Every time the last 15-minute candle is a bullish engulfing, then the last 15 minutes returned 5% or more within 1 hour

82 matches·all scanned symbols·2026-07-11 to 2026-07-18

More often lower: 51 of the 82 completed matches fell over the next 30 minutes (62.2%).

82 matches in 5,369,372 eligible market-minutes across 662 markets.

No reference rate is pinned to this report yet; republishing after the baseline ships will pin one.

▸ Replay an occurrence Run this study →

OCCURRENCES PER DAY · UTC · SUM 82

FWD_RET_30M DISTRIBUTION · 82 PRESENT · 0 ABSENT10 FIXED BUCKETS

REPORT INVALID

Integrity invalidated 2026-07-20. In this record_result.v2 pin, 32 of 50 visible sequence-step-1 evidence rows showed none or bearish instead of the required bullish value. The sequence evidence materialization used the final anchor bucket rather than each matched step bucket. That engine defect is fixed for new record_result.v4 scans. Do not use this pinned report as verified evidence. Corrected successor: q#3c0016a5 re-asks the same question over the full recorded corpus under record_result.v4.

Corrected successor: q#3c0016a5

What the record shows

22 of the 82 (26.8%) traded 5% or more above the match at some point inside 1h. That is a high-water mark, not a close. Matches appeared on 6 of the 8 UTC days scanned; the busiest day was 2026-07-15, with 20 of the 82 matches.

Evidence and methodologyExact search, pin, freshness, missing data and versions

Published 2026-07-18 by @admin · q#ed3ac65e

Exact search

Every time the last 15-minute candle is a bullish engulfing, then the last 15 minutes returned 5% or more within 1 hour

SCOPEtimes.anchor_time between 2026-07-11T07:06:45.000Z to 2026-07-18T07:06:45.000Z

STEP 1feature.candle_engulfing_15m = bullish

STEP 2feature.ret_15m ≥ 0.05within PT1H

Read-only. Every value above was confirmed by the author before the run. The prose on this page is commentary; these values are the research.

PINNED RESULT - NEVER CHANGES

v2 · nrm.v1 · ft.v1 · rev 3e792f17…397015e2 · q#ed3ac65e

This ID reruns the exact same search: same answer, any day, any account.

FRESHNESS

The record grows nightly. A later rerun is a new, labeled result; this pinned result never silently changes.

Outcome diagnostics

HEADLINE SPLIT (mfe_1h)

22 OF 82 (26.8%) traded 5% or more above the match at some point within 1h (mfe_1h, a high-water mark and not a close); 60 OF 82 did not. measured forward from each match.

PEAK DAY · SPAN

peak day 2026-07-15: 20 OF 82 (24.4%). span 8 UTC days scanned: 6 with matches, 2 with zero.

DISTRIBUTION SHAPE (fwd_ret_30m)

modal bucket -5..-2 pct: 20 OF 82 (24.4%). 51 OF 82 closed below 0 · 31 OF 82 at or above 0.

PRESENT / ABSENT BY METRIC

fwd_ret_30m: PRESENT 82 · ABSENT 0

mfe_1h: PRESENT 82 · ABSENT 0

mae_1h: PRESENT 82 · ABSENT 0

THRESHOLD COUNTS (fwd_ret_30m) - BOTH SIGNS, EXACT PARTIAL SUMS OF THE BUCKETS

≥ +121 OF 82

≥ +216 OF 82

≥ +57 OF 82

≥ +101 OF 82

≤ -137 OF 82

≤ -229 OF 82

≤ -59 OF 82

≤ -103 OF 82

EVERY ROW IS A PARTIAL SUM OF THE BUCKET COUNTS ABOVE: THE TABLE AND THE CHART CAN NEVER DISAGREE.

Methodology and versions

DEFINITION

2 steps · 1 scope

feature.candle_engulfing_15m = bullish THEN feature.ret_15m ≥ 0.05 WITHIN PT1H. the chips at the top of the page are this definition, verbatim.

EXECUTION

edge-triggered · 1m UTC grid

the predicate becomes true having been false; sequences chain greedy-earliest and anchor at completion; episode dedupe PT1H. deterministic at every worker count.

DENOMINATORS

5,369,372 of 6,672,960

eligible market-minutes scanned (one market for one minute) across 662 markets; 1,303,588 excluded where a measurement was not yet available - missing data shrinks the denominator, it never counts as zero; 15 market-days excluded by the source ledger. counts are matches within the scanned window, never a base rate.

OUTCOMES

forward from anchor · in-scan

closed threshold list ±1/2/5/10 pct, 10 fixed buckets across 3 forward metrics, computed over all occurrences pre-paging. an outcome absent at the record edge (horizon incomplete) is counted, never truncated or implied zero. nothing on this page is filterable by outcome. nothing here is a guarantee or trading advice.

OBSERVATION SCOPE

setup as-of anchor

setup values are as-of the anchor bucket close (point-in-time freeze); window aggregates cover their stated pre-anchor duration; every value the predicate consulted is at or before the anchor.

pinned 2026-07-18 · 82

the pinned result (counts, page-1 rows, outcomes, outcomes summary) never mutates. later dataset revisions are new runs, never merged.

SORT

times.anchor_time asc

the only record sort in v2.0; page limit 50. the pinned rows are page 1 of 82 - charts and rates derive from the summary over all occurrences, never from the page.

VERSIONS

q#ed3ac65e · rev 3e792f179546

schema research_query.v2 · normalization archive_normalization.v1 · features feature_defs.v1 · encoding record_result.v2. same key implies a byte-identical rerun.

DATASET REV · FULL

3e792f1795461fcf84a7f7aa3cb6b79cc132458a07f62cce576c6d79397015e2

Examples and replaysFirst 50 of 82 matches

These rows are examples, never the denominator. The counts and charts above cover every match. Each row opens 30 minutes before the recorded moment.

SYMBOLANCHOR · UTCFWD30MMFE1HMAE1HREPLAY

TACUSDT2026-07-16 21:22−2.94%+2.10%−4.26%replay →