evidence / Frontier measurements

True intelligence -- counterfactual, context, microstructure, forward-graded, cross-sport

The other evidence pages ask how well calibrated are we. This wave asks the questions a scout, a trader, or a skeptic actually asks: what if this player were gone, why did the win-prob move, when does the market do its work, who grades the graders, and does the same method survive across four sports. Every number below is copied verbatim from a committed out/*.json under `scripts/platformkit/analytics_showcase/out/`; every module carries edge_claimed: false in its own metadata. Nothing here is a dollar / ROI / edge claim -- these are calibration and intelligence analytics, and the honest not_buildable / underpowered / roster-confounded verdicts are displayed as prominently as the wins, because that is the whole point.