trading / confluences / b_onbreak

b_onbreak — overnight-range breakout alignment SURVIVOR-BENCHED

Online lane — practitioner · Cleared the screen's FDR gate; benched, not registered.

What it asks

Has price already broken out of the overnight range in the direction of the trade at the moment of decision?

Exact definition · verbatim docstring from backtests/conf_lab/bits.py (sha1 5473adbbf694)

Overnight-range breakout alignment (practitioner standard). 1 iff the
last CLOSED 1m close at the decision is strictly beyond the overnight
window's high (long) / low (short).

Overnight window = CLOSED 1m bars with OPEN timestamp >= (decision day - 1
day) 18:00 ET and < decision day 09:30 ET, read from
`view.history("m1", n=2000)`. Fewer than 30 bars in the window -> 0.
Sunday/holiday gaps get NO special case: >= 30 bars counts, otherwise 0.

Corpus diagnostic, stated not hidden: the POSITIVES-REPORT overnight-high
finding was MNQ-only and flat on MES.

Fixed parameters & provenance

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.

Screen analytics — screen-2026-08-07

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.

Per era · reported separately, never pooled into a verdict — each cell is a smaller sample than the headline

Funnel & fill conditioning · a bit that mostly predicts whether a limit FILLS is not the same discovery as a bit that predicts whether a trade WINS

Novelty vs the six live bits · high agreement means the bit is mostly restating a gate that is already in the stack

By stratum — does it work where the method actually trades? · the live A/B take rule is count >= 4, so the count>=4 rows are the only ones that could change a live trade

Config-grid involvement — 2026-08-07

Paired replay on the certified conflab-livegate-v1 stream, $250-nominal basis (~2.6% high versus engine dollars), 31 cells examined with ZERO multiplicity correction and no p-values. Every cell below is post-hoc on data the bits were screened on.

Status — SURVIVOR-BENCHED

The screen's strongest survivor and the one object the whole 2026-08 programme produced that deserves to live — as a BENCHED hypothesis, not a challenger. Pooled +0.166 R, matched p=0.0141 (98.59th percentile of its own null), BH q=0.197 at q=0.25, bootstrap CI [+0.018, +0.320] clear of zero, positive in BOTH eras, and — uniquely among the added bits — it keeps a real conditional effect (+0.120) in the count>=4 stratum where the live method actually trades. It was nominated for the challenger seat by the screen, then the 2026-08-08 tribunal declined to spend the seat: its best grid cell (CONFIG_7_k4) fails the study's own Stage-1 bar — winners-capped-at-1R flips the delta to −$5,937, the added set nets −$2,811 without its own top five trades, the whole delta annualises to $380/12m against a $1,500 bar, and its $17.57 marginal edge per added trade is under one tick of friction at the book's actual 5.38-point median stop. It is also the textbook fill-predictor: limits fill 70.6% of the time when the bit is on versus 82.2% when it is off, so part of the measured lift is fill selection by construction.

Retest path

Two things, both named by the tribunal. (1) An UNBURNED grid: a future pre-registered grid on post-2026-07-31 data in which a b_onbreak-additive cell clears all seven Stage-1 items — including cap-at-1R on the PAIRED delta and added-set ex-top-5 — at >= $1,500/12m pace. (2) A SHADOW-FILL LOG that costs no challenger seat: signals-only, no orders, no registration, logging whether count==3 + onbreak limit orders actually fill at >= ~80% with entry prices at-or-better than touch. If fill reality beats the assumed discount and a conservative replay keeps the added set at >= half its nominal delta, the ruling reopens.

Standing caveats

• Fill model optimistic: touch fills, zero net slippage, entry-bar SL/TP not checked inside the fill minute.
• Not a holdout — 2019-2026 was selected on, tuned on and reported on before this lab existed. A good number here means 'worth pre-registering', nothing more.
• MES only; the P1 edge is MES-centric and recent-regime.
• Masters miss four Sunday-evening blocks (May-Jun 2026); data ends 2026-07-31.
• Mean solo R is an average per candidate. The solo ledger is not a book and is never summed (PROTOCOL.md §0).
• Per-era cells are reported separately and are NOT pooled into a verdict; each is a smaller sample than the headline.
• Nothing on this page is TV-verified. A backtest number becomes actionable only after an independent TradingView Pine port reproduces entries AND per-trade PnL on a Deep Backtest export.
• Screening is not certification. Promotion requires all seven PROTOCOL gates, ending in a forward paired evaluation and a Pine referee.

← all confluences