Comments on: Building an Updated Statistical +/- Model for 2009

By: Scout

Scout — Mon, 20 Sep 2010 19:52:24 +0000

Where is Pops Mensah-Bonsu's family from exactly?

By: Neil Paine

Neil Paine — Thu, 23 Apr 2009 17:22:05 +0000

Actually, pretty much every linear weights formula out there has a negative coefficient for "shot attempts". In fact, I'd be worried about this one if it didn't have TSA/40 showing up as negative. Because the good parts of shooting -- the shots that actually go in the basket -- are already captured by Pts/40 and the V.I. by extension. All that's left after that is the cost of the possession you use by shooting, in terms of what your teammates could have scored had you passed off. More important to me is the fact that TSA/40 squared is actually positive and very significant. That basically proves there is a usage/efficiency tradeoff. It proves that high-volume shooters like your Allen Iversons, etc., the guys who have been roundly dissed by the stats community for low shooting percentages, are actually helping their teams by having a willingness to take on low-percentage shots. Because somebody's gotta take a shot during a possession (or at least try to). I mean, we can't just stand around and play the four corners forever.

By: TV

TV — Thu, 23 Apr 2009 06:05:17 +0000

Curious that TSA/40 has a negative coefficient, no? The conventional wisdom would have it that there is value to usage. Maybe this is just capturing a usage/efficiency tradeoff indirectly?

Or is it capturing efficiency more directly, in the sense that holding pts/40 constant and increasing TSA/40 implies lower efficiency..?

By: Jose

Jose — Wed, 22 Apr 2009 20:15:15 +0000

This thing shows how much the Spurs really miss Manu Ginobili. He wasn't healthy last year against the Lakers in the WCF and now he can't play at all. Without him, will the Spurs lose in the first round? I hope not.

By: Caleb

Caleb — Wed, 22 Apr 2009 15:59:20 +0000

Only a + 0.14 for David West? Seems low... in fact my Hornets look pretty sad based on this metric. Chris Paul with his amazing 12.65 and everyone else with negative numbers or barely positive numbers.

By: Eddy

Eddy — Wed, 22 Apr 2009 07:36:23 +0000

Great stuff, Neil.

Thanks for taking the time to compile all the data.

By: Gerrit

Gerrit — Wed, 22 Apr 2009 05:47:06 +0000

That D-Will is so low in both is really surprising. If it were just one metric showing him at a low value I could just shrug it off as a flaw in the metric, but seeing that both agree make me wonder if he isn't really the second best point guard in the league.

82 Games does some tracking of charges drawn.
http://www.82games.com/charges0607.htm
http://82games.com/FSORT10.HTM
http://82games.com/charge.htm
Is three years enough to run a good regression?

By: Neil Paine

Neil Paine — Wed, 22 Apr 2009 04:13:25 +0000

I just used pure age as the variable. And I agree, the Williams/Miller comparison is strange, but it actually mirrors what you see pure adjusted +/-:

http://basketballvalue.com/teamplayers.php?team=UTA&year=2008-2009

http://basketballvalue.com/teamplayers.php?year=2008-2009&team=CHI

Miller's 2-year APM is 5.79, vs. just 1.65 for D-Will, so the regression is really just doing its job, in terms of predicting the non-boxscore imapct from the regular stats.

In the future, I would love to add charges and any other numbers I can get my hands on... The problem, of course, is that the NBA doesn't bother to keep track of those numbers, so I'm ultimately limited by what's in the database. I recently heard a rumor that we'll be seeing stats like charges being officially tracked in the semi-near future, but I'll believe it when I see it.

By: Gerrit

Gerrit — Wed, 22 Apr 2009 02:50:13 +0000

For age did you do abs(age-27) or just pure age?

It also seems like the importance of versatility skews the results for point guards and centers. Seeing Brad Miller at +4 while Deron Williams is only at +2 seems really odd. In fact it's one of the few oddities that the metric produces.

If you decide to recalculate the metric could you also use charges drawn?

By: Jason J

Jason J — Tue, 21 Apr 2009 13:59:34 +0000

Great stuff, Neil! This is why I check the BBR blog every morning. I'd be interested to see if this adjusted methodology effects the all-time list you compiled in a significant way. A couple of rankings that caught me eye: Jason Richardson's SPM ranked much higher in his time in Charlotte than Boris Diaw's did, yet the Bobcats clearly performed better after the trade. I wonder if that's just an exception to the rule, or if there was an uptick in the SPM of Wallace or Okafur when Diaw arrived that we're not seeing in the numbers. Allen Iverson has a plus ranking - something neither Prince nor Stuckey can say. At first I assumed this was a clear exception, but then you see that the only Piston who has a really strong plus ranking is Rasheed. Their post-Iverson surge just happened to take place when Rasheed came back and got healthy. So this may be a case of people lobbing too much blame at AI not meshing and not giving enough attention to the impact of Wallace being absent from the line up.