One player on the universal scale, read through the park he plays in, with the confidence that sizes every number.
The OVERALL is one number, but it is never a closed one. It opens into the three components the player is made of, and it is read anchor-first: overall offensive value and the component quality that predicts it come before any traditional line. Batting average and RBI are held at PROXY, in view, never driving the grade.
The OVERALL is 82.4, and every layer under it is legible: the three components, their readings, and the tier of data behind each. His bat and his glove are MEASURED off Statcast, his approach is INFERRED and held at lower confidence because the feed cannot see intent, and the traditional line sits at PROXY where it belongs. A rating you can press to the component, the reading, and the tier is a rating you can argue with, which is the only kind worth having.
Illustrative engine read on the real KR hierarchy (OVERALL, the three components Production / Athletic / Baseball IQ, each reading on its own MEASURED / PROXY / INFERRED / UNSCORED tier, defense credited by the spectrum). Composite shortstop, demonstration figures.
The pitcher runs a contained Pitcher Sub-Engine onto the same three components and the same universal scale, so a pitcher and a center fielder can both be KR 88, the same comparable rating reached through nearly disjoint machinery. The anchor is run prevention read defensive-independent, then the stuff and command reads, and ERA, wins, and saves are held at PROXY.
The pitcher reaches KR 88.0 through defensive-independent run prevention, MEASURED stuff, and a command read still INFERRED, with the injury term broken out at its own confidence rather than blended in. ERA and wins sit at PROXY, because the runs that scored and the games credited to him are shaped by the defense and the offense around him. A pitcher and a center fielder can both be KR 88, because there is one currency and the machinery that reaches it is the engine's problem, not the operator's.
Illustrative engine read on the real Pitcher Sub-Engine (run prevention defensive-independent, stuff and command, the injury and workload term broken out at its own confidence). Composite starting pitcher, demonstration figures.
A two-way player runs both weighted slices, the position-player roll-up and the Pitcher Sub-Engine, combined by the two-way rule: the two roles are additive in team value but reported as two component reads, never averaged into a mush, because he is genuinely two players worth of value in one roster spot.
He is reported as an 84.6 hitter and a 90.1 pitcher, each with its own confidence, with an additive team-value note, and never as one averaged number. Averaging them would erase exactly the thing that makes him rare, so the engine refuses to. Two players worth of value in one roster spot, told as two reads, because that is what is true.
Illustrative engine read on the real two-way rule (the position-player roll-up and the Pitcher Sub-Engine combined additively in team value, reported as two graded reads with separate confidences, never averaged). Composite two-way player, demonstration figures.
A pitcher and a center fielder can both be KR 88, because there is one currency and the machinery is the engine's problem, not the operator's.
Player Intelligence gives you a single OVERALL you can open to the reading and the tier of data behind it, priced with the confidence that sizes the bet, honest about the box the traditional line hides.