THE CARD
Call Result The grade Detail Mahomes starts Week 1 HIT — started, all 4 quarters at-log 27 → Brier .5329 · last mark 68 → .1024 both printed, no netting Companion draft buy graded −0.2 picks entry 104.1, final 104.3 flat loss beside a hit call Rams under 11.5 first data point LAR 0–1 (27–7 in Melbourne) sixteen games grade it Cleveland (MISS, graded 8/26) formality confirmed Watson wire-to-wire at JAX closed The draft-season book, final: 2 wins, 2 losses, 1 flat — Jefferson WIN, Pierce WIN*, Nabers MISS, Bowers DEAD, Mahomes-companion −0.2.
A hit that takes its medicine
Mahomes started Monday night — took the first snap, played all four quarters, threw for 184 and two scores, ran for another, absorbed two sacks and scrambled four times. The backup logged zero snaps. The bubble-wrap scenario the desk had pre-flagged was a ghost: this was a full, unmanaged start, and the final injury report had said exactly that — full participation all three days, no designation.
The record still takes an ugly number, and that's the convention working, not failing: the desk's entry claim, logged in July, was 27% — the wrong side of the line — and the record grades the entry. Brier 0.5329, printed in the loss column's font. Beside it, the revision path: 62 from the day he cleared physicals in late July, 68 at the grade after the cutdown tripwire — right for six consecutive weeks, scoring 0.1024 at the last mark. This is the exact mirror of the Nabers grade, where the desk marked down into a loss and got no credit for the markdown. Same rule, both directions, no exceptions: you are graded on what you said when you said it.
A pre-registration that lost, graded like the rest
Last week Riccardo pre-registered the uncomfortable ranking: if his paperwork observables cleared Mahomes while the draft market kept fading him, the market wins. The week then ran the experiment: the market faded Mahomes literally to the end — the companion buy closed at −0.2 picks — while the Saturday filings said full participation, no designation. The paperwork was right. The market was wrong. The pre-registration graded against its author, and he owns it: when the question is a narrow compliance binary — will he dress and take the first snap — filings answer exactly that question, while markets price a fuzzy aggregate of risks around it. Filings dictate binaries; markets price spectrums. Logged at n=1, no overcorrection — but logged, in public, the same way the desk logs everything else.