The same engine that rates a golfer runs his week.
From the course advance to the post-round debrief, round ops runs off settled truth rather than re-grading him on a round. It is the operator's cockpit across the phases of a tournament week: the advance and pre-round plan, in-round live ops, the between-rounds packet re-planned by leaderboard position, and the post-round strokes-gained debrief. It references the durable read and hands each round to the Pipeline as one piece of evidence, never a verdict.
A plan drawn from settled truth. Not a fresh grade.
The advance is built from the course-and-field read, not a fresh evaluation. The engine takes the venue's archetype and setup vector, the golfer's course-fit against it, the field, the weather and tee-time wave, and the hole-by-hole demand, and outputs a pre-round plan keyed to where this course rewards and punishes his game. It carries its confidence gate, so a thinly covered course would be flagged.
Every line of the plan is drawn from upstream truth, his phases, his archetype, his fit, and never from a fresh grade. The engine already knows he is a bomber with a strong fit here; the advance just turns that settled read into a hole-by-hole plan and a tee-time note, at a stated confidence. The advance turns a settled read into a plan; it does not re-grade the golfer to build one.
Illustrative engine read on the real advance structure (venue archetype and setup vector, course-fit, hole-by-hole demand, weather and wave, confidence-gated). Composite golfer and course, demonstration figures.
Speak only when the decision is live. Re-plan by position.
In the round the engine speaks in caddie-usable language, minimally, and only when a decision is live. The support is a bounded decision set: the club-and-target and risk-selection choice for the shot in front of the golfer, not a running commentary. Between rounds it hands over a structured adjustment packet, and the leaderboard-position adjustment is the golf-specific piece: a golfer chasing and a golfer protecting a lead play different strategies, and the packet re-plans.
Press the scoring holes. Take the aggressive line on all four par fives and the wide drivers, and accept the added variance: a golfer four back needs to make ground, so the plan raises the risk on the holes where his length pays.
The anti-spam rules are locked: the engine offers a read only when there is a real choice with strokes on the line, and stays silent on the stock shots. The between-rounds packet re-plans by leaderboard position, because four back and one ahead are different games, and every output is confidence-gated. None of it re-grades the golfer. Speak only when the decision is live, and re-plan by position, never re-grade by round.
Illustrative engine read on the real live-ops and between-rounds structure (bounded decision set, locked anti-spam rules, the leaderboard-position adjustment). Composite situation, demonstration figures.
A round is an input. Not a verdict.
After the round the engine debriefs it in strokes gained by phase, against the field and the conditions, and against the plan rather than raw. It produces the KPG for the round and folds into the post-tournament packet. Then the loop closes: the round becomes evidence the Pipeline weighs and the course intelligence accumulates, and crucially the debrief updates evidence, it does not recompute the golfer's rating from one round.
The strokes-gained debrief reads the round against the plan: the length paid as designed, the putter was cold and is regressed hard rather than read as decline. The KPG grades the round, the KR does not move, and the round is handed to the Pipeline as one weighted piece of evidence. A round is an input, not a verdict. A round is an input the Pipeline weighs, never a verdict that overwrites the rating.
Illustrative engine read on the real post-round structure (strokes-gained debrief against the plan, the KPG, the plan-to-outcome loop and self-scout feeding the Pipeline without overwriting the KR). Composite round, demonstration figures.
Round ops scouts the course, prices the live decision, re-plans by leaderboard position, debriefs the round in strokes gained against the plan, and hands the round to the Pipeline as one piece of evidence. It never lets a single round overwrite a rating, because a week of operations is built on the durable read, not the last scorecard. The read runs the week; the week does not rewrite the read.
The week runs on the read. See what it feeds and references.
Round ops references the player and course reads to build the week, and its rounds feed the grades and the Pipeline. It is the operating layer that turns a settled rating into a plan, a live decision, and a debrief.