No number in golf is course-agnostic. The course is the test.
Production is always produced on a course, and courses ask for genuinely different games. The engine reads the golfer against the course as a demand profile, the archetype crossed with a seven-dimension setup vector. It reports his value on that course, the KR, and, separately, how well his game covers what the course asks, the fit. The best-fit course stays a hypothetical, surfaced only on demand, so the real number never competes with the flattering one.
The archetype crossed with the seven-dimension setup.
A course is typed in two layers, run together. The archetype (the family) sets the base demand shape, and the seven-dimension setup vector is the demand profile. Together they re-weight the golfer's phase and trait profile to this specific test. Pick a family and move the setup, and watch what the course asks for.
The demand profile is what the course asks of the four phases, before any golfer is priced against it. Distance is most predictive at the longest, most open tests; accuracy at the tightest. The setup can swing the fit adjustment by roughly a stroke per round. The course is the test, and every test asks for a different game.
Illustrative engine read on the real Course Demand Profile structure (the course archetype crossed with the seven-dimension setup vector). Composite course figures, demonstration values, setup weighting v0.
Two different objects. Never merged.
The KR is a magnitude, the golfer's value on that course. Course fit is a coverage read, how well his profile covers what the course asks. They are two different objects, and the engine never merges them. Here is the same bomber on two courses that ask for opposite games.
On Course Y his KR barely moves, because he is still an elite golfer and the KR is his value on that course. His fit collapses, because his shape does not cover what the course asks. A backer reads both: the magnitude tells you how good he is here, the coverage tells you how comfortable the bet is. A high KR and a poor fit are not a contradiction. They are two honest answers to two different questions.
Illustrative engine read on the real Course Demand Profile structure (archetype crossed with the seven-dimension setup vector, fit as a coverage read distinct from the KR magnitude). Composite golfer and courses, demonstration figures, fit adjustments v0.
OVERALL is the real read. The explorer never moves it.
OVERALL is the golfer read through the representative course distribution of his actual schedule, badges baked in, legend-anchored. It is the default and only required rating: there is no neutral-course Base number. The Course Explorer runs the identical math on any course you select and re-prices live, and OVERALL never moves.
Select a course and the on-course KR and the fit both re-price, live, off the same demand math. The OVERALL above them does not move, because it is the real read through the real schedule. The ceiling is a separate question with a separate answer, offered only when asked. The default output is the true one: the golfer through the course he actually plays.
Illustrative engine read on the real Course Explorer structure (OVERALL through the schedule distribution, live re-pricing on the selected setup vector, the best-fit ceiling on demand). Composite golfer and courses, demonstration figures, fit and setup adjustments v0.
Production is always produced on a course. The engine reports the golfer's value on the course he plays and, separately, how well his game covers what that course asks, and it holds the best-fit course as a hypothetical surfaced only on demand, so the real number never competes with the flattering one. A KR and a fit are two honest answers to two different questions, and merging them would hide the one a backer most needs.
The course prices the golfer. See what it feeds.
The demand profile and the fit read run under the field, the simulation, and the week's strategy. This read is the test every other golf number is produced against.