-0.0 NG/84 · 81 projected GP → -0.0 net goals over the season

1. Age. Every past season is first moved to his 2026-27 age along each component's career curve (Age adj: what a typical player gains or loses between that age and this one). Games from 2026-27 so far count like 2025-26 games, with no age adjustment. 2. Recency. Per component, the seasons are averaged with weights that fall off with age; Memory is the half-life in seasons, fitted walk-forward (penalty kill and EV defense change fast, finishing and playmaking slowly). 3. Shrinkage. Each component is pulled toward the average forward who reached the NHL at 22-23 (Prior). Trust = n / (n + K): n is the effective number of games behind the weighted rate, K how many games of evidence that component needs before it outweighs the prior (fitted walk-forward on 2013-2025-26 seasons, then every K is multiplied by 0.5: shrinking the ten components one at a time ignores that a player above the prior in one is usually above it in the others, which compressed the total by about a fifth). Projected = Trust × his rate + (1 − Trust) × Prior. The ten projected components add up to his NG/84, the number the standings simulation dresses him at. Proj GP = games played so far plus the games the standings simulation dresses him for (injuries from ESPN return dates; goalies split 65/35 starter/backup).