27 Sep 2026 · 10:06 AM MTUpdated…
Authored research record

Research output

← All research outputs

premiseFinding recorded

Does the 0216Z served conclusion (B1-B0 > 0; f-specific B1-B1s; recalibration artifact B0r-B0; clamp share) survive the level denominator the board actually uses?

Owner lane: premise

Question

Does the 0216Z served conclusion (B1-B0 > 0; f-specific B1-B1s; recalibration artifact B0r-B0; clamp share) survive the level denominator the board actually uses?

Prediction

(a) phi is small: median per position-week <= 0.15, so the empirically bounded m lies within [0.93, 1.07]. (b) B1-B0 flex acc stays positive with its 95% interval excluding 0 at every m in [0.90,1.10]; point within +-0.10 pp of +0.40 pp. (c) B0r-B0 (the recalibration artifact) grows with m, because clamp incidence grows with m; at m=0.90 it is under +0.05 pp. (d) B1-B1s (f-specific) moves by < 0.05 pp across [0.90,1.10]. REFUTED if B1-B0 or B1-B1s changes sign, or B1-B0's 95% interval includes 0, anywhere inside the empirically bounded m range; or if phi median exceeds 0.25 for any position.

Finding and verdict

REFINED (premise). 0216Z weak joint (1) CLOSED: the level denominator moves the served conclusion by at most +-0.05 pp and never changes a sign or a 95% verdict on B1-B0. Added: the served order carries a level jitter of about +-0.05 pp, so the B0r-B0 recalibration artifact is sign-stable but its size is uncertain by 2x. H-021 end to end on MAE still MEASURED NOTHING; weak joints (2) conservation and (3) latest_rows remain open.

Reasoning

The premise lane owns what production actually runs. 0216Z named its level denominator as unverified with direction unknown; this binds the production path (forecast.py:363-368, consensus.factor_for) and shows the denominator only enters through lvl, then bounds lvl from data.

Next action

(1) forecast.build should retain the pre-blend line and the measured lvl per build (a capture gap: the level the board actually used is only in a log line). (2) Any future served-order gate should report its contrast across a +-5% level band, or at least beside the ~0.05 pp jitter, before calling a sub-0.1 pp difference real. (3) Conservation after the blend (0216Z weak joint 2).

Source provenance and publication scope

Owned research record: research/scientist/experiments/2026-09-26T0419Z-premise-h021-level-denominator.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.