APBRmetrics

https://www.sloansportsconference.com/r ... in-the-nba

I mentioned interest in this topic recently (... and had a brief conversation about average RAPM for similar type players with a GM about 10 years ago...)

There are some confirmations of conventional wisdom and some notable findings. I'd suggest people read it for themselves but will pull some elements out.

Has anyone else in the basketball media written about this paper? If so, share a link.

"To better evaluate positional plus-minus, we redefine seven offensive
and three defensive positions in this study..."

"Overall, we adopt a hierarchical Bayesian linear model to estimate positional offensive and
defensive plus-minus in the NBA from 2015-16 to 2021-22. We then compare estimated plus-minus
ratings across positions and analyze how these positional values have changed over the years."

"Pick and Roll Attackers are the most valuable offensive contributors in most seasons."

My comments in parentheses.

(Think about teams emphasizing that play type such vs. those that don't.)

(Spot up shooter have an lowest average offensive plus minus among non-bigs. Post-up bigs have the highest average value among big types.)

"...the gaps between positional offensive plus-minus have
shortened over the years."

"Bigs defensive
plus-minus are consistently higher than those of other positions’ defensive plus-minus."

"...the variances of Post-up Big’s, Pick and Roll Attacker’s,
and Versatile Scorer’s offensive plus-minus seem to be considerably higher than other positions."

"There seems
to be a notable improvement for Roll and Cut Big’s offensive plus-minus as the posterior mean
increased significantly from 2015-16 to 2021-22. Post-up Bigs and Skilled Bigs also seem to
experience some progress on offensive plus-minus."

(The year to year changes in positional averages and variances could use more study. The use of distribution graphs instead of distribution tables and stats causes some difficulty on speed / accuracy of reading.)

"... the offensive
contributions from Spot-up Shooters may not vary drastically from player to player."

(Think about who spend a lot on spot up shooting vs. not and what they are getting.)

(Table 2 shows large changes in number of players and certain types gain / lose more than others.)

(Average number of stints at about 60,000 per season for league... so 2,000 per team and a bit less than 25 per game.)

(Any reactions here? Any notable discussion at the conference by those at presentation or poster or just cognizant & discussing ad-hoc? Any panelist on any basketball panel show any awareness / interest in this paper?)

As mentioned in the K Perkins thread, I think Plus Minus by position for individual players would be interesting. Probably 3 year to maintain sample size.