How to compare a later version
Use the same player and metric scope. Record both cutoffs, identify newly verified information, and explain any editorial change separately. A subsequent NFL outcome must not be fed back into a supposedly pre-draft prediction.
Source revision: 0967a87ef88fb21ebaeaab379ed2060052c825c0
Cohort version: 2026-11-15-29-eye-test.
Evaluation horizon
No success-rate backtest is claimed here. The selected achievement cohort retains its existing end-of-2025 observation rules. Broader-archive outcome labels and a uniform evaluation horizon must be established before predictive testing.