The 'effect of being up X'

[ilardi]

I think you can just divide the coach rating by 5 to get the impact of the coaches' rating on the player rating.

That certainly sounds plausible, but it's ultimately an empirical question, right? For example (whenever you have a few minutes to spare), I think it would be informative to see a comparison of RAPM rating for each of the OKC players with and without the coaching adjustment. Does Brooks's whopping +6.2 coaching effect simply get distributed on a minutes-weighted basis across the entire roster, or are some players disproportionately "penalized" by the Brooks coaching adjustment?

In fact, I think most of us are ultimately interested in knowing, for each NBA player: (a) what is his actual +/- impact on offense/defense?; (b) how much of that impact is attributable to coaching/team context variables?; and (c) how much is fully attibutable to the player after adjusting for such coaching effects? After all, when we look at other advanced metrics (e.g., true shooting %), we never adjust for coaching effects . . . and even if we did/could, we would still also want to know what the actual unadjusted metric value was, since that's the number that tells us what the player ACTUALLY DID. Make sense?

[DSMok1]

I think you can just divide the coach rating by 5 to get the impact of the coaches' rating on the player rating.

That certainly sounds plausible, but it's ultimately an empirical question, right? For example (whenever you have a few minutes to spare), I think it would be informative to see a comparison of RAPM rating for each of the OKC players with and without the coaching adjustment. Does Brooks's whopping +6.2 coaching effect simply get distributed on a minutes-weighted basis across the entire roster, or are some players disproportionately "penalized" by the Brooks coaching adjustment?

In fact, I think most of us are ultimately interested in knowing, for each NBA player: (a) what is his actual +/- impact on offense/defense?; (b) how much of that impact is attributable to coaching/team context variables?; and (c) how much is fully attibutable to the player after adjusting for such coaching effects? After all, when we look at other advanced metrics (e.g., true shooting %), we never adjust for coaching effects . . . and even if we did/could, we would still also want to know what the actual unadjusted metric value was, since that's the number that tells us what the player ACTUALLY DID. Make sense?

Very good points. A is the easiest, splitting to B and C is the next frontier.

[AcrossTheCourt]

Perhaps this is why "clutchess" has been difficult to find in players. Some of that noise might be due to how often you are behind/ahead, though the effect decreases the closer the game is.

[J.E.]

After all, when we look at other advanced metrics (e.g., true shooting %), we never adjust for coaching effects . . .

I feel like we should, though. And obviously adjust for who you're playing with, too.
Does Belinelli shoot 46% on 3-pointers on average NBA team? Does Stephenson grab 6 DReb/game on an average team? If we could create 'adjusted BoxScore stats' for each player we would get a more accurate SPM.
I feel that's pretty much the next frontier in terms of improving player metrics

[ilardi]

After all, when we look at other advanced metrics (e.g., true shooting %), we never adjust for coaching effects . . .

I feel like we should, though. And obviously adjust for who you're playing with, too.
Does Belinelli shoot 46% on 3-pointers on average NBA team? Does Stephenson grab 6 DReb/game on an average team? If we could create 'adjusted BoxScore stats' for each player we would get a more accurate SPM.
I feel that's pretty much the next frontier in terms of improving player metrics

Yes, but we need to see *both* (adjusted and unadjusted) forms of each metric to get the complete picture . . . i.e., it is important to see player APM numbers both with and without the coaching adjustment.

[colts18]

Maybe I'm looking at it wrong but isn't this some kind of evidence for momentum existing?

[v-zero]

It is in fact evidence of anti-momentum. That is, performance above expectation will tend to cause said performance, relative to the opponent, to worsen.