Skip to content

Complete Player-Prior RAPM Baseline

This is the recovered-coverage player-only control for HIPSTER PM. It preserves the full forward player-prior pipeline but disables lineup context and excludes all box-score prior features.

Returning players receive a regularized aging forecast built from prior RAPM, prior exposure, age and experience, draft/entry information, physical profile, and an age-by-prior-RAPM interaction. That final interaction is value- conditioned aging: it allows the expected age transition to vary smoothly with the player’s previously estimated value. The full preseason prior vector is then centered using prior-season on-court possessions.

First-year players receive the forward exposure-gated cold-start prior:

\[ \mu_i^{cold}=p_i^{low}R^{replacement}+(1-p_i^{low})R_i^{draft}. \]

For every season, the draft-rate ridge, exposure gate, replacement token, aging transition model, and RAPM regularization selection use only earlier completed recovered regular-season data. The target-season forecast is materialized before that season's RAPM update. No player score from box-score residuals, composition functions, or matchup functions enters this model.

Recovered-Coverage Results

Artifact: artifacts/models/forward_complete_player_prior_rapm/2023-24_to_2025-26/forward-complete-player-prior-rapm-2023-24-to-2025-26-20260813T210918Z-6520555d.

Possession RMSE Eligible game RMSE Full-game RMSE Winner accuracy Team NetRtg RMSE Pythagorean-win RMSE
1.198061 14.2105 14.4756 67.64% 3.6754 7.7561

Per-Season Regular Results

Season Possession RMSE Eligible game RMSE
2023-24 1.193156 13.8723
2024-25 1.202055 14.2200
2025-26 1.198795 14.4787

Frozen Playoff Check

Season Possession RMSE Eligible game RMSE
2023-24 1.191543 16.7687
2024-25 1.194370 16.8245
2025-26 1.192351 16.2592