Statistically Determining Defense

[Ghost of Podkolzin]

There will always be stuff the numbers cannot tell us. How many rotations did Luka miss. How many times did he not close out on a shooter. How many times did he not get back on defense.
I guess one could argue that the sum of all of it is +/- but even regressions in large sample sizes like 5-year RAPM cannot completly reduce "randomness".

In a way that is what the C.Wood debate comes down to. A player that is looking good if we are just looking at counting stats or tracking data. But where do I find the stat that tells me how many times Sweeney had to call out defensive assignments or rotations because even 60 games into the season he still didn´t know the scheme.

Ya, I don't know.  Let me play with the "contested shots" a bit.  So far my model really points to defensive production, but there has to be something pointing to engagement.  I watched all 82 games of the Mavs.  I never once thought "get Wood out of there; he's killing our D".  Perhaps I don't know shit, or perhaps Wood was banging Kidd's wife.  Stranger things have happened, like with Nash's ex-wife and Lebron's mom.  Has Kidd ever shown this disdain for another player when he was in BRK or MIL?  Something personal had to have happened.