H-183: the passing-TD market carries no model information because our line IS the market's implied total through the D-76 anchor (R2 0.91) and the price is a normal curve at a half-integer. Is that still what production runs, and what does the pass-TD line carry now?

Owner lane: premise

Question

H-183: the passing-TD market carries no model information because our line IS the market's implied total through the D-76 anchor (R2 0.91) and the price is a normal curve at a half-integer. Is that still what production runs, and what does the pass-TD line carry now?

Prediction

Written before the analysis run: shipped room pass_tds vs implied team total R2 well under 0.5 (the anchor cannot fire), anchored control R2 > 0.95; the shipped football-only line adds nothing measurable to the market's team total (increment interval spans 0) but still beats a constant climatology. REFUTED as a premise check if shipped R2 > 0.8.

Finding and verdict

REFINED. The conclusion survives and gets stronger; the mechanism is reversed. The line no longer copies the market: G-TDANC switched off D-76 on every production path (R2 0.146 against implied total, 0.989 with the anchor re-fired in memory). What is left is a football-only throwing-TD line that adds no measurable information once the market's team total is known (+0.0012 [-0.010,+0.012] team-game MSE), loses to that team total by 0.096 [0.052,0.143], does not beat a constant at the team grain and loses to one per QB (+0.069 [+0.006,+0.139]; line r=0.104 against outcomes with sd 0.43, too wide for its signal). The price-shape clause is second-order: normal against Poisson at the three book lines moves Brier by at most 0.006. The centre is the defect, not the curve. At 1.5 (81% of throwing-TD captures) the shipped normal puts 0.486 on the over against 0.438 realised; the qmap median shift is not replicated here.