NBA Action is fantastic. It really is. At the same time, it's over-marketed, excessively corporate and not always the greatest thing to watch. Here we'll chronicle the good and the ill.

Tuesday, August 21, 2007

Great Post on Defensive Rankings

While offense has been pretty easy to quantify, defensive basketball measures have been pretty unsatisfactory. Everyone understands that you can get a lot of blocks (Adonal Foyle) or steals () and still be a very poor defender. Measuring how well you defend involves effectively measuring the other player: how often they miss shots, turn the ball over, or generally fail to do effective things. This is pretty infeasible from aggregate box scores, because it's hard to tell who was guarding who, when and where the points came from, and other relevant factors.

82games, and the associated information revolution (cf. Popcornmachine, APBRMetrics, shot charts) has made tremendous strides in this direction.

Here's the latest incarnation: the defensive composite score. I really like where this is going; it throws out box scores completely and combines defensive +/-, d-rating (see b-r.com for an explanation) and counterpart PER into one metric.

The second round of results look good. The list of best defenders includes the obvious candidates - Garnett, Duncan, Bowen, Yao, Okafor - without which this measurement would lose all credibility. But it also captures the specialists and lesser known names like Tyrus Thomas, Devin Harris, Joel "The Vanilla Gorilla" Pryzbilla and Rajon Rondo.

Most convincing are the statistics for bad teams. Look at Washington (one player above the median), Milwaukee (5 players in the bottom 20%) and Seattle (all players in the bottom half of the league).

0 Comments:

Post a Comment

Subscribe to Post Comments [Atom]

<< Home