Regularized adjusted plus-minus
RAPMA regression method that isolates a player’s on-ice impact after adjusting for teammates, competition and situation.
RAPM uses regression to separate a player’s effect on outcomes (like expected goals) from the influence of the teammates and opponents they share the ice with, plus zone starts and score state.
Regularization shrinks noisy estimates toward the average, making RAPM more stable than raw plus-minus for comparing players in different roles.