29 May 2006

All I Need is a Metric(le)

Over at Gladwell's, Malcolm's blogging his latest New Yorker book review of "The Wages of Wins." Yet another attempt to apply Moneyball-style analytics to basketball. I'm all for improving the tools GMs use to evaluate talent, if only to keep the Tsakalidis of the world out in the rest of the world. But there's a problem when those stats tell me that Josh Childress is the most underrated player in the league.

His 'fro, yes. No matter how many props it gets, it's always going to be underrated. Much better than Ben Wallace's do. Shane Battier comes in at number 8 on the list. I forgot, were we rating players or unusual hair?

And of course, KG is the best player in the league, hands down. His team can't win, but that's beside the point. His win shares are huge! Bron-Bron gets within a whisker of the ECF before his 22nd bday, Nash turns Boris Badenov into a breakout star and makes it to the WCF, but KG is the stud. I know he's revolutionary, and at 30 is only a bit past his peak performance years, but he's got to learn to win. Do his teammates suck? Yep. But I'm not sure they wouldn't beat the Cavs-sans-King. Handily.

Like everyone else, I'm going to jump on the fact that one of the economists, David J. Berri, wrote a paper in which he contended that the real MVP for the '97-'98 season was Dennis Rodman. I love Rodman as much as the next earth-bound guy who's only skill is boxing out. But seriously...there's a reason no one picks me first and it's the same reason (writ VERY, VERY small) why no one who actually, you know watches basketball thinks Rodman was ever MVP.

Maybe if Malcolm and The Economists came out for a game sometime (or, in Malcolm's case, just got a little sun) they'd develop a different appreciation for it.

3 comments:

Anonymous said...

Richard, you ignorant slut.
I really don't have time to smackdown every stupid thing somebody says on their blog, or smackdown juvenile commenters, but...
The point of quantifying is because too many GMs and people who actually *watch* basketball don't know what the hell they're looking at. And what most people are watching and pay attention to can be ascribed to the Ewing Theory of Economics: Yes, we make a lot of shots (money), but we take a lot of shots (spend a lot of money).

It's no surprise that the top-10 over-rated are volume shooters--high FGA, relatively low FG% or specialists that don't really fill any other stat categories (Hamilton). And it's no surprise that most of these guys would be top 25 in NBA jersey sales too.

Your critique of the underrated list just illustrates your overall basketball ignorance--as the childish, superficial fixation on hairdos and the lack of any substantive argument belie your true standing as a dilettante.
Had you bothered to do any research, you might have found that Childress shot 55.2% from the field and 49.2% 3PT with Royal Ivey as his starting PG.
Meanwhile, your beloved Raja shot 45.7 and 44.2% (and 2 less rpg in 7 more mpg) with the world's greatest point guard. And I suppose Raja was a "loser" who "didn't know how to win" last year when he played for Utah.
And neither of us has "proof" that Bell is somehow a superior defender than Childress.

Gladwell and his band of pointyheads ALREADY have a different appreciation for the game, beyond that of those shiny-object jock-sniffers with which you have so brilliantly aligned yourself.
Instead of your pseudo-macho posing, you might want to open a book and *see* the game you're missing.

R.A. Porter said...

First off, why do my friends have an obsession with jocks and their contents?

On to more "serious" notes. I was in no way implying that quantification is a bad goal, merely that when one's measurements so defy conventional wisdom, it might be time to step back and re-examine the metrics. Perhaps perceptions do need to be overturned, or perhaps our metrics need to be rejiggered. My specific problems with the Win Score and Win Shares is in their predictive ability and my perception, wrong or not, that they undervalue actual wins.

Addressing the latter topic first: I've had the discussion with O before that what fantasy leagues need is some punishment and reward for victories and losses. It's possible (though I imagine nigh-impossible in practice) to put together a great fantasy roster exclusively from losing teams. As coaches are wont to say, "a W is a W." And like it or not, the Wolves aren't getting many of those anymore. From my perspective in the desert (the heat does distort one's vision, mind you) value = wins.

As for predictive ability, let's compare Josh Childress' '04-'05 and '05-'06 seasons as well as Raja Bell's. Now, I'm abusing the Win Score formula a bit here, but I believe I'm retaining its spirit. For the season just completed, Childress had a Win Score of 6.75, and Bell had 4.95. This sounds about right to me, as Childress is a young stud on a bad team and Bell was primarily retained for his "tenacity". This puts the 'Fro in the top 10-15% of players and the Clothesline in the top 25% or so. However, in the season prior, Childress posted a 6.25 to Bell's 2.20. Granted, it's but a single player comparison, but Win Score definitely wouldn't have helped Brian Colangelo last summer. In retrospect, he looks prescient, but the goal of quantitative measurements is to help reduce the need for GMs to be seers.

Finally, as to my dis of Childress as being underrated...by what measure is he underrated? He was selected sixth in the '04 draft, was All NBA Rookie 2nd team, and will probably make a few All-Star games in his career. Maybe there's some analysis missing - was his PER significantly lower than his WS? is there a massive anti-fro media campaign I've missed? -- or is it just a qualitative impression the authors have that Josh Childress is unloved?

R.A. Porter said...

Nathan, in his most charming mode (ie: calling me an idiot in an IM,) pointed out that in fact the under/overrated lists do consist of "...the players who’s Wins Score rankings differ the most from their NBA Efficiency rankings". I of course saw that the first time I'd read Malcolm's post, but inconveniently forgot it by the time I posted. So I am corrected. By the Win Scores metric, these players are measurably mis-rated.

I still doubt the predictive ability of these measurements in a vacuum, though they can certainly be added to a GM's arsenal of analytical tools. In fact, mixing and matching several of the flavors of metrics - PER, WinVal, Win Shares, etc. - might allow teams with different philosophies the opportunity to rate "their kind" of players.