Forward Centered Aging Bounded Hierarchical Portable-Matchup Contextual RAPM

This candidate tests whether an explicit, leakage-safe player-rating reference improves the recursive age-informed portable contextual state. It is identical to the aging bounded portable-matchup model except that every season's complete player prior vector is centered at its prior-season possession-weighted mean.

Because every RAPM row contains five positive and five negative player indicators, adding a common constant to every coefficient leaves lineup predictions unchanged. The centering rule therefore turns an otherwise arbitrary coefficient location into a documented pre-season reference:

\[ \widetilde{\mu}_{i,t} = \mu_{i,t} - \frac{\sum_j p_{j,t-1}\mu_{j,t}}{\sum_j p_{j,t-1}}. \]

The centered prior enters the same age-informed returner branch and exposure-gated draft/replacement cold-start branch as the uncentered model. Its contextual term remains:

\[ C(A,B)=h(x(A))-h(x(B))+q(x(A),x(B)). \]

Evaluation Status

The full through-2025-26 recursive evaluation is in progress. Its frozen regular-season, playoff, full-game, and team-level metrics will be compared with the uncentered aging candidate before this model is promoted.

Use the training guide to reproduce the candidate.