APBRmetrics

I was looking at an EZPM file recently and the usage100 column and started wondering how much usage affects total offensive impact or APM (of any stripe). I couldn't quickly find any past discussions of Usage100 at Evan's site. Evan, it is correct that usage100 is a value completely separate from OFF100, right? Do you have any thumbnail estimate of how much the average Usage100 is of "total offense" or total EZPM or how it correlates to offensive RPM? I am having trouble getting on track on this subject and wondered if any one else had thoughts about it. I wondered if it might help in the consideration of very high (Westbrook) and low usage guys (like a Demarre Carroll, for lack of a better example at the moment) on RPM to know the estimate impact of that usage, via the usage 100 estimate. For Westbrook his usage 100 positive impact (+3.39) is about twice as big as his Offensive 100 impact (+1.73). Carroll is +0.70 on OFF100 but -0.96 on usage 100. Seems worth knowing or thinking about this split in offensive impacts, especiaslly if usage were to change due to change in role or team.

There is also usage adjusted adjusted rating and UARPM from nbacouchside. There might be an opportunity for useful triangulation between these different metrics imo, but I haven't looked into it enough yet.

nbacouchside: “I adjusted the UARPM100 numbers using a per minute adjustment for each player on the team so that total team plus-minus was equal to team SRS (basically point differential adjusted for strength of schedule) via Basketball-Reference”

How much did you have to adjust on average? I have wondered how skew apm type data is from srs. Most skewed teams?

Given recent work by bbstats on usage, fg efficiency and rapm https://mobile.twitter.com/bbstats
I am bumping this back into view. There might be something in the works cited above that could be useful when viewed next to the new work. I intend to try to chew on it more later.

What if you computed offensive rpm for a player (or all players) for the set of plays where he'd get usage credit or better yet true usage credit and compared that to rpm of that player for his non-usage plays? Would this form of situational rpm be helpful to the usage-rpm discussion? I think it might. Not necessarily a clear cut simple answer as the two play sets are related but if a players usage is helping at least part of the gain should be seen in the difference in efficiency between the sets (in typical cases).

Some low usage players may be lower usage players when playing with a particularly high usage player and not so low otherwise. If one looked at the rpm in these two situations, one would have info about how situational their rpm is, that is how much it is related to usage, I think.