trading / confluences / b_avwapday
Corpus lane — RG-07 · Zero variance on the filled rows — not scoreable.
Is price on the signal's side of the session VWAP anchored at 09:30 ET?
RG-07 anchored-VWAP day gate. 1 iff price at the decision is on the signal's side of the session VWAP anchored at 09:30 ET. VWAP is built from CLOSED 1m bars of the decision's own session only, typical price (h+l+c)/3 weighted by volume; the reference price is the last closed 1m close at the decision instant. No bands, no trend clause — the live c2 uses `vwap_conf='band-trend'`, and this is deliberately the plain side-of-AVWAP form the source describes. Missing session data -> 0. Evidence tier: OPINION (the backlog's own label for RG-07).
Every parameter below was fixed in the design document before any candidate value was computed. No sweep, no tuning: a bit that needs tuning to pass has failed.
| Field | Value |
|---|---|
| Source component | RG-07-avwap-chop-day-gate |
| Evidence tier | OPINION |
| Backlog rank | #6 |
| anchor | "09:30 ET" |
| bands | false |
| Design doc | backtests/conf_lab/SCREEN-2026-08-07-DESIGN.md |
| Doc sha256 | e2ba7ca77acce01c03ebcdb7afb51ffade7bb7a591303a2923329dba13a91d94 |
Conditional association on mean solo R, stratified on the count of the OTHER live bits, against a per-bit prevalence-matched null (50,000 i.i.d. Bernoulli draws at this bit's own prevalence, scored on its own strata). Population: 2,882 candidates, 2,289 filled, MES, 2019-06-04 → 2026-07-31.
DEGENERATE — not scoreable at all. Prevalence is exactly 1.0000 on the filled rows: every single candidate that filled was on its own side of the session AVWAP, so the conditional estimator has only one arm and no effect, CI or p-value exists. This is not a weak result, it is the absence of a measurement. It consumes no FDR slot. In hindsight the reason is structural — a P1 breakout candidate has by construction just pushed through a level in its own direction — which puts it in the same family as the b_closeconf tautology, found by a different route. Its evidence tier was the slate's weakest to begin with: OPINION, the backlog's own label.
Only in a form with variance: bands, a distance threshold, or an anchor that is not the session open. That is a NEW bit with new parameters and a new pre-registration, not a retest of this one.