them came well after his body changed noticeably and dramatically. Bonds was 36, 37 and 39 during those seasons; no other player on Sinins' list was older than 33. Another baseball researcher, Sean Forman of baseball-reference.com, says that from 2000-2004 Bonds had the greatest stretch of five consecutive years that any player has had in major league baseball history. Again, keep in mind that Bonds was 35 at the start of that run. Both Sinins and Forman are looking at overall offensive performance. Bonds' added strength and bat speed not only translated into more home runs but raised his complete hitting game, altering how pitchers could pitch to him and enabling him to do things that no one else in baseball history has done.

Clearly, Bonds is a special athlete. He was a great player long before 1999, but cheating has enabled him to vault from being a great player to being a transcendent player and has irrevocably tampered with baseball's record book, not only in home runs but also in slugging percentage, walks and many other categories--unless, of course, you choose to believe that he has not gotten bigger or that his increase in size is not the result of using performance-enhancing drugs.

If you still believe that Bonds has not gotten bigger, then you are an excellent candidate for the Flat Earth Society or one of those groups that believes that the moon landings were staged. For the rest of us who realize that Bonds has in fact gotten much bigger, the question is how much of that growth is attributable to performance-enhancing drugs. It is absurd to say that such drugs don't work; if that is true, then why are athletes in almost every conceivable sport using them? Do economists know more about physiology than elite athletes and the doctors/chemists who are working for them? Steroids and other performance-enhancing drugs are just that--performance-enhancers; they enable users to work out longer and harder, with shorter recovery times. It should be obvious that this represents a tremendous advantage. Anyone who makes it to the major leagues is already an elite athlete. An elite athlete who uses artificial means to enable himself to train longer and harder will end up having advantages in strength and explosiveness, which directly translate into power and bat speed. Steroids will not help the average person to hit home runs but an elite athlete who already is gifted with great hand-eye coordination receives a great boost by using such substances. That is why so many athletes are cheating. Major league baseball has finally put a steroids testing program in place, albeit one that the World Anti-Doping Agency (WADA) considers to be weak, but there is no reliable test for Human Growth Hormone (HGH). In other words, just like the cheaters were ahead of the game in the 1990s and early 2000s it is likely that they are still ahead of the game now.

20 comments:

Reggie said...

Great article! Thanks for writing it!

Reggie

Anonymous said...

I linked to this from your latest WOW post. The post you are citing on the WOW was entitled "More on Bonds - Lifting Comments From Sabernomics.com." As the title indicates, almost all of the post is lifted directly from a different site. That could not be more clear.

The basic premise of your piece is that the WOW folks said a lot of things about Barry Bonds. In fact, they simply noted approvingly what others had said.

For instance:

"The first claim is based on a statement by Mark Silva, who is erroneously referred to by Wages of Wins as a "doctor" "

In fact, the WOW did not erroneously refer to Silva as a doctor. As the post notes very clearly, those are the words of J.C. Bradbury, author of The Baseball Economist, who said that in the comment thread of his blog, sabernomics.com.

Later you say:

"It's only one step from Mr. Silva's remarks to this quote, made by a Wages of Wins reader and cited in the aforementioned Wages of Wins article: "...please don’t show me a picture of the 1986 Bonds compared to today..."

Again, the article was on Sabernomics.com. The commentor was JC Bradbury. It's true he is a WOW reader, and in fact has posted there. However, you seem to be going to great lengths to find the WOW guilty of a possible crime committed by someone else.

The fact that you got all this wrong doesn't necessarily invalidate your point vis a vis economists and their love of numbers although I disagree with you on that score. However it is, all in all, imho sloppy journalism.

If I were you I would consider taking it down or rewriting it completely. Every post you make is a calling card for people looking to hire you as a freelance writer. Accurately attributing qoutes and collecting facts are the bricks and mortars of good journalism.I don't think this piece reflects well on your professional talents, which from other things I have read seem strong.

I am a WOW fan, and I don't agree with the argument you are making here, or in your latest wow post, but this comment is certainly made in the spirit of constructive criticism and I hope you take it that way.

David Friedman said...

Thank you for sharing your concerns, although I don't understand why you are not confident enough in your views to attach your own name to them. Nevertheless, let's examine your claims.

I agree with you that attributing quotes and facts correctly is important but you are wrong that I have failed in this regard. The WoW author describes the Sabernomics article as doing "a wonderful job correcting some of the stories that have been told about Barry Bonds." At the end of his post, the WoW author concludes, "It is unfortunate that more members of the media fail to note the observations offered by Bradbury and his readers." That means that WoW endorses all of the errors that I cited in my post; in fact, he not only endorses these errors but he considers it "unfortunate" that more people do not think this way. Your critiques about sloppy sourcing are more properly directed at WoW for offering such effusive praise to such poorly thought out material.

I quoted directly from the WoW post, when I alluded to Bradbury's comments I made it clear that he is a WoW reader and I provided a link to the WoW post so that my readers can check out the whole thing for themselves. If I "approvingly" post a link to something that is inaccurate then readers would be correct to question why I have put such content on my website and that is exactly the criticism that I am directing toward WoW.

An analogy would be if a New York Times article includes a quote that is inaccurate (like referring to Silva as a doctor) or if it cites some dubious conclusions (such as asserting that PEDs do not enhance performance). The NYT is responsible for everything that appears in its pages, whether an NYT writer wrote it himself or quoted somebody else. Likewise, WoW is responsible for, as you put it, "approvingly" linking to an article that makes several highly questionable assertions about steroids and also inaccurately refers to Silva's professional status.

I find it interesting that you mentioned more than once that you disagree with me but you spent all of your time critiquing my method of citing the WoW post but you did not indicate the nature of your disagreement with my fundamental points, namely that PEDs work and that there is overwhelming, if circumstantial, evidence, that Bonds took them and that his size and performance increased after he allegedly began doing so.

Anonymous said...

Look, my problem is that you have turned a possibly very valid criticism of something sports economist and author J.C. Bradbury, said into a diatribe against another sports economist, David Berri. Thats why this entire post reads very strangely to me. I mean if you are going to attack Berri, is the fact he thought something Bradbury wrote was interesting where you want to start?

IF you wanted to write a post about how wrong people are to say that Bonds performance might not have that much do with steroids, why not simply critique this piece.

http://www.ajc.com/opinion/content/opinion/stories/2007/07/30/bondsed_0731.html

What really does Berri have to do with this? Is the fact that he likes possibly flawed scholarship more newsworthy than the flawed scholarship itself? You are just setting up a strawman, and attributing quotes very sloppily in order to make it work.

Also, in your most recent post, which was actually what I was meaning to comment on before I got sidetracked, you say you take Berri to task here in this post, "for asserting, among other things, that performance-enhancing drugs do not in fact enhance performance."

Berri simply didn't say that or even imply that. What he said was, "People have argued that the surge in home run production we see from Bonds as he aged could only have been achieved via steroids. Apparently, though, Aaron also saw his home run output surge in the latter years of his career." In his first post on Bonds, he says "Bradbury goes on to site the role of league expansion, changing stadium dimensions, changes in strike zone enforcement, and even alterations to both the ball and bats being played. In other words, there could be other factors beyond steroids that could explain what is going on with Bonds."

Where exactly does he say that performance enhancing drugs don't work? And how do you go from a comment Bradbury made about HGH, to Berri having made a blanket statement about PED's.

And also, what are your thoughts on the scientific literature regarding the performance enhancing effects of HGH. It it actually a PED?

I don't raally know what the answer is re Bonds. Having read Bradbury's piece I am convinced that there were other factors in play. But I agree with you also that PED might have more to do with it.

Courteously,
Owen

David Friedman said...

Let's take this from the top. My post about Bonds was not a "diatribe against" David Berri. In fact, Berri is not mentioned even once. I read the WoW post about Bonds/steroids and noticed that it reflected something that I have previously observed about the modus operandi of many (but of course not all) economists and engineers, namely that they think that they know everything about everything. My post refuted four aspects of the WoW post. The first aspect is that Silva's credentials were wrongly represented. The second aspect is that Bonds has not gotten any bigger because, allegedly, the pad that he wears around his elbow has been the same size for many years. The third aspect is that PEDs allegedly do not work. The fourth aspect is that Bonds' late career increases in production should not raise any suspicions.

Your primary concern seems to be that you think that I have somehow damaged WoW's credibility by mentioning WoW in connection with Bradbury's writing--"blaming" WoW, to use your word, for what Bradbury said. The problem with your reasoning is that WoW, by your own admission, "approvingly" cited Bradbury's take on these subjects.

Among the things that the WoW post in question "approvingly" cites is this quote: "Anyone who has bothered to look at the scientific literature knows that HGH has no performance-enhancing effects. This is the consensus opinion of the exercise physiology profession." That is why in my post I take WoW to task for asserting that PEDs do not enhance performance. Also, although HGH is one of the substances that Bonds allegedly took, he also allegedly took several other PEDs, so even if HGH is not a performance enhancer then Bonds still may have received a boost from the other illegal substances that he took. Since every major sports organization that I know of views HGH as a performance enhancer I'd be interested to know why the Bradbury and the economists at WoW think otherwise. The burden of proof is in their corner.

It is certainly possible that a multitude of factors are in play to some degree regarding the increased home run production that started in the late 90s--however, the preponderance of evidence, as documented in Game of Shadows, is that Bonds took PEDs and that his usage of those substances altered his body and increased his production. Marion Jones never flunked a drug test but has now admitted that a substantial proportion of her achievements were fraudulent, so just because Bonds has never flunked a test we do not know that he was clean. In fact, many of the things that he allegedly took are undetectable and/or can be used in a cycle that masks detection.

I think that the WoW writer should have been much more thorough in checking out what is known about Bonds and PEDs before "approvingly" citing Bradbury's article.

I only mentioned Berri by name in my 20 Second Timeout post about his article about Kobe Bryant. That article asserts that Bryant is not the best basketball player in the NBA. WoW seems to like to attack what it views as conventional wisdom and that is fine--I like to do that sometimes, too--but only if you have the goods to back it up. As I pointed out in my 20 Second Timeout post, the same system that Berri uses to conclude that Kobe is not the best player in the NBA also concludes that Dennis Rodman was more valuable to the 1996 Chicago Bulls than Michael Jordan. That does not pass the "laugh" test. Furthermore, casual readers may not realize that Berri's methodology regarding basketball analysis has been seriously questioned by other statisticians and economists (like Dan Rosenbaum).

It is my opinion that the sloppiness that pervades the WoW posts about Bonds/PEDs and Kobe Bryant is indicative of some flaws in WoW's overall approach--namely, the WoW writers assume that conventional wisdom is completely wrong and that they know everything about everything. From what I've seen and heard, Berri's responses to criticisms by Rosenbaum and others are extremely condescending--and, at times, deceitful, such as when Berri denied what he had written in his book regarding Jordan and Rodman. I agree with Rosenbaum that WoW has attempted to do some interesting things but the attitude and arrogance of the WoW writers is a serious drawback.

Anonymous said...

My primary concern is that you continue referring to a post by JC Bradbury, on his own blog Sabernomics, as if it had written by the WOW people. JC Bradbury is a sports economist, but he is not one of the authors of the WOW. He has posted there, but 99% of the posts at the WOWJ are by David Berri. This in my mind is misattribution. I mean the name JC Bradbury doesn't even appear in your piece! Isn't that sort of amazing.

Your response to this is that Berri citing this post approvingly is basically like him saying it himself. I find this argument curious. It really doesn't make any sense to me. Why no just keep things simple, and say you disagree with Bradbury's post?

It is true is that Bradbury did post at the WOWJ an article about HGH. I understand you have a problem with the case he makes in this post. That's is what you should be trying to refute.

http://dberri.wordpress.com/2007/04/24/rumors-experts-and-human-growth-hormone/

Regarding Berri and the WOW, I have to admit I sort of frustrated with your post. It's basically the same kind of thing I have seen in a 100 places.

It's factually incorrect that Berri said Rodman was better than Jordan in his book. Don't believe me, read what Matthew Yglesias had to say on this question.

"Since I'm partially responsible for the confusion, I should say that now having read the whole book, it's true -- The Wages of Wins doesn't argue that Dennis Rodman was better than Michael Jordan. It does argue that Rodman was an underrated player and that, generally speaking, the value of scoring is overstated and the value of rebounding understating. They conclude, however, that the CW on Jordan is more-or-less right."

What you contend is true, and what has basically become the CW about the WOW in many pasts of the blogosphere, is in fact completely wrong. Berri thinks, and always has thought, that the statistics show Jordan was the better player. That's because he was much higher above the mean for his position than Rodman was for his. He has been entirely consistent on this point, though I sdmit consistently misunderstood.

Given that, reading your Kobe post, it's just sort of baffling. You say, "If you are signing on for Bryant not being the best player based on WoW's methods then you are also signing on for Rodman being more valuable than Jordan." This is just incorrect, plain and simple.

(Another thing that bothers me. Why do you put the word "revelation" in qoutes in your post? The suggestion is that Berri used that word in his post. Did he?)

I have probably written 100 posts in various places on how good Kobe is. I honestly know his statistics inside and out. You put less stock in statistics than I do, which I understand and accept. There is more to the game than the numbers. But the statistics are very clear. Kobe has no statistical advantage even over other elite shooting guards, let alone everyone else in the league. Last year, he was worse per minute than Manu. Before his injury. Wade was performing at a much higher level also. It's not even close.

You will say that the numbers don't matter. which is fine, and that will be the end of it That's almost always how it goes. But at the end of the day, its really hard to look at something like this and say that Kobe is indisputably the best player in the league.

http://www.basketball-reference.com/fc/pcm.cgi?req=1&cum=0&p1=ginobma01&y1=2007&p2=bryanko01&y2=2007

What really does Kobe have on Manu, other than the ability to play a lot of minutes? Manu is a winner you know. Pretty clutch in the playoffs. Led Argentina to the Gold medal. European titles. etc.

Re Rosenbaum, I just wish you held him to the same standard you hold the WOW folk to. I think the slides he presented at a recent Harvard conference and posted on APBR might be something interesting for you to look at. I would love to see a post on those.

I would also note that while there may be sports economist who disagree with the WOW folk, there are many more who value his work. He has published work on the topic in peer reviewd academic journals. He has gotten an academic publisher to print his book. And he has had other economists use his metric. Wolters used WP to investigate the impact of race on officiating. Berri also organizes the biggest sports economics conference n America, something he has been involved with for years. So I don't know that its exactly true that a lot of sports economists doubt the validity of his work.

I don't think Berri is condescending. But I am partial. The fact that he responds to nearly every polite question in his comments has swayed me a great deal on this point.

Losing the thread here, I hate posting in these little text boxes. That is enough for now...

David Friedman said...

I really do not understand your confusion or concern regarding the format of this post. You spent a great deal of time in your most recent comment indicating how respected a figure Mr. Berri is; that makes it even worse that he devoted space on his website to Bradbury's questionable arguments. My post spent more time dealing with why those arguments are wrong than with assigning "blame" (to use your word). If I did assign "blame" it was directed in a general sense to a style of reasoning/expression that I have frequently observed among economists and engineers. My focus was much more on clarifying the situation regarding Bonds and PEDs than engaging in a chicken/egg search to determine if the erroneous viewpoints originated at WoW or with Bradbury. Readers can make their own judgments about the fact that WoW clearly and unhesitatingly accepts these viewpoints.

I clearly referred to Bradbury in my post, calling him a WoW reader, when I quoted directly from his comment about pictures of Bonds and Bonds' hat size. I am much less interested than you appear to be in assigning "blame" to a specific person (though I do "blame" the style of thinking and the condescension that I detect in the WoW posts). My post focused on what is wrong with the arguments cited in the WoW post. Since I included a link directly to the WoW post, anyone who does not clearly understand exactly who wrote what here is, quite frankly, being intentionally dense, so I am puzzled that you accuse me of incorrectly attributing anything.

In the book, Berri plainly says that by his metric Rodman was more effective than MJ on a per minute basis. MJ played more minutes and produced more wins according to Berri but I can assure you that any GM or coach would find the idea of Rodman being better/more effective than MJ to be quite laughable.

The use of "revelation" is clearly meant as sarcasm and not as a literal quote. I think that you are not giving my readers enough credit; they can understand the difference.

In the comparison between Kobe and Manu that you cite, Kobe leads in PER 26.1 to 24.1. That is a difference of almost 8%. How big of an advantage is it reasonable to expect the best player in the game to have over one of the 15 or 20 best? In track and field, the difference might be a fraction of a second. Furthermore, Kobe produced that PER in almost 41 mpg, while Manu produced his in less than 28 mpg. What kind of production would Kobe be capable of if he only had to play 28 mpg? Could Manu sustain a 24.1 PER over 41 mpg? These things are pretty obvious to anyone who watches basketball with a serious eye, like the coaches, scouts and players who I regularly interview, most of whom say that Kobe is the best player in the NBA. How many games do Berri and his cohorts watch? Basketball players are not robots and they are not just economic numbers that can be studied in abstract. I very much believe in the value of statistics--but as a tool, not as a god.

I don't hold Berri to a higher or lower standard than anyone else. I found two articles on his site to be incorrect--the Bonds one and the Bryant one--and I wrote a post about each. To suggest that I can't write about WoW unless I write about every single other thing that may be incorrect holds me to an impossible standard. Your argument is analogous to something someone said to me after I wrote an article in which several Hall of Famers told me that Roger Brown should be in the Hall of Fame. This person said to me that there are more worthy candidates who I should have written about first. Whether or not that is the case, what happened was I researched Roger Brown, interviewed many people and reported what they told me. Must I keep that article in my files until I write about every other more worthy candidate? Who determines which candidates are most worthy? I am not aware of what Rosenbaum said/wrote that you find so objectionable but I will look over the information that you provided when I get a chance--but the fact that Rosenbaum may be incorrect about one thing does not invalidate his critique of Berri's basketball analysis.

I corresponded with Berri briefly a while ago (prior to these posts) and I had no problem with how he interacted with me--but as an APBR member I am aware that his responses to Rosenbaum have not been so warm. I have never met Rosenbaum and do not know him personally but I have read a lot of his APBR posts and seen how he interacts with a variety of people, so I find it difficult to believe that he started whatever trouble exists or existed with Berri. In any case, there is a difference to responding to neutral or positive comments at his own website and the writing style that he uses in the posts, in the book and in his dealings with critics like Rosenbaum. Rosenbaum also has lamented that Berri has declined to participate in APBR metrics and discuss his work with the members of that forum. In short, I don't have a personal beef with Berri but I disagree with his basketball analysis, with WoW's take on Bonds/steroids and the general methodology that I observe many economists/engineers apply to certain issues, particularly issues outside of their expertise.

I do appreciate your comments and thank you for taking the time to share them.

Anonymous said...

On this engineers post, I think I have said my piece on the fact I find it very confusing.

I agree there is a big divide between adherents of observation vs. statistical analysis. That is really, a very very large topic, one that I don't want to get bogged down in debating. You and I hold different views on the subject, and I think they are unlikely to change.

Re Rodman and Jordan, you said above in your comment.

"In the book, Berri plainly says that by his metric Rodman was more effective than MJ on a per minute basis."

Berri addressed this misunderstanding at great length in his post, On Jordan and Rodman Again. The link cuts off so I won't paste it but here is what he says:

"...it really doesn’t matter much if Kaufman is taking sides or is actually a genuine Switzerland in this debate. What might matter is whether or not I know what we said in The Wages of Wins.

This is the quote Kaufman takes from page 144 of our book. “Per 48 minutes played, Rodman’s productivity even eclipsed Jordan. Rodman’s WP48 of .0.415 was four times the production offered by an average player in the NBA, and even surpassed the 0.386 WP48 posted by Jordan.”

If Kaufman were to read the very next line he would see: Of course when one looks at standard deviations about the average, Jordan was still more productive than Rodman."

Later in the post he says:

"In other words, although in terms of Wins Produced or WP48 Robinson or Rodman might eclipse Jordan in a given year, we still find MJ to be the best when we consider the supply of talent each played faced at his position."

I think about it in baseball terms. A first baseman who hits 40 home runs is more productive than a shortstop who hits 39. But he is far less valuable, since there are many power hitting first baseman, and very shortstops capable of that kind of production.

Now, look at what you wrote.

"David Berri's initial response was to deny ever making such an assertion; that is kind of like Rafael Palmeiro wagging his finger defiantly and then flunking a drug test (maybe Palmeiro should have just said that WoW denies that performance-enhancing drugs enhance performance). As Salon.com's King Kaufman pointed out, "alas and alackaday for Berri, when you write something in a book it stays writ and the statement that Rodman was more valuable on a per-minute basis than Jordan--that is, a better player who just didn't play as many minutes--was on Page 144" of the book The Wages of Wins. The exact quote on that page is "Per 48 minutes played, Rodman's productivity even eclipsed Jordan."

I mean you compare Berri to Palmeiro. That's ridiculous. And you basically trot out the same argument Kaufman made nearly a year ago, without any recognition of the fact that Berri has responded in full.

Basically, when I read your post I wasn't impressed. At the end of the day, I am a WOW fan. But I am genuinely curious to see new posts on the topic and to hear fresh thinking on it. I don't think your post qualifies as that.

And what Rosenbaum posted at APBR was extremely disappointing to me. I was looking forward to him presenting a strong challenge to Berri and bringing something fresh to the table. It did not work out that way.

Alright, enough from me...

David Friedman said...

Berri wrote in his book that Rodman's per minute productivity eclipsed Jordan's. If a torrent of criticism informing him that this is absurd led him to subsequently try to explain his way out of or around that statement, that's fine, but to deny that he wrote this in the first place is like Charles Barkley saying that he was misquoted in his autobiography. The larger issue is that Berri's system overvalues rebounding, which is how this erroneous ranking of Rodman comes about in the first place. As Rosenbaum and others have pointed out, there is a lot of subjectivity in the WoW basketball rankings, meaning that players can be moved up or down based on certain adjustments (this is fully explained in the material that I linked to in my 20 Second Timeout post, so I am not going to get into all of it here).

When you admit that you are a "fan" of WoW that pretty much says it all. This is not about being a "fan" of a statistical analysis system. This is about trying to analyze basketball in the most accurate fashion possible. Being a "fan" of WoW is no different than being a Steve Nash "fan" or a "fan" of any other player who makes comments from a biased perspective. At least you openly acknowledge your bias.

Anonymous said...

This is the qoute from the book.

“Per 48 minutes played, Rodman’s productivity even eclipsed Jordan. Rodman’s WP48 of .0.415 was four times the production offered by an average player in the NBA, and even surpassed the 0.386 WP48 posted by Jordan. Of course when one looks at standard deviations about the average, Jordan was still more productive than Rodman."

So Kaufman was clearly in fact qouting Berri out of context, and you are following his lead. I mean don't you think it a bit strange to qoute a paragraph but leave out the final sentence. It's like telling another guys joke, leaving out the punchline, and then talking about how he just isn't that funny.

I am a WOW fan now, but I was on the fence for a long time. And it was that kind of stuff from his critics that helped sway me in part.

I have read Rosenbaum's critique, and parts of it are reasonable and he is civil there. He has been extremely intemperate at other times though. He has attacked Berri very harshly at APBR. He also anonymously posted some very vituperative comments on NBA Babble right after it started up. I helped with that site, and I have to say his conduct did not win my admiration, and was also kind of bizarre to boot.

Still, I was hoping and expecting some very interesting work from him in the wake of the Harvard conference, and he definitely did not deliver.

Re Rodman, he was an excellent player and he should be in the Hall of Fame. Every team he played significant minutes for won a lot of games, bar one. He was the best rebounder of all time and one of the best defenders of all time. And he was on the two best teams of all time for which he averaged more than 6.5 offensive rebounds per 40. To me that sounds like a great player, two dimensional, not better than jordan, but certainly great.

David Friedman said...

Berri flatly wrote in his book, as I quoted, that Rodman was more valuable to the 1996 Bulls on a per minute basis than Jordan. If Berri meant to say something else then the passage in the book should have been written clearer. You have alluded to Berri's point allegedly being misunderstood. It seems more accurate to say that his point has been laughingly dismissed to such a degree that Berri subsequently chose to distance himself from his own conclusion, which was also displayed in the form of a graph in the book. All of this business about Jordan being more valuable because of the position he played and first basemen versus shortstops is just a sideshow. Berri's system overvalues rebounding and that is why it determined that on a per minute basis Rodman was more valuable than Jordan in 1996. Maybe rebounding is undervalued by conventional wisdom but Berri's correction seems to be more than a bit extreme.

I don't know anything about NBA Babble or what Rosenbaum presented at Harvard so I cannot comment on either of those things.

I agree that Rodman was an outstanding player worthy of HoF consideration. Chamberlain and Russell were the two greatest rebounders of all-time but Rodman is right there on the short list after them with guys like Jerry Lucas and Moses Malone. Rodman is the greatest mid-size (6-6) rebounder of all-time. I never even remotely suggested that Rodman is not a great player; however, he was not as valuable--on a per minute basis or any other--as Jordan. This is the kind of thing that makes me ask if Berri and company have actually watched any of these guys play or talked to any knowledgeable people (players, coaches, scouts) who watched them play. Maybe a person can design great economic metrics from some inner sanctum without ever dealing directly with the businesses and customers who make the economy go but you cannot hope to really understand basketball without actually watching it. One year, Rodman missed several games due to suspensions and during that stretch Jordan averaged over 10 rpg. Could Berri or anyone else have predicted prior to that that Jordan could do such a thing? A player's value consists not just of his actual statistical contributions but many other factors. Jordan and Bryant attract a lot of defensive attention that makes it easier for other players to get open shots. Berri argued that Bryant's teammates are not as bad as some people say but he gives no indication that he understands that a good part of their production is made possible by Bryant's presence (and, as I noted before, Bryant played well over 40 mpg, so don't tell me that they were getting most of their production when he was not in the game). If Jordan missed several games could Rodman have suddenly averaged 30-35 ppg? If the Bulls had not gotten Rodman they could have signed a good/above average power forward and the Jordan/Pippen duo still would have led the team to championships. They would not have won 72 and 69 games like they did with Rodman but if they had a guy who got 8-10 rpg they would have still been the best team. Even in 1995 with a rusty Jordan and no power forward (Pip played power forward in the playoffs) the Bulls still gave Orlando a battle, losing one game at the last seconds when Jordan, fresh off the baseball diamond, uncharacteristically turned the ball over.

David Friedman said...

I'd like to go back to one of your earlier comments. You asked what does Kobe really do better than other top shooting guards and cited Ginobili as your test case. I responded by pointing out that Bryant is clearly better than your own hand picked example by a significant percentage in PER (nearly 8%) and that on top of that Bryant maintains this greater productivity while playing significantly more minutes.

Your response to this was:

"I agree there is a big divide between adherents of observation vs. statistical analysis. That is really, a very very large topic, one that I don't want to get bogged down in debating. You and I hold different views on the subject, and I think they are unlikely to change."

I am not an "adherent" of anything other using all of the available information to make the best possible evaluation of players/teams. If you have read my earlier posts/articles about Bryant then you know that I have explained in great and specific detail exactly why I--and many other informed NBA observers--consider him to be the best active NBA player. To say that you don't want to get "bogged down in debating" is simply a cop-out that can be interpreted in the following ways: (1) You don't believe that I am capable of understanding the logic behind your reasoning vis a vis Bryant/Ginobili; (2) you have no effective rebuttal to the points that I made in response to your comparison of the two players. There is nothing to get "bogged down" about; my response was very direct and I am open to hearing a reasoned rebuttal. Why do you say that our views are unlikely to change? Does that mean that you are 100% certain that you are right and/or that you believe that I am irrational and therefore would not be open minded enough to consider other evidence?

Reasonable arguments could be made in favor of several players being dubbed the best player in the NBA. It is my opinion--and the opinion of many other expert NBA observers--that the best and most persuasive arguments are in favor of Bryant.

Anonymous said...

He didn't flatly say it. And he didnt change his mind afterwards. It's all there in the book. Nothing has changed. What should have changed is your understanding of his argument, but that doesn't seem to have progressed much since early 2007, or whenever the Kaufman piece came out. Really if you read the book you should know that at no point in time did Berri assert Rodman was better than Jordan. There is a whole section of the book dedicated to showing that Jordan was probably the best player of all time! How could he think that?

On the observation v stats, question, I think you misunderstood me. I was speaking generally about the general divide between statthink and intuition. You find these kinds of debates in many, probably most fields. What is the best method of medical diagnosis? Play poker by feel or by the numbers? Fundamental buy and hold or trading? In many places there is a divide between people who feel you should stick to the numbers and those who think you do better if you look beyond them and trust your gut or your experience. I think in some fields stats work well, in others intuition is more important. It's complicated, and a debate that has been going on full force since Aristotle. That's what I was referrring to re too much to debate in a comment thread.

Re Manu and Kobe, I am happy to discuss, here is what I can say. Manu last year was better than Kobe per minute. He has a significant 3% edge in TS%. He is a better rebounder. He was a better defender. Kobe is a higher scorer, but Manu is also a high scorer and is more importantly a better scorer. This isn't wins produced, this is just looking at the raw stats. It's very clear. And it translates in the +/- data. If you look over at 82games, Manu has a large advantage as well. His net +/- was 9.3, Kobe was at 5.9. And that is despite the fact Manu is working against a much tougher comparison than Kobe.

IMHO, Kobe's greatest strength is his ability to play 40 minutes per game. That is the only reason you might prefer him to Manu, who is injury prone. Although Manu is outstandingly productive when he is in there, the Spurs have to find someone to play twenty minutes at sg every night, while the Lakers have only 8 min's to fill. The Spurs of course have an extra seven million to spend, so that evens the score a bit, but it might be reasonable to choose Kobe for that reason. That said, I think it makes a lot more sense to have Manu and 7 million dollars to spend on a quality player like Barry. You are hedged against injuries for one.

But for one game in which you needed 30 minutes, certainly I would choose Manu.

Kobe is the biggest star in the NBA, not the best player. He plays in a big media market. He has won championships, although Shaq deserves more credit for those. He is extremely charismatic. He is a great showman. He is very memorable. These qualities and attributes explain in great part why he is considered the best player in the league even though it's difficult to discern that superiority statistically.

If you can show me any stat that suggests Kobe is better I am happy to discuss it, but other than PER, which I think even Hollinger lately has suggested needs some tweaking, you won't find any. Manu really is much more productive per minute, and Kobe really does play more minutes. That's basically the numbers story.

Also, I understand I am not going to change your mind about Kobe. But hopefully you can understand my perspective here. When one player is a more efficient scorer, a better rebounder, a better defender, and has a demonstrable advantage in the +/- numbers, how can I think the other guy is better than he is?

David Friedman said...

I have read the book and your account mischaracterizes what Berri says regarding Jordan (which is ironic, considering that is what you are accusing me of doing).

First, Berri does say that Rodman's per 48 minute production exceeds Jordan's (which is also the same claim that you make about Manu versus Kobe--I'll get to that later). Berri argues that because it is harder to accumulate credit in his system as a shooting guard than a forward (because his system overvalues rebounding--that is my comment, not his) that Jordan was more standard deviations ahead of players at his position than Rodman was ahead of players at his position. None of this changes the simple fact that Berri's system says that Rodman was more effective in the 1996 season on a per minute basis than Jordan was. I watched that team pretty closely and I do not agree that Rodman was more effective, productive and/or crucial to the team's success than Jordan was. I doubt that anyone who understands basketball would agree with Berri's assessment.

Second, there is a section in which Berri asserts that Kevin Garnett is the best player of the previous 10 years (1995-96-2004-05), though he hastens to add, "we are not married to our criteria or our answer." A few pages later, there is a section titled "The Jordan Legend" in which Berri compares Jordan to various players, including Wilt Chamberlain and Magic Johnson. At one point (page 141) he writes, "Does this mean Jordan was the greatest player ever? Once again we have two hands and we can't make such a definitive statement." A few sentences later, he says, "So one could make an argument that Magic is the best ever." Then a few sentences after that, he adds, "In other words, we believe Wilt was very good, and maybe the best that ever played the game." All of these quotes occur within two pages of each other, so if you had posted a comment saying that Berri called Jordan, Johnson or Chamberlain the best ever I suppose that you would be right--and you would be wrong. For something that is supposed to be an exact--or reasonably exact--science there sure is a lot of wiggle room here. Any reasonably competent fan could tell you that Jordan, Johnson or Chamberlain might be the best player ever. Certainly, a subjective case could be made for each one without having to do any kind of sophisticated calculations.

Berri says one thing on one page and then offers another interpretation on the next page. Are we supposed to take all of it seriously or none of it? I find the idea that Rodman was more productive on a per minute basis than Jordan in 1996 to be an assertion that does not pass the proverbial "laugh test." The fact that Berri later says that by a different metric Jordan was more productive and that he may be the best player--unless Magic is or maybe Wilt--does not change the fact that his system says that Rodman was more productive in 1996 on a per minute basis than Jordan. So, I am comfortable with what I quoted and the way that I presented it. If someone wants to take Berri's word for it that Kobe is not the best player in the game then he should also be comfortable with the idea that Rodman was more productive on a per minute basis in 1996 than Jordan was. That's all I said in the post on this issue, nothing more and nothing less; the reader can look at this and make his own judgment.

I have explained a few times in various 20 Second Timeout posts/comments which criteria lead me to say say that Kobe is the best player in the game today. I consider 10 categories that a scout would look at when evaluating a player. Here is how I analyze Kobe's performance in each one (these are not necessarily listed in order of importance):

1) Finishes at the hoop with either hand
2) Dribbles well with either hand
3) Has excellent post moves and footwork
4) Draws fouls and shoots FTs very well
5) Has three point range
6) Can get off a good shot attempt even against good defense
7) Rebounds well for his position
8) Reads double-teams well and makes the correct passes, which don't always lead to assists for two reasons: the second pass out of the trap often leads to the assist and it is not possible for anyone to get an assist if the shot is not made
9) Excellent defender, as acknowledged by the league's head coaches in All-Defensive Team voting
10) Tremendous inner drive and will to win

From a scout's perspective, Kobe has no weaknesses. Most of the other top players are deficient, relatively speaking, in at least one area (Duncan's free throw shooting, LeBron and Nash's defense, etc.). There are 4-5 players for whom one could make a case as the "best" player, so this is a somewhat subjective exercise. I understand that, but I believe that the best case can be made for Kobe. If you talk to players, coaches and scouts around the NBA--which I am fortunate enough to be able to do--many, if not most, of them say that Kobe is the best player. Some would argue that he is not the MVP because the MVP should play on a team that wins at least 50 games but that is a whole other discussion.

So, while a statistical case can be made on Kobe's behalf (see below), my viewpoint is formed by watching and evaluating players the way that a scout does. This is a subject that greatly interests me and I did a two part article called "The Scout's Eye View of the Game" that described my experiences talking with scout Kevin Mackey about how he does his job.

Now that I have explained my perspective, let's take a closer look at what the numbers say. According to the NBA's official EFF stat, Kobe was the second best player in the NBA last year with a 27.65 rating (Garnett had a 29.17 rating). Ginobili ranked 56th (17.32). On a per minute basis, of course, Ginobili moves up and Kobe drops but Kobe is still ahead of him.

As you already noted, Hollinger has Kobe at 26.13 (2nd in the league) and Manu at 24.18 (3rd in the league); he has Wade first at 29.04 but Wade only played 51 games, which is not enough to qualify for league leadership in most categories, so I would not include Wade.

Roland Beech of 82Games.com produces several ratings: on court, off court, net and Roland Rating, which adjusts the net rating to account for what the other nine players on the court produce. Beech attaches this disclaimer to the Roland Rating: "These ratings represent a player's value to a particular team and are not intended to be an accurate gauge of the ability and talent of the player away from the specific team." So, the Roland Rating is not really meant to determine who is the best player in the league; it gives an indication of how valuable a player was within the context of his own team but it is not at all certain that the player would be equally valuable in a different situation. Nowitzki ranked first in Roland Rating last year, Manu was fifth and Kobe was sixth.

Each of these three rating systems has different weaknesses that their creators freely acknowledge (and this comment is lengthy enough without me describing them). Still, it is telling that Kobe was second in two of the systems (first in one if you properly discount Wade, who barely played half a season) and sixth in the other. None of this has anything to do with his charisma or the market he plays in or anything else. Many fans--and even some alleged experts--declare that all Kobe does is shoot a lot and score a lot but if that were true then he would not rank so highly in all three of these systems (as an aside, it is interesting that two-time MVP Steve Nash, who finished second in last year's voting, does not rank in the top ten in any of these three systems, despite the fact that he is often praised for his "efficiency," especially in contrast to Kobe's alleged inefficiency).

Per minute stats have some usefulness but they are not very meaningful when trying to assess who is the best player in the NBA. The best player in the NBA must, by necessity, be someone who can sustain a high level of play for 38-40 mpg; otherwise, he is a part time player. You blithely dismiss the significant difference in Kobe's minutes versus Manu's but there is no reason to believe that Manu can sustain his level of play for an entire season while playing 40 mpg--and there is plenty of reason to believe that Kobe could perform at an even higher level in fewer minutes. I also reject the assertion that Manu is better defensively than Kobe; Kobe is bigger, stronger and just as quick, if not quicker, than Manu. He is capable of guarding a wider range of players than Manu.

If Manu and Kobe switched places, the Lakers would be much worse because Manu could not fill Kobe's role as a 40 mpg, 30 ppg player. Meanwhile, the Spurs would be better, because Kobe would step right in as a high level 40 mpg player--and I would definitely prefer an extra 8 mpg of Kobe over an extra 8 mpg of Brent Barry.

Anonymous said...

I thought I posted a response to this. I don't see it there. I was having a problem with your password system, so I guess it didn't go through...

Have to say, I don't like your list. A majority of the categories are scoring related. Why don't you simply say, he must have a ts% above 56% and a score more than 25 points per 40 to be considered a good scorer? The idea that you would put passing out of a double team above moving without the ball seems ridiculous to me. Your omission of ability to play without committing turnovers seems incredible also. What about ability to play in transition? What about offensive rebounding? Surely a much more important offensive skill than some of the ones on your list? What about the ability to defend the pick and roll? What about basketball IQ?

I do think your list is indicative of the way scouts think though, they basically are focused on scoring. The general rule in the NBA is that points per game is the only important statistic. You see it in aalary decisions also, and in playing time. And that's a big reason why I agree with Berri that NBA decision making is fundamentally flawed. How smart are these guys when if their decision making is overwhelmingly driven by one statistic?

Re Jordan, Chamberlain, and Magic, he points out that the full box score didn't exist when chamberlain played, so he can't calculate WP. That whole section is heavily qualifed. He outlines the assumptions that go into his conclusion very clearly. And he speaks very circumspectly, as economists often do. It's interesting, sometimes you find him condescending, but here you find him wishywashy.

Re the EFF stat. It's absolute garbage. It's not even worth talking about. I don't think much of PER either, although it's slightly better. They both have the same flaw. A player can increase his rating by taking more shots, even if he does it a low efficiency.

And according to PER Rodman was a below average player for his career, which tells you something about the weakness of that approach.

Finally, Berri does not say Rodman is better than Jordan. And the definition of productivity you are using is yours, not Berri's. Rodman was more productive in terms of raw statistics. Rodman averaged 15 rebounds per game in 32 minute that year. But in terms of how good they were relative to the average for their postion, Jordan was much better.

It's very clear throughout what he means,but I sense I am not going to change your mind on this issue either...

David Friedman said...

If you sent a previous comment it never made it through to the page where I moderate comments.

I should have realized that to fully appreciate my skeletal "scouting report" one must actually watch basketball and understand what the terminology means. For instance, when I say that Kobe rebounds well for his position this incorporates offensive rebounding, which is not a primary responsibility for a shooting guard who operates on the perimeter and who often is expected to get back on defense when a shot goes up as opposed to crashing the offensive boards.

Most great players commit a high amount of turnovers simply because of how much they handle the ball. Look in the NBA Record Book and you will see that turnover records are dominated by all-time greats. More important than turnover stats is what kind of judgment a player exercises and how productive he is overall. Nash is renowned for his efficiency but he turns the ball over more frequently than Kobe. Yes, Nash generates more assists but Kobe scores more points; they both handle the ball a lot and are very productive.

Defending the pick and roll is incorporated in #9. Perhaps I could have been more specific about Kobe's defensive contributions but I tried to keep this list to a manageable size while at the same time giving an idea of how scouts evaluate players.

Moving well without the ball would be a worthwhile addition to the list. Kobe is excellent in this category as well; he knows the ins and outs of the Triangle and makes good off the ball cuts (he could also function in other offenses, of course).

Scoring is not even close to the only thing that I consider when I evaluate players; my list included passing, defense, rebounding and heart/desire. I gave some specific examples of offensive skills that scouts look at as opposed to just saying that Bryant scores a lot of points. Many NBA players score a lot but do not have the well rounded offensive game that Bryant has in terms of footwork, ballhandling skills and shooting range.

Berri is "wishy washy" in the language that he chooses in the particular section of the book that we are discussing but yes, I do think that Berri is arrogant--and wrong--to crunch some numbers and conclude that NBA decision making is "fundamentally flawed," as you put it. He knows more about player evaluation than NBA GMs, coaches and scouts? This is exactly what I meant when I said that economists and engineers think that they know everything about everything. If an NBA GM walked into an economics conference and claimed to know more about the subject matter than economists would anyone take him seriously? NBA player evaluation is a lot more sophisticated than Berri assumes and he is wrong to correlate salary with value. Salaries are affected by a lot of factors. Sometimes teams knowingly overpay to avoid losing a player; the way that the financial business of the league is conducted is impacted by the salary cap rules and other factors. You also have to remember that players sign long term deals that escalate, so their salary in a given season may not correspond to their relative "value" that year. I don't think that anyone who knows what he is talking about would say that the top ten players are the 10 highest paid players. Jordan was not the highest paid player in the NBA until late in his career. Do you think that GMs did not realize that he was the best player? Obviously, it would be ideal if every player was paid exactly what he is worth but our society does not work like that or else movie stars would not make a lot more than teachers, police officers, etc. Behind the facade of a bunch of numbers Berri is basically making a Joe Six-Pack argument: "These stupid GMs don't know what they are doing. Why do they pay Player X so much money?" Yeah, sometimes GMs make mistakes, as do managers in any business but Berri does not give NBA GMs nearly enough credit. At APBR Metrics, Rosenbaum discusses at length why Berri is wrong to take this approach.

Last year, NBA EFF ranked Kobe second and Iverson--a frequent target of criticism by Berri--27th so the stat obviously considers something other than just scoring. As I've said, though, neither NBA EFF nor any other single stat is the reason that I say that Kobe is the best player in the NBA.

There is something wrong if you are correct that the plain language in Berri's book does not mean what it says. Berri clearly wrote that Rodman was more productive on a per minute basis than Jordan. Those were his words, not mine, so you are right that you will not be able to convince me that he did not say this. Nor will you be able to convince me that this statement is accurate. Now your defense of Berri seems to be that Berri used very cautious language that could be interpreted to mean a lot of different things. He also seems less "circumspect" at his website and in his reactions to his critics.

Anonymous said...

What do you think Berri is saying in the last sentence of the paragraph?

"Of course when one looks at standard deviations about the average, Jordan was still more productive than Rodman."

I just don't understand why we can't hash this out.

On Gm's and salary in the NBA, there is overwhelming evidence on this subject. Every researcher, not just Berri has concluded that what drives pay is points per game. I don't think that makes sense. Neither does Berri. Nor do most people in the basketball stat community.

The most important factor at the team level is shooting efficiency. Yet when GM's hire players, the pay much more attention to scoring volume. Yhis doesn't compute.

You can defend them if you like, but the data is extremely clear, as anyone, Hollinger, Berri, Oliver, etc will tell you.

As for Rosenbaum, it's unsrprising that you like him as your stat guru. In the stats v human judgement debate, like you he favors the latter, which is fine. Personally, I have a little trouble separating what he says from the fact that he works in the NBA (although he is no longer being paid). It would be difficult for him to say anything else.

And actually, it would be totally impossible to do your job the way you do, with a lot of interaction iwth NBA staff, if you believed the things I do. No value judgment there, but certainly a reason I know I am unlikely to make much headway here.

On your list five categories are directly related to scoring, and dribbling is a fairly big part of scoring as well. I am not going to make too much of the list, since I understand you were just sketching thing out. I mean passing out of the double team is in there, and general passing ability isn't. I get your idea. Passing is important, and you have to think about it situationally, not just in in the aggregate...

But in general, I think most of what you need to know, for a player who has been playing in the league for 3 years, is what is his TS%, how much does he score, and how many turnovers does he commit. This, offensive rebounding, and assists tells you pretty much everythign you need to know about a players offensive ability.

Kobe was not anywhere near as good as Nash last year. Nash is a much more efficient player. His TS% was almost tops in the league, 7% better than Kobe's. His assists more than compensate for his turnovers. And he makes his teammated better. Put Kobe on the Suns and they are worse as a team.

And finally, about Berri, it may be he was less than polite when all this started, but having read every post on the site, I can say he has been a model of circumspection for a long time now...

David Friedman said...

In the last sentence of the paragraph, Berri is saying that Jordan is more productive than Rodman because Jordan is further ahead of his fellow shooting guards in productivity than Rodman is ahead of his fellow power forwards. Berri also said somewhere in the book that Jordan contributed more wins because he played more minutes. None of this changes Berri's assertion, in very plain English, that Rodman was more productive--Berri's exact word--than Jordan on a per minute basis. I also don't understand why we can't "hash this out." In my opinion, the reason is that I am taking what Berri wrote at face value, while you are trying to interpret it in some way.

I'm not sure who constitutes "most people in the basketball stat community." If you go to the APBR Metrics site, there are a lot of people there who dismiss Berri's work because Berri has not published his numbers in an academic journal, because he is either unresponsive to or haughtily dismissive of their questions/critiques and because the WoW methods have very poor predictive value even compared to the other stat systems that you dismissed in your previous comment.

Did the Spurs sign Bowen years ago for his "scoring volume"? Is that why the Nuggets signed Camby? Why do the Cavs value Eric Snow even though fans are dismissive of his contributions? Obviously, not every general manager does a good job and even the good ones make mistakes. It is not realistic to expect player evaluation to be any more flawless than anything else in the business world. Berri's categorical statements to the effect that NBA personnel people, as a group, don't know what they are doing and that he--Berri--does not need to watch any games to better understand how to build a team are both absurd contentions. In fact, those statements are so bizarre that they should make anyone cast doubt on his work even before reading his book or looking at his numbers. Think about someone making analogous, categorical statements about anything else: all the experts are wrong and I can do better than they do without even obtaining first hand information about the subject at hand. Then you wonder why I might say that economists (and engineers, but that's a separate story) think that they know it all.

Rosenbaum uses plenty of numbers--he just dismisses (or at least disputes the value of) Berri's WoW work. Since Berri is apparently one step (and a small one at that) below God for you, it is unsurprising that you dismiss Rosenbaum's work without apparently being too familiar with it.

Why would it be impossible for me to "do what I do" and think a different way? Well, I guess the fact that I actually have more direct, first hand knowledge of the league gives me a different perspective from Berri and you but somehow I doubt that is what you meant. Go to 20 Second Timeout and read my work; I hardly hesitate to criticize players, coaches or executives who I think are doing a poor job--but I do so fairly and after considering all of the relevant information.

You are right that my list is a general, sketched out one. I could easily add another 10 or more categories to make it more specific. My overall point is that scouts evaluate a player's skill set in many specific categories; I just gave a sampling of them. Scouts also look at a player's body type and athleticism but that is less of an issue here because we are talking about players who are not only already in the league but are in fact among the league's elite players; body type and athleticism are more important factors when one tries to figure out if a high school, college or international player can make it in the NBA.

No stat system that I've seen--I'm not counting WoW here--says that Nash is more "efficient" than Kobe. Nash is a more accurate shooter, in no small part due to the fact that he can be much more selective with his shot attempts because he has better teammates and is not expected to score as much. Nash commits more turnovers and more turnovers per 48 minutes than Kobe--and if you want to talk about a truly bogus stat then bring up assist/turnover ratio. There is not an either/or relationship between the two; those numbers have just been cobbled together for some reason. Granted, a very poor ratio there might indicate a problem, but it should be obvious that someone whose primary role is playmaking is going to have a better ast/to ratio than a scorer because all players are going to have at least 1-2 turnovers a game just based on the occasional offensive foul, dribbling violation or getting the ball stolen. A playmaker picks up his assists in volume; Kobe is a good passer for a shooting guard but his role is not to get 9-10 apg.

The stats that you cite to determine an offensive player's ability completely ignore how he actually plays--that is, what is his role on his team, can he get off his own shot and so many of the things on my list that you so casually dismiss. Would you say that Steve Kerr is a better offensive player than Kobe based on shooting percentages and turnovers? Your point of view is not new to me; I know other numbers crunchers who feel the same way. None of them understand that it is important to watch a player perform to see how he assembles his stat line. Based on your way of thinking, you must have looked at J.J. Redick's college numbers and assumed that he will be a great pro.

Put Nash on the Lakers and they will be much worse, because the Lakers have too many players who can't catch and who don't move without the ball (and who also play poor defense). If you watched the Lakers last year then you saw Kobe pass out of double teams only to have guys fumble the ball and/or miss shots. Then you saw Jackson tell Kobe to stop passing and start shooting, after which Kobe averaged more ppg after the All-Star break than anyone has in the past four decades, carrying the floundering Lakers into the playoffs.

Put Kobe on the Suns and they would actually have a chance to beat the Spurs. They would be more explosive offensively and they would be better defensively.

Anonymous said...

Ok, I will just punt on Rodman-Jordan. You have heard my perspective

Berri has published a LOT of work in academic journals. He is publishing all the details of Wins Produced in an academic paper in January. He is a tenured economics professor. He has been the lead organizer of one of the biggest sports economics conferences in the world for the last three years. I don't know who on APBR has published in an academic journal. I do know that Rosenbaum has never done so, and he himself admits he can't "hold Berri's jockstrap" when it comes to academic credentials. If you can understand the presentation Rosenbaum gave, I would love to hear your explanation of what team adjustments he actually was using in his charts. While he suggests he was using the same team adjustment Berri uses in order to predict wins better than Berri with NBA efficiency, this is not the case. in fact he was using a different team adjustment. And that isn't kosher.

Re Bowen-Camby-Snow I don't quite get your line of attack. My argument is that all the evidence leads to one simple conclusion. Raw scoring, i.e. scoring without regard for efficiency, primarily drives salary and playing time decisions in the NBA. This is the finding of every single study of the matter that I know of. There is an enormous "scoring bias," in the aggregate. Players are rewarded for scoring more points, regardless of how many shots they take, rather than actually playing well. Other factors, turnovers, reboundings, etc, have very little impact. So, in the aggregate, NBA decisionmaking must be flawed, since scoring efficiency (both fg% and ft%) are the most important factors in determining wins. The fact that GM's do smart things, like hiring Camby, doesn't change that fact.

I have read your work. I don't think you are a homer at all. That's not what I meant. You certainly are very objective. You call it as you see it. My point was that you work within the conventional wisdom. You are also active in shaping it as an opinion leader, but there is a traditional framework that is your base. You have to write and think for a large audience, which sees the game in a certain way, and that shapes your approach as a journalist .You also have some constraints. You have to be entertaining for instance. Us stat geeks, we can just pontificate, since no one is listening for the most part anyway.

I liked your most recent article. In different bits and pieces, I think you gave Kobe a fair evaluation. You did note he was 13-32 and 18-27. As a stat geek, I would say this. Kobe had a substandard shooting game. His ts% for the game was 51.2. That compares poorly to his average last year and his career average. But he was much better than the rest of the Lakers, who shot 46.2 and clearly missed Lamar Odom in the lineup. In the rest of the box score he contributed solidly. his eight rebounds, four assists, and 4 steals offset his five turnovers nicely. Overall his Winscore for the game was 7 in 42 minutes, so per 48 minutes his WS would have been 8. This compares favorably to the average shooting guard production of 6.1. (that is for the period 94-04 anyway) But it really is not a stellar game. (not that different from what you conclude). If he had performed to his season average last year, i.e. .199 winscore per minute, he would have posted a 8.358 WS in 42 minutes and the Lakers would, all other things being equal, have won the game (somewhat unfair comment I know, but I will throw it out there) So his game, while spectacular and exciting, was not anywhere as good as he is capable of playing. Granted, he was playing a great defensive team in the Rockets.But he didn't knock it out of the park. Mcgrady btw also had a winscore of 7.

Manu, in his first game posted a WS of 10 in 29 minutes. Most people would disagree that he had a better game than Kobe I think, but to me his contribution was clearly superior. You have to throw in the fact they were playing the Blazers of course, but I give the first round to Manu.

Re your list. At twenty items, with no sense of how each is weighted against the other, your list would begin to get a bit unwieldy I think.

There is no question Nash is the more efficient scorer. His TS% was much much higher than Kobe's last year. 65% to 58%. That's like the difference between Kobe and Jamal Crawford. Kobe is far more prolific. And that has a lot of value, considering that he scored last year as an individual more efficiently than almost every team in the NBA. However, he is not more efficient. And Nash was, overall, a much better player last year. 9.5 assists per 48 is a huge advantage. While he scored 12 less points per 48, he also took ten less shots and 6.7 less free throws. And his production relative to other point guards is far more impressive than what Kobe offers relative to other shooting guards. Kidd and Nash were head and shoulders above every other point guard in the NBA last year.

I don't know I would agree assist/turnover ratio is a bogus statistic. Perhaps scoring/turnover would be a better one. You have to discount people's scoring when they commit a lot of turnovers. For instance Eddy Curry. He is an extremely efficient scorer. However, it costs him a lot of turnovers to achieve it. This hurts his production a lot. An extra turnover per game is actually very significant. It's not a throw away. What really is bogus is how little attention people pay to turnovers.

Re Steve Kerr - He was an above average player. However, the numbers show he was not nearly as good as Kobe. I am not even going to bother to demonstrate that.

Let me say here, if I haven't made it perfectly clear already. Kobe Bryant is comfortably in the top 5% of all basketball players in the NBA. However, he is not clearly the best player. IMHO, he clearly is NOT the best player. And that's basically what we are debating.

Re Redick. according to Berri's draft system, which he thinks needs a lot of work still, JJ was the 19th best player in his draft. So, off the numbers, there was no reason to assume he would be a great pro. It's not just scoring efficiency that counts. Renaldo Balkman on the other hand...

I think if you put Nash on the Lakers, they would actually be a fair bit better. It's an impossible hypothetical to debate. As for the opposite case, I will take a crack at it. Let's just say I strongly disagree that Kobe would actually improve them offensively, even if he did make them more "explosive." The Suns had a higher ts% last year as a team than Kobe did as an individual, in the best shooting year of his career. So when he joins the Suns, its going to be very difficult to help them there. He commits fewer turnovers, but not many. This is actually where assist/turnover comes into play. His turnover differential doesn't come close to making up the gap between them in the assist department. There really isn't any question about it imho, Nash is much much better offensively. The Suns WOULD be better defensively and in rebounding with Kobe, but they would be worse overall. That is my view.

Also, the suggestion in "explosive" is that if Kobe catches fire, the Suns can beat the Spurs, which they can't do otherwise. My view, is that unless the Suns suffer an injury to the big three, they are very close actually. The Spurs had only a slight edge in differential last year, although they didn't try to optimize it, while the Suns stars played big minutes. The Spurs have more depth. But they would be barely favored going in to a series against the Suns, as they were last year. And they did win that series, but it was very close thing, and the Suns were hurt badly by the suspension. It very easily could have turned out differently.

Bottom line, in hopes of making this a constructive exchange. I dislike your animus for Berri, and for engineers and quants in general. Unsurprising since that is what I am. I respect your opinions and your style of analysis. I think you say most of what needs to be said, while also being extremely entertaining and readable. I can't deny that what fans want from the game ultimately is not stats and numbers, it's stories and human personalities. And you fill that need much better than I could, or than any stathead could. What I would say, respectfully, is that you and feel almost any person writing about the NBA could benefit from paying closer attention to what a player's TS% is and looking at all of his box score stats relative to his position average.

I love the game of basketball. I was an all-league high school player (rebounder extraordinaire of course, captain and chief statistician also) and I ran my college club team. What i like about basketball is the team aspect, and the need to be a complete player. The monetary incentives in the NBA, the scoring bias I alluded to, are in direct conflict with the spirit of team play. Players need to take more shots to make more money, rather than doing what they do best, and what is best for the teams. Ultimately, I believe that the NBA would be a much "better" league, more fun to watch I guess, if people were compensated according to WP. They would be incentivized to play "the right way." Obviously, I am making a value judgment about what the right way to play is, and am judging the NBA to be far from that standard, but I don't think i am off base.

Anyway, that is where I am coming from, and that is basically my wish for the impact stat-think can have on the NBA. Call me a dreamer...

David Friedman said...

My only point about Berri and academia is that--at least until January--the WoW info has not been published in an academic journal. A number of the folks at APBR Metrics disagree with Berri's premises, methods and conclusions. I've never questioned his status as a professor or his knowledge regarding economics; I am only questioning his competency regarding sports. I didn't see Rosenbaum's presentation, so I don't know what he said on that occasion. He previously outlined at APBR Metrics his objections to Berri's approach and Rosenbaum's comments in that forum seemed tactful, measured and reasonable to me.

My "line of attack"--though I'm not sure I would use that exact phrase--with Bowen, Camby and Snow is to give three examples of players who are neither showy nor offensive powerhouses but who have been amply compensated (and given heavy minutes--at least until recently for Snow) on good teams. There are other examples, also. I don't agree that good GMs overvalue scoring. I think that Berri and you do not understand how the NBA salary structure has come about. The salary cap, the luxury tax and when players become free agents all affect the marketplace. In other words, a player's salary does not merely reflect his absolute, objective value. A team may overpay to prevent another team from obtaining a certain player for a variety of reasons; another player may be "underpaid" because he became a free agent at a time when many teams were over the salary cap and could not afford to sign him. Yes, those are just hypothetical examples but my point is that a lot of factors drive the salary structure and the issue is more complicated than just saying that player x is the highest paid and therefore GMs think that he is the best player.

Furthermore, I'm not sure that it is true that scoring points is not the most important--or at least a very important--thing. So the "bias" you speak of may not in fact be a bias at all. I don't believe that NBA decision making (or decision making anywhere else) is flawless but I also don't think that it is as deeply flawed as Berri and you suggest.

Thank you for clarifying that you did not mean that I am an NBA "homer." However, I don't agree that my opinions fall entirely in the realm of conventional wisdom. I am influenced by what informed people (executives, coaches, players, scouts) tell me but I form my own views and I sometimes disagree with what the experts tell me. For instance, one time Alex English told me that he thought that Chauncey Billups was a good MVP candidate. I disagreed with that but I ran the quote without commentary because it was part of my profile of English and that was not the place to interject my opinion (which had been expressed previously in a different article).

I also don't agree that my audience shapes my viewpoint; just check out the Kobe discussions at 20 Second Timeout :) My audience, or at least the vocal portion of it, often disagrees with me. I guess that it is good that you find my writing entertaining but my primary goal has always been to be informative and analytical.

Using different means it does seem that we reach essentially the same conclusion about Kobe's first game of the season. My larger point in that article was to debunk the notion that he is somehow not "committed." He was not as effective or as efficient as usual but either by watching the game or reading the stats it is apparent that he gave a very committed effort.

I thought that Manu played very well but I did not really try to compare Manu's performance against one team to Kobe's performance against another. If I am going to compare them then I'll wait until they play each other or until they have several more games under their belts this season.

I agree that a 20 item list could get unwieldy. That is why I stopped at 10. My point was to give readers an idea of how scouts evaluate a player without necessarily portraying this in an exhaustive fashion. Scouts look at players meticulously in a variety of very specific areas. It comes down to more than just a player's shooting percentage, turnover rate or any one stat.

Nash is a more efficient shooter than Kobe. I've never denied that. I disagree that Nash is a more efficient scorer. Kobe can score in more ways and against a greater variety of defenses than Nash can; for instance, Kobe can post up, he can elevate over double teams and he can finish in traffic over bigger players by dunking (Nash does have some nice wrong-footed floaters but a dunk is a higher percentage shot and is also more likely to draw a foul). Looking at their free throw percentages, Nash is a better shooter than Kobe but not by the margin that their field goal percentages suggest. Kobe has to take more shots and shots with a higher degree of difficulty than Nash does due to their roles on their teams. Yeah, MJ could shoot a lot and still make more than half of his shots but that is one reason why I don't say that Kobe is better than MJ. Kobe's shooting percentage is very good considering the number of threes that he takes (MJ did not shoot the three that much for most of his career). I definitely do not agree that Nash was a much better player last year (or any other year) than Kobe. Nash has somewhat of an edge as a shooter and a clear edge as a passer (though Kobe is a good passer for a shooting guard and usually leads his team in assists). However, Kobe is a better rebounder and defender and a more dominant presence on the court. Nash helps his teammates but he also needs teammates who move without the ball and can catch and finish. Kobe can dominate in any situation, regardless of teammates or opponents (though his team will not necessarily win).

Team turnovers are important but individual turnovers have to be broken down to be meaningful. For instance, an offensive foul or a dribbling violation is not as costly as a bad pass or a lost ball that can immediately be converted into a score. The other team gets the ball every 24 seconds (or sooner) no matter what and the great players tend to turn the ball over more than the lesser ones because the great players handle the ball a lot. If you watched the Lakers-Houston game, Kobe's turnover when Wells stripped him was more costly than an offensive foul would be. The Wells play led to a fast break layup attempt (and a foul by Kobe); an offensive foul just leads to the other team inbounding the ball and trying to score against a set defense.

Nash's assists would drop if he joined the Lakers because the other Lakers can't catch and can't shoot. Kwame's hands would not improve when Nash passes to him; the perimeter players would not shoot any better. Nash might score more as a Laker but then his shooting percentages would drop as defenses keyed on him (like they key on Kobe now). In Phoenix, teams often let Nash shoot to guard against him passing to Amare or Marion; no one is going to let Nash shoot to prevent him from passing to Kwame. Also, a scout would note a player's physical characteristics. Nash's body would not hold up to the demands of carrying the Lakers (being double-teamed and guarded more physically than he is now). The Suns with Kobe would be almost unstoppable. Kobe's assist average would increase to 6 or more and Barbosa would pick up the rest of the slack. Meanwhile, how would opponents guard Amare, Marion and Kobe? It would look much like the West's win over the East in this year's All-Star Game (Kobe, Amare and Marion looked like a good combo in that game).

The Suns are close, relatively speaking, to the Spurs, but they have not beaten them yet in a playoff series and I don't think that their recent moves have made them better in that regard (losing defender Kurt Thomas, gaining Grant Hill).

Your last comment is interesting, because Free Darko's critique of WoW essentially said the same thing but cast a negative spin on it, saying that WoW is biased against players like Iverson who do not "play the right way" in some people's opinion. I share your dream that players and teams play the right way but I disagree with the premise that "scoring bias" is quite as out of whack as you suggest. It certainly is not out of whack among the league's best teams, so you can safely watch the Conf. Finals and Finals at the very least :)

Thank you for contributing some very interesting food for thought.