I've got a standardized per-36-minute rate for players, in which the standards are per 100 points and per 44 rebounds per game (per team). With additional adjustments, the goal is to estimate a player's Sco, Reb, etc rates for any team; but especially for a historically average one.
So I take a Euclidean distance from the avg for each stat and get a difference measure. Other similarity scores may do it such as to approach 1.00 for 100% identical, but in this exercise, zero is identical.
Our hypothetical all-average player is thus .00 different from himself:
Code: Select all
dif per-36 rates tm Sco Reb Ast PF Stl TO Blk
.00 aaavg Joe nba 15.0 6.6 3.0 3.0 1.1 2.0 .8
.25 Max Strus Cle 13.9 5.6 3.4 2.8 1.0 1.6 .7
.26 Naji Marshall NOP 14.1 7.6 2.9 2.4 1.3 2.1 .5
.29 Deni Avdija Was 14.0 6.8 3.6 3.1 1.2 1.9 .4
.32 Kevin Huerter Sac 15.1 6.3 3.3 3.0 1.1 1.2 .5
.38 Shaedon Sharpe Por 15.2 5.6 2.8 3.0 1.1 3.0 .4
.39 Keldon Johnson SAS 14.8 7.3 3.3 2.8 .9 1.4 .3
.40 Jaime Jaquez Mia 15.1 5.4 2.9 3.1 1.5 1.9 .3
.41 Donte DiVincenzo NYK 15.7 5.1 2.6 2.4 1.6 1.9 .5
.43 Buddy Hield Ind 15.4 4.4 2.8 2.9 1.0 2.2 .9