A coach's record is mostly his roster. The engine rates what is left.
A head coach's win total is entangled with the roster he inherited or was handed and the regime around him, so the engine points inward and rates the coach on what he actually adds: the development his players realize beyond their projected curves, the in-game decisions that move the margins of close games, and the record read with the regime and roster separated out. It rolls the pillars up to a rating on the KR scale, read against a coach legend, with a career-versus-current split because regimes change, and it never moves a player's current number. The coach is the institutional term of the three-bet doctrine: the same player is a different bet under a different staff.
Point the engine inward, and separate the coach from what he was handed.
Two composite coaches at the identical .620 record. The engine points inward, the same doctrine that grades a player, and separates the coach from what he was handed. The two are not the same bet.
Two coaches at the same win rate are different bets once the roster is separated out, and the engine will not credit a coach with a number the Team KR says the roster produced. The archetype and identity triggers place the coach the way the front-office reads place a general manager, and the career-versus-current split carries both the long arc and the current regime. Developing players and acquiring them are different skills, and a staff is often strong at one and weak at the other.
Illustrative on the real coach-rating framework (the inward turn, the pillar roll-up to a KR-scale rating read against a legend, the career-versus-current split, the regime-and-roster separation, the develop-versus-acquire split, the archetype). Composite coaches, demonstration figures.
Not whether his players are good. Whether they got better than projected.
A composite staff, its players tracked from the day they arrived. The development pillar is not whether the players are good. It is whether they got better than the curves they were projected for on entry.
The lift is largest at the technique-heavy, slow-developing positions, the residual is confidence-gated on the sample of players and years (a two-season read is a lead, a decade is a fact), and it is separated from talent acquisition. It adjusts the trajectory and its confidence, never a current KR.
Illustrative on the real coaching-development residual (realized versus projected across a sample, the upward re-pricing of acquired prospects, the position-specific magnitude, the confidence gating, the never-touches-a-current-KR rule). Composite staff, demonstration figures.
A good staff does not make the roster better. It makes better decisions with it.
A good staff does not make the roster better. It makes better decisions with it, and those decisions move the outcome at the margins that decide close games.
The three bets are kept separate and each priced, because conflating them is how a coach gets credit for a roster or a player gets blamed for a landing spot. The coach is the load-bearing input to the third bet, and counterfactuals are a separate, explicitly labeled query. The coach is the third bet of three, not the first.
Illustrative on the real in-game and institutional layers (the coaching actor's decisions on the expected-points and win-probability differential, the quality scaling and the coaching-change confidence event, the adjustment-outcome tracking, the three-bet doctrine with the coach as the institutional term). Composite coach and situation, demonstration figures.
A win total is a joint output of the roster and the coach and the bounce of the ball, so reading a coach off his record credits him with a roster he may not have built and a variance he did not control. The engine points inward and rates the coach on what he actually adds, and it measures that as a residual: his players' development against the curves they were projected for, his in-game decisions against the expected points each choice was worth, his adjustments against the outcomes they produced, and his record against what the Team KR said the roster should have done. It separates developing players from acquiring them, because they are different skills and a staff is often strong at one and weak at the other, and it separates the current regime from the career arc, because a coach's number moves when his staff and his situation change. It rolls those pillars into a rating on the same scale and doctrine as everything else, gated on the sample, banded by its confidence, and it never lets any of it touch a player's own number, because the coach is the third bet of three and not the first. Rate the coach on his residual, keep the roster's share with the roster, and the sideline stops being a place where reputations are laundered into facts.
Separate the coach from the roster. Rate what is left.
Head Coach Intelligence points the engine inward, measures a coach as the residual of his players' development, his in-game decisions, and his record against the roster, and prices him as the institutional bet, never touching a player's own number.