How much of kicker fantasy error is FG attempts, how much the conversion/distance arithmetic, how much PATs, realised team points and level -- and do the premises H-201 rests on (line runs 0.44 low; FG line indistinguishable from a constant; kdef_rows is the shipped line; shuffled truth returns to shipped) hold when run?

Owner lane: Not recorded

Question

How much of kicker fantasy error is FG attempts, how much the conversion/distance arithmetic, how much PATs, realised team points and level -- and do the premises H-201 rests on (line runs 0.44 low; FG line indistinguishable from a constant; kdef_rows is the shipped line; shuffled truth returns to shipped) hold when run?

Prediction

Frozen in protocol.json before any arm. H-201 as registered: O1 attempts truth -0.9 to -1.6 pts per kicker-week (25-40% of MAE); made-by-distance a further -0.3 to -0.6 over attempts; PAT truth -0.2 to -0.5; realised team points -0.5 to -1.0; hindsight level -0.03 to -0.10; shuffled controls within +/-0.05 of shipped. Premise lane added: P1 bias interval contains 0.44; P2 fitted FG MAE vs constant spans 0; P3 served-shape (name '') minus registered within +/-0.02; P4 the +/-0.05 shuffle expectation is FALSE for O1, O2, O3, O5 and no shuffled arm beats shipped; O7 conversion truth worth less than O1 attempts truth.

Finding and verdict

TESTED. Of kicker fantasy MAE 3.722 (market-anchored walk-forward 2023-25, 1,634 kicker-weeks), actual FG ATTEMPTS at the fitted conversion is worth -1.607 [-1.814, -1.395] (43%); actual conversion and distance at the FITTED attempts only -0.483 [-0.616, -0.365] (13%), so attempts carry 3.3x the conversion arithmetic and the surprise condition (attempts under -0.4) did not occur. Given attempts, actual made-by-band adds a further -0.97 [-1.05, -0.89] (O3-O1, 95%). Realised team points in place of the implied total: -0.336 [-0.429, -0.242] (9%) -- the kicker's week is the TD/FG split of his team's drives at fixed points, not the points. PAT truth ALONE: +0.022 [-0.061, +0.099], spans 0, MSE +0.54; with attempts known it adds -0.38 (O8 -1.989 [-2.183, -1.790]). Post-hoc (disclosed): FG-points and PAT-points residuals correlate -0.18 (cov -1.24), so the PAT term hedges the FG term and fixing it alone removes the hedge. Hindsight level +0.016 [-0.011, +0.041]: the 0.44 under-projection is real in the mean and worth nothing in MAE. H-201's registered prediction was WRONG on five of six arm ranges (O1 slightly larger than its range; O3-over-O1 twice its range; PAT, team points and level all smaller) and on its shuffle control. Every searched value is model-relative (kicker_week arithmetic with one truth substituted), never a bound on football.

Reasoning

The kicker line reads one market scalar; everything that decides a kicker's week at a fixed total -- whether drives stall into field-goal range or finish, which is where the opposing red-zone defence and the own coach's fourth-down decisions sit -- enters through nothing.

Next action

Filed as a new row: at a fixed implied team total, does the SPLIT of drives into TDs and FGs have a pre-kickoff signal on EITHER side of the ball (own fourth-down go rate, opponent red-zone TD rate allowed), measured against realised FG attempts?