H-071 asks to refit D-77 GRADIENT_SHARE {rush_yds 0.60, pass_yds 0.19, rush_att 0.17}. Before sweeping a constant, does it reach any player line on the rows the release scores?

Owner lane: Not recorded

Question

H-071 asks to refit D-77 GRADIENT_SHARE {rush_yds 0.60, pass_yds 0.19, rush_att 0.17}. Before sweeping a constant, does it reach any player line on the rows the release scores?

Prediction

At the release config every swept GRADIENT_SHARE arm changes exactly 0 of ~6,867 rows on every field and the gradient site is called 0 times, because rows_for carries no mkt_team_total and latest_rows withholds it (G-TDANC). Under the keyed positive control every non-1.0 arm moves > 1,000 rows on carries/rush_yards and the QB rows on pass_yards. H-071's MAE sweep is therefore unmeasurable on the production path and its refit prediction is moot. REFUTED if any release-config arm moves any row.

Finding and verdict

MEASURED NOTHING on the production path, exactly, and that is the result: D-77 GRADIENT_SHARE has zero reach at the release config (0 site calls, 0 of 6,867 rows moved by any of 12 arms), because no production row carries mkt_team_total since G-TDANC (2026-09-15). The positive control that the 09-25 referee lacked works (keyed market, 1,632 site calls, up to 6,257 rows moved). H-071 as registered is moot and is CLOSED; a refit is meaningful only if PW_MARKET_ON=1 is ever adopted, which is a market-as-input decision (D-320/D-380), not a constant refit. Consequences for the register: H-138 (one row's missing market silences the team anchor) and H-130's 'anchor survives on the published object' describe a site that the release never enters. Secondary: the rush_att fraction reaches carries only, never yards or fantasy points, contrary to its code comment.