"Regularized Adjusted Plus-Minus (RAPM) is a method for evaluating skater performance that seeks to improve on the known issues with on-ice metrics (such as raw on-ice Corsi For %, etc.). Raw on-ice skater ratings are inherently impacted by the teammates a player plays with, the opponents they play against, a player’s percentage of zone starts in the offensive or defensive zone, back to back games, etc. There are many aspects of the game that players have no control over. The goal of RAPM is to control for these aspects and provide a better measure of a player’s offensive or defensive ability.
RAPM seeks to account for all these factors by using a statistical technique called linear regression (specifically, a regularized linear regression called ridge regression). This allows us to control for all teammates, opponents, score state, zone starts, etc. at the same time further increasing the accuracy of a specific player’s rating. The method can be used at various strength states and for various metrics. On Evolving-Hockey, we provide 2 strength states (EV & PP/SH) with 3 separate metrics (Goals, Expected Goals, and Corsi). Each regression is run for each season and players are separated by team (traded players are split into multiple versions of themselves – one version for each team they played for in each season)."
RAPM accounts for many of these variables you're mentioning.
Wait, you said Celebrini is scoring at a higher pace and I proved you wrong. Now PP points don't count? All goals count the same.
