Methodology

Regularized Adjusted Plus-Minus (RAPM)

RAPM is a player impact metric that estimates how much a player contributes to their team's scoring differential per 100 possessions, after controlling for the quality of teammates and opponents on the court. Unlike box score stats, RAPM uses no individual player statistics. A player's impact is estimated solely from the points their team scores and allows while they're on the floor.

How it works

RAPM uses ridge regression across every possession in a season. Each possession is treated as an observation where the 10 players on the court (5 offense, 5 defense) are the explanatory variables and the points scored are the outcome. By processing hundreds of thousands of possessions, the regression isolates each player's individual contribution while accounting for who they played with and against.

Each player receives two ratings: one for their offensive impact (ORAPM) and one for their defensive impact (DRAPM). A player's total RAPM is the difference between these two values. A positive ORAPM means the team scores more when the player is on offense. A negative DRAPM means the team allows fewer points when the player is on defense (negative is good for defense).

Multi-season RAPM

Single-season RAPM can be noisy for players with fewer minutes. To improve stability, we also compute 2-year and 3-year RAPM by including possession data from prior seasons with recency weighting (most recent season gets the highest weight). Multi-season RAPM better captures sustained performance and is less susceptible to small-sample noise.

Credits

Our RAPM implementation is based on the methodology described by Jeremias Engelmann, the creator of RAPM and co-developer of Real Plus-Minus (RPM). For a deeper technical explanation, see his 2026 article on building adjusted plus-minus.