Malcolm Gladwell reviews the aforementioned book, The Wages of Wins. It's an interesting piece, though at this point I think all serious basketball fans already know that statheads think Allen Iverson is overrated. If you want to see a truly controversial claim, check out this 1999 paper by one of the book's authors looking at the 1997-1998 MVP race. The leading candidates were Michael Jordan and Karl Malone. They conclude that Malone had a superior season, worth 18.83 wins as opposed to Jordan's 16.44 wins. Fair enough.
They say Jordan actually only had the sixth-best season. Numbers four and five were David Robinson and Tim Duncan respectively. Then it gets a bit wild. Number three on the list is Jayson Williams! And number one on the list . . . better than Jordan, Barkley, Robinson, and even Malone -- Dennis Rodman whose 20.79 wins make him far and away the league's top player in the 97-98 season.
May 22, 2006 | Permalink
link to the gladwell review?
Posted by: right | May 22, 2006 12:15:37 PM
"I think all serious basketball fans already know that statheads think Allen Iverson is overrated."
Let me be blunt: This is not an argument. This is cant. It's silly and it's insulting.
Posted by: Marty Lederman | May 22, 2006 12:35:15 PM
Petey: Ha, ha.
But, seriously, everyone knows that this is what all the stat guys say. It's getting to the point where, in a sense, I think Iverson is becoming underrated though it always depends (of course) on which audience you're talking about. I think some of this "efficiency matters just as much as scoring" business is overstated. It's easier to be efficient when you don't need to carry a team.
I mean, take this Rodman stuff. If you had him taking 20 percent of your team's shots, he'd turn into a horribly inefficient offensive player. Naturally, the thing to do is to not have him do that. But somebody needs to take the shots. One of the virtues of having a scoring monster on your squad is precisely that it lets you get away with giving lots of minutes to someone like Rodman without crippling your offense.
Similarly with Ben Wallace who's only emerged since being in a situation where the rest of his teammates are good enough to play four-on-five offensively.
This review bothers me. In my experience, if you run a simulation and get counter-intuitive results the first thing you should do is consider whether your simulation is flawed. Gladwell doesn't seem to do that. Perhaps that is because the the authors make such a convincing case for their algorithm but he doesn't even mention any of their arguments for their model.
If I'm not mistaken, their algorithm seeks to turn player stats into "wins" by considering the relative value of different stats across the league. If this is so then I can think of a simple test: total the wins of all players on each squad and compare that to the actual team wins.
Another test (more of a thought experiment really) is to take the 2 best players according to their algorithm at each position and consider what their record would be as a squad if they all played together.
Posted by: WillieStyle | May 22, 2006 1:21:07 PM
That's a pretty impressive stunt for Rodman, considering he only averaged 4.7 points per game in 1997-8.
Posted by: Al | May 22, 2006 1:21:08 PM
WillieStyle: I don't know about the book, but the 1999 paper matthew cites does exactly that:
"The idea that Dennis Rodman offers production
that either surpasses or is nearly equivalent to the
illustrious Michael Jordan rests entirely upon the
validity of the model and methods presented in
this investigation. Does the method utilized to
evaluate the players result in an accurate depiction
of the player’s contribution? In an effort to
answer this question, the estimated wins of each
player was summed across each team. This summation
was then compared to the actual wins
each team registered in both the 1997–1998 regular
season and the playoffs. The results are reported
in Table 11.
The results appear to indicate that the methodology
paints an accurate depiction of each player’s
production. Of the 29 teams listed, the
difference between the summed player wins and
the actual team wins is less than one for nine
organizations. For only four teams is the difference
in excess of five victories and not one team
registered an error of ten wins. The average error,
in absolute terms, was 2.6 wins. Such evidence
would suggest the evaluations of Jordan, Malone,
and Rodman are accurate estimations."
Posted by: Al | May 22, 2006 1:26:07 PM
"if you run a simulation and get counter-intuitive results the first thing you should do is consider whether your simulation is flawed. Gladwell doesn't seem to do that. Perhaps that is because the the authors make such a convincing case for their algorithm"
I'm waiting for my hard copy New Yorker to read Gladwell's piece, but just to re-itereate for the new arrivals in the crowd, algorithms and stats are for baseball, not basketball. Stats in basketball are incredibly imperfect measures of value.
Any basketball stat simulation will be fundamentally flawed.
Jayson Williams, indeed.
(This would be the perfect crowd for a carefully constructed joke about Jayson Williams and Dick Cheney going on an outing together...)
There are occasional debates here in Detroit on who is better - Big Ben or Worm. Ben usually wins because of his leadership but Rodman's numbers sure look better. Worm was infamous for alot of reasons and poor free throw shooting was one of them. If Ben consistantly shot at the percentage Worm did the Pistons would be unstopable.
Posted by: LowLife | May 22, 2006 1:35:13 PM
Well, that's what Dean Oliver calls the skill curve, I think; the more possesions a player uses, the lower his efficiency. The trick is that every player has a specific skill curve and some players (Iverson for example) can use a surprisingly high number of possesions and still maintain a relatively high efficiency. And of course, there are players who can only maintain a reasonably high efficiency by using a very low number of possesions. Still, my take is that some ratings (PER, for example) take account of this pretty well. On the book, let's say that Barry is not very well thought of in the basketball statsheads community. They are destroying his book in the Apbrmetrics forum as we speak.
Posted by: Carlos | May 22, 2006 1:42:25 PM
BTW - I like how Bo Outlaw gets on that list of the top-10 players of 1998. Bo Outlaw!
Jayson Williams at least was an All Star in 1998, so his selection among the best players that year is not THAT inexplicable.
Posted by: AlAl | May 22, 2006 1:47:25 PM
Speaking as someone who has gone on the record in this very forum as believing that rodman deserves a place on my starting 5 for earth against the space aliens (along with russell, bird, magic, and michael), i can hardly criticize an analytic construct that emphasizes rodman's value.
nonetheless, petey, although perhaps overstating a tad, is on the money: basketball stats aren't nearly as meaningful as baseball stats.
or, as i never tire of repeating, russell said that early on he realized that the act of shooting occupies about 4 minutes of a basketball game, so he thought he'd emphasize the other 44, and even though some of what goes on in that other 44 shows up in stats, a lot of it doesn't....
Posted by: howard | May 22, 2006 2:09:38 PM
Building a rating that shows how many wins a basketball player produces is extremely difficult, by the way. Even well-built ones like Oliver's are very context dependent. That's why you see Iverson producing in 03-04 a third of the Player Wins that in 00-01. PER and others on the other hand, are more limited but its results are less variable. By the way, I don't agree with Petey that stats are "incredibly" imperfect measures in basketball, but maybe it's a disagreement on what "incredibly" means.
Posted by: Carlos | May 22, 2006 2:12:07 PM
Carlos, link to the discussion?
Petey, having just been to TPM, your 12:17 is brilliant.
on reflection, i do think it's worth pointing out that team stats can be useful in basketball: opposition shooting percentage, for instance, is a good indicator, as are things like offensive rebounding and turnovers forced....
Posted by: howard | May 22, 2006 2:25:02 PM
Al: I don't know about the book, but the 1999 paper matthew cites does exactly that
Well then I'm just flabbergasted.
Posted by: WillieStyle | May 22, 2006 2:30:05 PM
Here's the link
Posted by: Carlos | May 22, 2006 2:42:26 PM
Clearly, when confronted with evidence that Dennis Rodman was more valuable than Michael Jordan, we need a new definition of value.
I'm kind of serious; I'd never say that - well, let's set Rodman aside - Pippen was a better player than Jordan, but I've intimated that I think you could argue that Pippen was more valuable than Jordan to the success of those Bulls teams - which is to say, I think you could argue that Pippen + average shooting guard *might* have been better than Jordan + average small forward. But that's kind of a facile argument, because it assumes that a mature Jordan would've played exactly the same way without Pippen around; teamed with another superlative scorer - one superior to Pippen - a Jordan mature enough to suppress his ego could've focused even more on defense and point-play. In other words, would Pippen or Jordan have had the better career playing alongside Clyde Drexler instead of each other? In an unrealistic world where everything stays the same, I'd take Pippen+Drexler, but in a more realistic and complex model, I'd take Jordan+Drexler. If Rodman's value, or Bruce Bowen's value, is inherently wrapped up in the very specific role they play(ed), and I think it is, then their value isn't something they can take with them in another system that requires them to do more or different things - in which case, how much of the credit ought to go them vs. the coaches/teammates/overall system?
I'm obviously not saying anything new by saying that the relationship between individual and team has yet to be adequately explored by statistical analysis. I don't think it's impossible, but it might require more fuzzy thinking to get right than a stathead would prefer.
Incidentally, does baseball statistics explore these issues at all, or are they so caught up in the "recurring individual matchups" phenomenon that they don't really look at the effect of one's teammates, especially on the offensive end - i.e., why do the Red Sox bat Ortiz before Ramirez and not the other way around, other than amateurish observations about player personalities.
Posted by: Quarterican | May 22, 2006 3:25:43 PM
Sure Pippen was valuable,if he decided to take floor for the final play or if he didn't have a migraine. Rodman was so much better and more important than Scottie.
That's really not a reply to anyone or anything, I just like to point out that Pippen sucks.
Posted by: keatssycamore | May 22, 2006 3:39:11 PM
I'm not sure what's so shocking about this -- defense is undervalued in every major American sport. Remember what a scandal it was when Charles Woodson won the Heisman over Peyton Manning? How often do pitchers or goalies win league MVP awards? For that matter, how often do closers win Cy Youngs?
Has there ever been an NBA MVP who was primarily given the award for his defensive play?
It's actually not hard to get a basketball stats model that has Rodman rated highly... almost any model that uses offensive rebounds as part of offensive value and missed shots as part of offensive opportunities will do so. Rob Neyer did basically the same thing somewhere in the late 90s.
It's a flaw in the model. Rodman was an extreme player at the end of his career -- almost never shot, got tons of boards. Because of that, a straightforward points/opportunities ratio will list him highly.
The flaw here is that the reason why Rodman has few opportunities is not because he's efficient, but because he was basically not part of the offense -- in essence, he's getting credit for being a liability.
Posted by: nrd | May 22, 2006 3:49:13 PM
Aaron, yes: one william felton russell, who won 5 MVPs.
he's the only one i'm aware of....
Posted by: howard | May 22, 2006 3:53:44 PM
He's not as much of an offensive liability as you make it sound because 1) he would screen the hell out of people and 2) if you just left him alone, then he just went and got an offensive board.
Posted by: keatssycamore | May 22, 2006 3:55:49 PM
Hold on. I'm sorry. Did someone cite, as evidence that Pippen sucked - as a person or a basketball player - in comparison to Rodman, the notion that Pippen sometimes quit on his team or didn't grit through adversity?
Did you happen to catch "Lost In The Desert: Dennis Rodman, The Spurs Years"? Highlights included much pouting, open distraction and disdain for coach and teammates during timeouts, frequent and inexplicable removal of shoes, and the perturbation of too-nice-for-his-own-good David Robinson.
Posted by: Quarterican | May 22, 2006 3:58:06 PM
The comments to this entry are closed.