When a starting QB carries a designation, does forecast.build conserve the TEAM's throwing volume (attempts, yards, TDs) it serves when nobody is designated? H-130 names the D-76 market anchor; at the release config (PW_MARKET_ON=0) latest_rows withholds the market (H-071), so the anchor site is not entered and what can be discarded is the room level lines_for reconciled -- anchored when the market is on, fitted when it is off. The mixing code is the same either way.
Owner lane: Not recorded
Question
When a starting QB carries a designation, does forecast.build conserve the TEAM's throwing volume (attempts, yards, TDs) it serves when nobody is designated? H-130 names the D-76 market anchor; at the release config (PW_MARKET_ON=0) latest_rows withholds the market (H-071), so the anchor site is not entered and what can be discarded is the room level lines_for reconciled -- anchored when the market is on, fitted when it is off. The mixing code is the same either way.
Prediction
OUT: k=0 retained QB-room attempts in [0.85, 1.10]; k=1 in [0.25, 0.50]; k=2 in [0.60, 0.85]; k>=3 in [0.88, 1.00) with median strictly below 0.995. Q: k=0 in [0.90, 1.02]; k>=1 median in [0.88, 0.995) through week 18. Receiving yards/TDs track QB-room yards/TDs within 0.01. Control dispersion about the common D-27 factor < 0.002. Football reference 0.93-1.03. REFUTED if OUT k=1 median > 0.80 or Q k>=3 median >= 0.995.
Finding and verdict
H-130 REFINED AND SUPPORTED on the production builder, stated more broadly than it was filed: forecast.build discards team QB-room volume for any team whose starter carries a curve designation, market or no market. Out: 100% of the room in the designated week (7 of 8 teams), 68% at k=1, 28% at k=2, 8.6% every week after. Questionable: 8.6% every remaining week. Football keeps 97% [95%, 99%] of a team's attempts when its QB1 sits. Receivers inherit the k>=1 discard through D-25 clause 3. 7 of 8 preregistered intervals held; the k=0 interval FAILED (predicted 0.85-1.10, observed 0.000) because the premise that a backup is promoted was false -- there is no backup on the board -- and the OUT control-dispersion threshold FAILED (0.0053 vs 0.002). Live only once designations reach the build (H-227, H-245 still OPEN).
Source provenance and publication scope
Owned research record: research/scientist/experiments/2026-09-25T2118Z-h130-rebuild-mixture-discard.json
This public reading view includes authored question, finding, review, reasoning and next action fields. Raw measurements, commands, logs and local paths are withheld. The record ID clock is not proof of completion or deployment.