1991-92 NBA RAPM

1991–92 NBA Season
Regularized Adjusted Plus/Minus
249-game sample 371 players ridge regression offensive & defensive split
RAPM isolates each player’s per-100-possession impact on both ends of the floor by regressing all simultaneous lineup combinations against observed point differentials. The 95% credible interval (Low / High) reflects estimation uncertainty — players with fewer possessions have wider intervals. Positive = better than average per 100 possessions.
# Player Tm O Poss ORAPM DRAPM Total Low High
Using the 249-game sample, team win totals were projected three ways: MLE (maximum likelihood — raw sample rates scaled to 82 games), Bayes (Bayesian estimator with a 0.500 prior — shrinks extreme small-sample records toward league average), and compared against ground truth (actual full-season records). The Bayes estimator substantially outperforms MLE, particularly for teams with few games sampled.
GB = games behind division leader · RMSE computed over all 27 teams vs actual wins

Squared Statistics

This analysis draws on the Historical RAPM Project — the only possession-level lineup database for the pre-play-by-play NBA era, reconstructed game by game from video since 2021.

View the Project → Data & Consulting →

Data available for the 1984–85 through 1995–96 NBA seasons · squared2020.com

Leave a Reply