How to actually use this

Every study here tells you what happened before, under a clear condition. That is context for how you prepare and review, not a signal to trade off of. Here is where the line sits.

Use this research for

Pre-market preparation and building realistic expectations
Comparing today's structure with a historical category
Understanding common versus uncommon market behavior
Reviewing whether a trade idea matched the market condition
Improving journaling and post-session review
Avoiding overreaction to normal market behavior

Do not use this research for

Automatic entries, exits, price targets, or stop placement
Predicting the next bar or assuming a percentage must occur today
Replacing risk management
Replacing an independently validated trading strategy
Treating any historical frequency as a forecast or guarantee

The library shows you the history. Pro helps you build the habit.

This library tells you what NQ has done before under a given condition. Net Alpha Pro helps you turn that into a repeatable routine: how you prep before the open, how you review after the close, and how you manage risk in between. We are coaches who trade, not salespeople who teach, so Pro is process and discipline, never signals or promises.

Static research packet archiveEvery library study as a full research packet with extended aggregate tables.

Aggregate scenario matricesCondition-by-condition historical tables for structured preparation.

Preparation checklistsDaily worksheets that connect each study to a pre-market routine.

Responsible-application lessonsHow to use historical context without turning it into a signal.

Daily educational commentaryEnd-of-day narrative that references certified findings in context.

Review frameworksJournaling and post-session review structures built on the research.

See how Net Alpha Pro turns HTA research into a structured daily preparation and review process

Net Alpha Pro is educational. It is not a data terminal, live market-data display, signal service, or source of guaranteed outcomes. Historical behavior may not repeat.

How this research is built

The publication layer never invents statistics. Full methodology →

Certified dataStatistics are computed only from hash-verified, certified historical datasets with documented defect registers.

Frozen definitionsSession windows, metrics, and eligibility rules are fixed before results are computed.

Uncertainty by defaultEvery rate carries its sample, Wilson 95% interval, and multiple-testing control.

Reviewed releaseAutomated QA, a rights-safe publication preflight, and founder approval precede every release.

Grade	Meaning
A. Established	Large sample; stable across periods, definitions, and volatility conditions.
B. Established with caveat	Generally stable, with one meaningful sensitivity stated prominently.
C. Conditional	Valid only under defined conditions, regime-sensitive, or smaller sample. Directional findings from the 2016–2026 rising market stay at C until re-tested on falling and sideways sub-periods.
Exploratory	Interesting but not established; excluded from headlines.
Restated	A prior figure changed; old and new remain linked with the cause explained.

Version history

v1.0 · 2026-08-13 · Public launch. First public release: 12 NQ studies selected for launch from the certified library (criterion: reader value and demand). Statuses flipped to PUBLISHED with founder approval recorded: Reid Nabua (co-founder), 2026-08-13. All other studies remain internal pending approval. Lead gate added: full reports unlock with first name + email. No research figures, samples, or definitions were changed in this release.

Every publication, data refresh, and restatement is recorded here. Restated figures always link the prior value and the cause.

Library versions

Date	Version	Change
2026-08-09	Manifest 0.36.0	Copy cleanup only, no research figures changed: removed every em dash across the site (study text, cards, charts, methodology, definitions and changelog), replaced with commas or hyphens for a cleaner, plainer read. Still 105 candidate studies, all CANDIDATE_PENDING_APPROVAL.
2026-08-09	Manifest 0.35.0	Added the NQ 4-Hour Candle Structure Census (Wave 12): ten studies putting the entire 4-hour candlestick playbook to a fair test over 2,222 trading days. Zero directional signals survived multiple-testing correction, big bodies, long wicks, outside bars, engulfing, swings/BOS and level rejections all landed on the coin-flip line. What replicated is non-directional and useful: after a compressed 4-hour bar the next bar is larger ~79-84% of the time (vs ~30-35% after a shock) as range size mean-reverts; inside bars carry a small range tilt but are not a coiled spring; deep impulse pullbacks coexist with new extremes; and the 4-hour chart trends only ~21-27% of the time. A full-history model missed the frozen +0.020 gate and the 5m/15m expansion-compression edge did not transfer to 4H. Router frozen, 0 contracts. Library now holds 105 candidate studies across NQ and GC. Still pending founder approval.
2026-08-09	Manifest 0.34.0	Trust-and-correctness pass (no research figures changed). Fixed seven charts whose bars or reference lines ran off-canvas: the volatility, volume and overnight clocks now read as clean multiples (e.g. 4.9×, not 4.9%), and the London/Asia and prior-day location maps are rescaled to fit. Shareable filter links now work, the instrument, timeframe, category and search state is written into the URL and restored on load. All clock times are relabelled Eastern Time (ET), daylight-saving aware, instead of the fixed EST tag. Methodology no longer claims a single certified window (it uses multiple certified panels, each study states its own), and the restatement page now records the real Wave-6 to Wave-7 correction instead of “none yet.” Still 95 candidate studies, all CANDIDATE_PENDING_APPROVAL.
2026-08-09	Manifest 0.33.0	Added the NQ Multi-Bar Sequence Census (Wave 11): four studies, and the first strongly confirmed positive finding in several waves. After hard current-state matching, an expansion bar followed by a compressed bar raises the odds of another high-range bar, positive in all 17 calendar years on both 5m and 15m, with a clean dose-response and persistence out to three bars. It is memory about size, not direction, and it authorizes 0 contracts. On the negative side: the popular directional candle narratives (three-bar impulses, failed reversals, large-body runs) stayed on the coin-flip line, a small-body “run” claim survived raw FDR but collapsed to zero after matching, and a full sequence-history range model improved AUC in all 8 folds (+0.0151 on 5m) yet missed the frozen +0.020 gate. Router frozen, 0 contracts. Library now holds 95 candidate studies across NQ and GC. Still pending founder approval.
2026-08-09	Manifest 0.32.0	Added the NQ Swing Structure Census (Wave 10): four studies putting the popular market-structure playbook to a fair test. With causal pivot confirmation and equal-distance geometry, HH/HL and LH/LL alignment, break of structure, BOS reclaim and failed swings all landed on the coin-flip line for the next-hour path (0 of the preregistered tests survived FDR), and a looser threshold did not change it. Swing state duration is banked as descriptive (the high 15m persistence is partly a slower-confirmation artifact), the failed-low case is a mild non-significant tilt, and adding the whole swing feature set lifted model AUC only +0.0086 with a worse Brier, so the gate failed. Router frozen, 0 contracts. Library now holds 91 candidate studies across NQ and GC. Still pending founder approval.
2026-08-09	Manifest 0.31.0	Added the NQ Prior-Day x Overnight Confluence study (Wave 9): four studies from ten hypotheses on whether where the overnight range sits relative to yesterday predicts the RTH day. A clean six-state location map is banked as description; a couple of large raw signals (accepted-outside persistence ~70%/62%, a -41-point sweep-and-reject contrast) dissolved under controls, current 09:30 location explains the persistence, the sweep flips positive in holdout after matched controls, and fair equal-distance path races sit at a coin flip. The model challenger passed 0 of 6 gates, so the router stays frozen. 0 contracts. Library now holds 87 candidate studies across NQ and GC. Still pending founder approval.
2026-08-09	Manifest 0.30.0	Added the NQ Multi-Session Handoff study (Wave 8): six studies on how the overnight book carries into RTH. Confirmed: RTH forms ~64% of full-day highs and ~57% of lows (overnight makes the rest), and the overnight range is almost always broken on at least one side (RTH stays fully inside just 4.3% of days). Context-only: opening outside the London range re-enters ~82% of the time, but the equal-distance fade is a coin flip. No-go: London-vs-Asia shape and one-sided overnight closes did not predict the RTH day type, and adding overnight structure to the day-type model passed 0 of 6 walk-forward gates, so the router stays frozen. 0 contracts. Library now holds 83 candidate studies across NQ and GC. Still pending founder approval.
2026-08-09	Manifest 0.29.0	Hover-to-define now underlines only the first appearance of each term on a page, keeping the prose clean. Removed a line from the homepage intro. No research figures, samples, or grades changed; still 77 studies (71 NQ + 6 GC), all CANDIDATE_PENDING_APPROVAL.
2026-08-08	Manifest 0.28.0	Hover-to-define. Terms from the Definitions page are now underlined on the study pages; hover or tab to one and its plain-English definition pops up, so readers never have to leave the study to look up a word. No research figures, samples, or grades changed; still 77 studies (71 NQ + 6 GC), all CANDIDATE_PENDING_APPROVAL.
2026-08-08	Manifest 0.27.0	Terminology pass, education-first voice. Every hard shut-down verdict now reads NO-GO or fail; the harsher one-word verb was removed throughout (verdict labels, captions, tags, and the glossary), and one internal study ID was renamed to match. No research figures, samples, or grades changed; still 77 studies (71 NQ + 6 GC), all CANDIDATE_PENDING_APPROVAL.
2026-08-08	Manifest 0.26.0	Expanded the Definitions page to 82 terms (added shock, HAR, quartile, AUC, p-value, Mann-Whitney and Wilcoxon tests, ablation, coverage, inside/mother bar, engulfing, streak, realized volatility, contract roll, and more) and re-sorted the whole glossary into a single A-to-Z list with letter headers. No research figures, samples, or grades changed; still 77 studies (71 NQ + 6 GC), all CANDIDATE_PENDING_APPROVAL.
2026-08-08	Manifest 0.25.0	Added a Definitions page: plain-English glossary of the terms and shorthand used across the studies (instruments, sessions and clock, price-structure and auction terms, statistics and testing, evidence grades and governance), reachable from the top nav. No research figures, samples, or grades changed; still 77 studies (71 NQ + 6 GC), all CANDIDATE_PENDING_APPROVAL.
2026-08-08	Manifest 0.24.0	Footer tidy. Combined the “Educational content only, not financial advice” disclaimer and the historical-research / risk note into a single paragraph in every page footer. No research figures, samples, or grades changed; still 77 studies (71 NQ + 6 GC), all CANDIDATE_PENDING_APPROVAL.
2026-08-08	Manifest 0.23.0	Compliance line added. Every page footer now carries a clear “Educational content only, not financial advice” disclaimer (HTA is a trading-education company, not a financial advisor or broker-dealer; nothing here is a recommendation or solicitation). No research figures, samples, or grades changed; still 77 studies (71 NQ + 6 GC), all CANDIDATE_PENDING_APPROVAL.
2026-08-08	Manifest 0.22.0	Time-zone clarity pass. Every clock time is now labeled U.S. Eastern Time (ET): all “ET” labels were changed to “ET”, each study page carries a “times shown are U.S. Eastern Time (ET)” line, and the finder and methodology state it plainly. No study figures, samples, or grades changed; still 77 studies (71 NQ + 6 GC), all CANDIDATE_PENDING_APPROVAL.
2026-08-08	Manifest 0.21.0	Display refresh. Simplified the finder cards to a headline number, a small chart, and one line, so the graphic pegs the click instead of a wall of text. Built full study pages with a simple chart for the 40 studies that were still card-only, so all 77 studies now have a page. Removed the “Common research paths” and “What we’re studying next” sections from the homepage. Copy tightened to the HTA voice. No study figures changed; still 77 studies (71 NQ + 6 GC), all CANDIDATE_PENDING_APPROVAL.
2026-08-08	Manifest 0.20.0	Added the NQ Frozen-Box Recapture study (Wave 7) and corrected Wave 6. Wave 7 re-tested the midpoint-recapture finding across seven box structures (opening range, initial balance, prior-day, Globex, Asia, London, rolling compressed). The raw ~60-67% opposite-boundary statistic replicated everywhere but collapsed to ~51/51/50% under equal-distance barriers, and a plain midpoint cross did just as well, so the effect is geometry, not a rotation edge (0 of 6 families qualified). The Wave-6 midpoint-recapture study was accordingly marked Restated: its raw figure stands, but its causal rotation/rebalancing interpretation is superseded. 0 contracts, router frozen. Library now holds 77 candidate studies across NQ and GC. Still pending founder approval.
2026-08-08	Manifest 0.19.0	Added the NQ Auction Migration & Balance study (Wave 6): six studies testing whether a sophisticated “mature balance” state beats simple compression. It does not, against same-clock, range-matched controls the complex label added no expansion advantage (87.3% vs 88.3% displacement). The useful finding is midpoint recapture: after a box displacement reclaims the frozen midpoint, resolved sequences rotate to the opposite boundary ~85-97% of the time (all eras, highly significant), but this generalizes beyond balance and the natural fade fails on costs (high win rate, negative expectancy). Midpoint migration inverted in holdout (NO-GO) and a displacement-volume filter failed FDR (NO-GO). Banked as structural context; 0 contracts, router frozen. Library now holds 75 candidate studies across NQ and GC. Still pending founder approval.
2026-08-08	Manifest 0.18.0	Added the NQ Boundary Extension & Path Quality study (Wave 5): the economics test of Wave 4's extension prior. Entering toward the developing high/low with a 25%-of-range target and stop passed 0 of 6 frozen checkpoint-side tests and 0 of 96 exploratory grid cells after a 1-point cost. The reason is path quality, ~45-51% of near-extreme signals traverse both a +25% and -25% move and ~30% of eventual winners stop out first, so eventual extension (a state statistic) is not a clean target-before-stop path (an execution statistic). One 11:00 near-high configuration was logged as watchlist-only; a noon range conditioner failed replication. NO-GO on the boundary-chase branch; 0 contracts, router frozen. Library now holds 69 candidate studies across NQ and GC. Still pending founder approval.
2026-08-08	Manifest 0.17.0	Added the NQ Intraday Auction Development study (Wave 4): eight studies on how the regular session builds its high and low. Final extremes are U-shaped in time (~63% form in the first or last hour); by 10:30 about 73% of days have exactly one boundary finished; the median range-completion clock is ~60% by 10:30 and ~80% by noon; closing near the developing extreme predicts a further material extension (64-72% vs ~45% control, +20-28pp, all eras); and for judging whether an extreme is already final, close location captures nearly all the skill while the opening-range break/retest sequence adds nothing. Reconstructed on 3,883 fully observed sessions (7 fail-closed exclusions); no entries tested, 0 contracts, router frozen. Library now holds 65 candidate studies across NQ and GC. Still pending founder approval.
2026-08-08	Manifest 0.16.0	Added the NQ Auction Day-Type study (Wave 3): a frozen five-class RTH day-type taxonomy (two-sided rotation and sustained acceptance are ~81% of days; strict trend days just 2.6%) and a walk-forward morning classifier that reaches 73.6% five-class accuracy by noon at a 0.60 confidence threshold, versus a 41.9% baseline. Logged as a SHADOW decision-support candidate, no entries tested, 0 contracts. Library now holds 57 candidate studies across NQ and GC. Still pending founder approval.
2026-08-08	Manifest 0.15.0	Added the NQ Price Structure Census Wave 2: the level-break lifecycle (27,333 first breaks, 180 preregistered hypotheses). It banked the wave's strongest new prior, a rejected opening-range break warns of a two-sided day (+9 to +23 pp opposite-boundary odds), plus the rebreak path prior (72-81% re-attack within six bars) and a candid trade-test that failed a deep OR-acceptance long once costs and yearly stability were applied. Capital verdict NO-GO, 0 contracts. Library now holds 54 candidate studies across NQ and GC. Still pending founder approval.
2026-08-08	Manifest 0.14.0	Added the NQ Price Structure Census (Wave 1): a 170-hypothesis preregistered census of popular candlestick and named price-structure patterns on the certified 2010-2026 panel (3,890 RTH days). The honest headline: not one single-bar pattern family, hammer, engulfing, pin, outside bar, impulse, swing failure, or close-location, produced a stable directional edge out of sample. Two non-directional priors survived and are published as studies (acceptance-state persistence; expansion→compression re-expansion), alongside the inside-bar myth-buster and a candle-anatomy baseline. Capital verdict NO-GO, 0 contracts. Library now holds 49 candidate studies across NQ and GC. Still pending founder approval.
2026-08-07	Manifest 0.13.0	Added Queued Research Wave 001: 12 preregistered NQ studies from the certified 2010-2026 minute panel (4,011 RTH quality-filtered days), three with full pages. The wave banked strong structural priors on opening-range acceptance, prior-day level failure, correct-direction path risk, initial-balance maturity, and trend occupancy, and it published honest negatives that ruled the Monday gap-fill myth a no-go, the opening-engulfing continuation, and the near-high-close edge. Capital verdict was NO-GO, 0 contracts. Library now holds 44 candidate studies across NQ and GC. Still pending founder approval.
2026-08-06	Manifest 0.12.0	Added GC (gold) as the second instrument: 6 certified GC Globex studies from the GC Globex Minute study (2010-10-14 to 2026-03-06, 2,882 quality-filtered days), three with full pages. Same honest pattern: strong structure, 0 tradeable overnight edges. Library now holds 32 candidate studies across NQ and GC. Still pending founder approval.
2026-08-06	Manifest 0.11.0	Added 6 certified NQ Globex minute studies (2014-10-13 to 2026-03-06, 2,229 strict days), three with full pages, including the 0-of-5,423 negative result. Library reached 26 candidate studies.
2026-08-06	Manifest 0.10.0	Added 9 certified volatility studies from the NQ Volatility Mechanics study (2010-06-07 to 2026-03-06, 2,155 strict days), three with full pages. Library reached 20 candidate studies.
2026-08-06	Manifest 0.9.0	Initial internal preview: 11 candidate studies drafted from certified Session Study aggregates (2016-01-03 to 2026-06-02). All candidates pending founder approval.

Dataset windows

Window	Status	Note
2016-01-03 to 2026-06-02	Certified (Session Study)	Hash-verified back-adjusted panel; 10.4 years; powers the session-structure, gap, opening-range, and range studies.
2010-06-07 to 2026-03-06	Certified (Volatility Mechanics)	Hash-verified 1-minute archive, 5.3M bars; 15.7 years, 2,155 strict days; powers the volatility, session-volatility, and clock studies.
2014-10-13 to 2026-03-06	Certified (NQ Globex Minute)	Hash-verified 1-minute archive, 2,229 strict days, 2.07M Globex minutes; powers the NQ overnight studies.
2010-10-14 to 2026-03-06	Certified (GC Globex Minute)	Hash-verified GC 1-minute archive, 2,882 quality-filtered days; A/B grading clears the known GC calendar-spread contamination; powers the gold overnight studies.
2014-10-13 to 2026-03-06	Certified (4-Hour Candle Structure Census W12)	Preregistered 4-hour census on 2,222 trading days (11,039 4-hour bars, 18:00 ET anchor; 71 shortened blocks excluded, no interpolation), with BH-FDR discovery control, current-state matched controls, discovery/validation/holdout eras, and a walk-forward model ablation. Powers the ten Wave-12 studies. 0 directional FDR hits; volatility-state and inside-bar range effects retained; model gate missed; router unchanged; 0 contracts.
2010-06-07 to 2026-03-06	Certified (Multi-Bar Sequence Census W11)	Preregistered sequence census on 3,883 exact RTH days (302,874 5-minute bars, 100,958 15-minute), with BH-FDR discovery control, current-state matched controls, discovery/validation/holdout eras, 17-year annual stability, and a walk-forward model ablation. Powers the expansion-compression, multi-bar-direction, compression-run and sequence-model studies. One confirmed range prior (size, not direction); directional families and the small-body run failed; model gate missed; router unchanged; 0 contracts.
2010-06-07 to 2026-03-06	Certified (Swing Structure Census W10)	Causal swing reconstruction on 3,883 exact RTH days (28,952 5-minute swing events), with pivot confirmation (k=2), an independent 1,000-event path audit, equal-distance geometry, and a walk-forward model ablation. Powers the market-structure-path, swing-durability, failed-swing and swing-model studies. 0 FDR discoveries; model gate failed; router unchanged; 0 contracts.
2010-06-07 to 2026-03-06	Certified (Prior-Day x Overnight Confluence W9)	Exact prior-Globex+RTH reconstruction on 1,694 fail-closed days (Discovery 128 / Validation 627 / Holdout 939), with a six-state overnight-vs-prior-day taxonomy, matched-control and equal-distance hardening, and a Router-v1 challenger. Powers the confluence-model, location-map, accepted-outside and sweep-reject studies. Router unchanged; challenger 0 of 6 gates; 0 contracts.
2010-06-07 to 2026-03-06	Certified (Multi-Session Handoff W8)	Exact-complete overnight (Asia/London/pre-NY/Globex) reconstruction and its RTH handoff on 1,694-1,810 exact overnight+RTH days, with no interpolated overnight minutes. Powers the session-of-extreme, boundary-survival, London re-entry, and overnight-model studies. Router unchanged; overnight structure added 0 of 6 model gates; 0 contracts.
2010-06-07 to 2026-03-06	Certified (Frozen-Box Recapture W7)	Same certified 3,883-session panel; the midpoint-recapture sequence measured across seven frozen box families (5,345 events, 1,811 matched pairs) with distance-neutral barriers and matched generic-cross controls. Powers the geometry-falsification study and the restatement of the Wave-6 recapture finding. General law NO-GO; descriptive location statistic banked, 0 contracts.
2010-06-07 to 2026-03-06	Certified (Auction Migration & Balance W6)	Same certified 3,883-session panel; a strict mature-balance state (205 events) matched to same-clock, range-percentile compression controls, plus midpoint-recapture ordering, migration-direction, volume/range filters and a fade-trade hardening. Powers the balance-vs-compression and midpoint-recapture studies. Research pass; midpoint recapture banked as a rotation state, no trade, 0 contracts.
2010-06-07 to 2026-03-06	Certified (Boundary Extension & Path Quality W5)	Same certified 3,883-session panel; a preregistered boundary-chase rule (next-5m-open entry, 25%-of-range target and stop) tested across 10:30/11:00/12:00 and discovery/validation/holdout eras, with path-ordering (MAE/MFE), cost stress and a 96-cell target/stop grid. Powers the path-quality and trade-test studies. Verdict NO-GO; 0 gate passes, 0 contracts.
2010-06-07 to 2026-03-06	Certified (Intraday Auction Development W4)	Exact-minute HOD/LOD and range-development reconstruction on 3,883 fully observed RTH sessions (7 early sessions fail-closed excluded for missing raw minutes). Opening-range lifecycle fields matched the frozen Wave-3 ledger exactly (12 fields x 11,649 rows). Powers the auction-development, extension, and extreme-survival studies. Research/habitat-prior pass; capital verdict NO-GO, 0 contracts.
2010-06-07 to 2026-03-06	Certified (Auction Day-Type W3)	Same certified 3,890-day NQ panel; five-class day-type labels independently re-implemented (400/400 match) and a walk-forward classifier over 8 annual folds. Powers the day-type taxonomy and classification-clock studies. Shadow decision-support only; capital verdict NO-GO.
2010-06-07 to 2026-03-06	Certified (Price Structure Census W2)	Hash-verified NQ 1-minute panel (source 4aa066…a2a2d1); 27,333 first breaks, 9,718 retests, 8,488 rebreaks; 180 preregistered hypotheses under BH-FDR. Powers the level-break lifecycle, two-sided-auction, rebreak, and OR-acceptance trade-test studies. Capital verdict NO-GO.
2010-06-07 to 2026-03-06	Certified (Price Structure Census)	Hash-verified NQ 1-minute panel (source 4aa066…a2a2d1), 3,890 RTH full-quality days; 170 preregistered structural hypotheses under BH-FDR. Powers the candlestick/price-structure census, the inside-bar myth-buster, acceptance persistence, expansion→compression, and the candle-anatomy baseline. Capital verdict NO-GO.
2010-06-07 to 2026-03-06	Certified (Queued Research Wave)	Hash-verified NQ 1-minute continuous panel, 4,011 RTH quality-filtered days; powers the Wave 001 studies on opening-range acceptance, prior-day level failure, path risk, initial balance, pullback depth, and trend occupancy. Twelve questions frozen before results; capital verdict NO-GO.

Restatements

One recorded restatement so far. The Wave-6 midpoint-recapture finding was restated in Wave 7: its original causal (rotation-edge) interpretation was superseded after a geometry-falsification test showed the effect was a distance artifact. The corrected study carries a restatement banner and links the prior figure and the cause on its own page. When any published figure changes, because of a data correction, definition fix, or window refresh, the old figure, new figure, and cause are recorded here and on the study page.

How NQ overnight gaps behaved: fill rates by gap size

Most NQ opening gaps were small, and small gaps usually filled, but gaps of at least a quarter of the prior day's range filled during regular hours only 42.4% of the time, and more often kept going than came back.

Grade BEstablished with caveat, the pooled fill rate is stable, but it conceals a strong gap-size effect that reverses the practical conclusion for large gaps.

Study HTA-NQ-GAP-001 · Certified dataset 2016-01-03 → 2026-06-02 (10.4 years) · 2,643 eligible sessions · Wilson 95% intervals · multiple-testing controlled (BH-FDR, q = 0.05) · publication v1.0 · updated 2026-08-06 · times shown are U.S. Eastern Time (ET)

The research question

After NQ opened away from the prior session's regular-hours close, how often did price trade back to that close (“fill the gap”), and how did the size of the gap, measured against the prior day's range, change that rate?

Why this question matters

“Gaps usually fill” is one of the most repeated claims in trading education. If it is true only for certain gap sizes, then traders who apply it uniformly are preparing with a statistic that does not describe the situation in front of them. Knowing the size-conditional rates helps a trader frame the morning realistically instead of anchoring on a folk statistic.

What was tested

For every eligible session, the overnight gap was measured as the distance between the new session's open and the prior regular-hours close, normalized by the prior regular-hours range. A gap counted as “filled” when price traded back to the prior close. Fill rates were measured cumulatively at the end of each successive session window (Asia, London, pre-NY, regular hours), and continuation was measured as the close finishing beyond the open in the gap direction. Rates were then computed within gap-size buckets.

Eligible sample and period

2,643 eligible sessions from 2016-01-03 through 2026-06-02. The median gap was small. 0.046× the prior regular-hours range, which is itself a finding: most sessions open near the prior close. Larger-gap buckets shrink quickly (642 sessions at ≥0.10×, 198 at ≥0.25×, 73 at ≥0.50×) and are labeled with their sample class below.

Main findings

Fill by end of regular hours, by minimum gap size (share of the prior regular-hours range). Whiskers show Wilson 95% intervals. The chart's underlying values are exactly the aggregates shown. Nothing more. Buckets are cumulative (≥), so larger buckets are subsets.

Measure	Rate	95% interval	Sample	Class
All gaps. Filled by end of regular hours	61.1%	[59.2%, 62.9%]	2,643	OK
All gaps. Filled during overnight (by end of London)	68.0%	[66.2%, 69.7%]	2,643	OK
All gaps. Continued through the regular-hours close	53.9%	[52.0%, 55.8%]	2,643	OK · FDR-sig
Gaps ≥ 0.10×. Filled by end of regular hours	54.8%	[51.0%, 58.6%]	642	OK
Gaps ≥ 0.10×. Continued through the close	61.2%	[57.4%, 64.9%]	642	OK · FDR-sig
Gaps ≥ 0.25×. Filled by end of regular hours	42.4%	[35.7%, 49.4%]	198	Thin
Gaps ≥ 0.25×. Continued through the close	61.1%	[54.2%, 67.6%]	198	Thin · FDR-sig
Gaps ≥ 0.50×. Filled by end of regular hours	34.2%	[24.4%, 45.7%]	73	Sparse

Gap direction itself was balanced: 51.2% of gaps were upward [49.2%, 53.1%], not significantly different from a coin flip. The median gap was 0.046× the prior regular-hours range.

Across all sizes pooled, 61.1% of gaps were filled during regular hours (n=2,643). The size buckets above show how sharply this splits by gap size.

Whether NQ gapped up or down was essentially even (51.2% up, not distinguishable from a coin flip), the informative variable was gap size, not direction. Certified from the same session set.

Conditions where the pattern strengthened

The smaller the gap, the stronger the fill tendency: sessions below the 0.10× threshold (roughly three-quarters of all sessions) drove the pooled fill rate above 60%. Much of the filling happened before New York hours at all. 68.0% of all gaps had already been filled by the end of the London session, meaning the overnight tape did a large share of the work.

Conditions where the pattern weakened or reversed

Size reversed the conclusion. At ≥0.25× the prior range, filling (42.4%) was less common than continuing through the close (61.1%). At ≥0.50×, fills dropped to about a third, though that bucket holds only 73 sessions and its interval is wide [24.4%, 45.7%]. Gaps of a full prior-day range or more occurred just 14 times in a decade; no claim is published for that bucket.

Stability and sensitivity

The large-gap continuation tendency survived multiple-testing control (BH-FDR, q = 0.05) on all adequately-sampled buckets. The finding is structural (about distance and mean-reversion to a reference price) rather than directional, which historically made this family of findings more robust across market regimes than direction-based ones. A dedicated sub-period re-run (2016–2020 vs 2021–2026) is scheduled before the grade can be considered for A.

Using this research responsibly

Use this research for

Pre-market preparation: size today's gap against recent range before forming expectations
Recognizing that a large gap historically behaved differently from a small one
Setting realistic expectations about how often fills happened, and when (often overnight)
Reviewing after the session whether the day matched its historical category
Avoiding overreaction when a small gap fills. Historically the common outcome

Do not use this research for

Automatic gap-fade entries or exits, this study contains no entry, stop, or target logic
Assuming any individual gap must fill or must continue
Predicting today from a historical frequency, the sample showed tendencies, not certainties
Replacing risk management or an independently validated strategy

Net Alpha Pro members get this study's full research packet, the gap-size scenario matrix, and a preparation checklist that connects it to a daily routine. See how Net Alpha Pro works →

Limitations

Large-gap buckets are thin (n = 198 at ≥0.25×) to sparse (n = 73 at ≥0.50×); their intervals are wide and marked accordingly.
The panel is a back-adjusted continuous series. Chosen deliberately so contract rolls do not appear as false gaps. Absolute price levels are therefore continuity-preserving, not historical tick prices; ratios and fill events are unaffected.
“Filled” means the prior close was traded, not that a fade trade would have been profitable. No costs, execution, or path are modeled.
The 2016–2026 window contains a predominantly rising market; while gap-fill mechanics are structural, the sub-period re-run remains scheduled.

Methodology summary

Sessions follow fixed CME/ET boundaries keyed to the CME trading date. Gap = open minus prior regular-hours close, normalized by the prior regular-hours range. Every rate carries its numerator, denominator, Wilson 95% interval, and Benjamini-Hochberg FDR flag; buckets below minimum sample thresholds are labeled Thin/Sparse and excluded from headline claims. Full methodology →

Advanced detail: definitions and thresholds

Eligibility requires a complete prior regular-hours session and a valid session open. Fill checkpoints are evaluated cumulatively at the end of the Asia, London, pre-NY, and regular-hours windows. Sample classes: OK ≥ 384, Thin ≥ 100, Sparse ≥ 30, Insufficient < 30 (no publication). Continuation = regular-hours close beyond the open in the gap direction. Multiple-testing control: BH-FDR at q = 0.05 across the study family.

Certification
Study ID: HTA-NQ-GAP-001 · Publication: v0.9-draft · Specification: SEC7 (frozen, session-study family) · Dataset: certified canon, 2016-01-03 → 2026-06-02, hash-verified · Eligible sessions: 2,643 · Interval method: Wilson 95% · Multiple testing: BH-FDR q = 0.05 · Exploratory/confirmatory: confirmatory within frozen family · Restatement status: none · QA: automated gates passed (internal record) · Rights preflight: aggregate-only, reconstruction test passed · Final approver: Reid Nabua (co-founder), 2026-08-13.

Related studies

NET ALPHA PRO · MEMBERS

Turn this study into a routine

The HTA Research Library explains what historically occurred under defined conditions. Net Alpha Pro helps members turn those lessons into a repeatable preparation, review, and risk-management process, including this study's full packet, scenario matrices, and daily checklists.

See how Net Alpha Pro turns HTA research into a structured daily preparation and review process

Net Alpha Pro is educational. It is not a data terminal, live market-data display, signal service, or source of guaranteed outcomes.

Measure	Rate	95% interval	Sample	Class
First RTH break of London range → returned inside ≤ 5 min	67.2%	[65.2%, 69.2%]	2,088	OK
… returned inside ≤ 15 min	77.7%	[75.8%, 79.4%]	2,088	OK
… returned inside ≤ 30 min	83.0%	[81.3%, 84.5%]	2,088	OK
RTH stayed entirely inside the London range (no break)	21.3%	[19.8%, 22.9%]	2,654	OK
RTH broke both sides of the London range	5.4%	[4.6%, 6.3%]	2,654	OK

Measure	Rate	95% interval	Sample	Class
Bottom-quartile RTH range → next RTH expands	63.4%	[59.6%, 66.9%]	666	OK · FDR-sig
Below-rolling-median RTH range → next RTH expands	67.6%	[65.0%, 70.0%]	1,329	OK · FDR-sig
Bottom-quartile Asia range → next Asia expands	61.0%	[57.3%, 64.7%]	670	OK · FDR-sig
Bottom-quartile London range → next London expands	60.3%	[56.5%, 63.9%]	667	OK · FDR-sig
Bottom-quartile full-overnight range → next expands	60.0%	[56.3%, 63.7%]	673	OK · FDR-sig

Measure	Rate	95% interval	Sample	Class
Globex-open range broken by Asia	93.1%	median break time ~20:21 ET	2,229	OK
… break accepted at the session close	52.1%	of breaks	2,076	OK
… returned inside within 30 minutes	74.5%	of breaks	2,076	OK
Asia range broken by London	91.9%	accepted 51.5% / back inside 30m 76.6%	2,229	OK
London range broken by Pre-NY	59.6%	accepted 55.2% / back inside 30m 75.3%	2,229	OK

Measure	Rate	95% interval	Sample	Class
09:30 ET opening block range vs a typical block	4.69x	conservative across periods	2,155	OK
09:30 ET opening block volume vs a typical block	30.36x	conservative across periods	2,155	OK
09:35 ET block range vs a typical block	4.03x		2,155	OK
Closing approach (15:50-16:00) range vs typical	2.73x		2,155	OK
Pattern stability	stable	held in discovery, validation, and holdout	2,155	OK

Hawai'i Trading Academy | Trading Coaching & Education

Know what NQ usually does.

How to actually use this

Use this research for

Do not use this research for

The library shows you the history. Pro helps you build the habit.

How this research is built

How HTA research is built

1 · Certified data

2 · Frozen definitions

3 · Uncertainty by default

4 · Evidence grades

5 · The publication boundary

6 · Reviewed release

Version history

Library versions

Dataset windows

Restatements

How NQ overnight gaps behaved: fill rates by gap size

The research question

Why this question matters

What was tested

Eligible sample and period

Main findings

Conditions where the pattern strengthened

Conditions where the pattern weakened or reversed

Stability and sensitivity

Using this research responsibly

Use this research for

Do not use this research for

Limitations

Methodology summary

Related studies

Turn this study into a routine

First breaks of overnight ranges usually came back

The research question

Why this question matters

What was tested

Eligible sample and period

Main findings

Conditions where the pattern strengthened

Conditions where the pattern weakened or reversed

Stability and sensitivity

Using this research responsibly

Use this research for

Do not use this research for

Limitations

Methodology summary

Related studies

Turn this study into a routine

Compression preceded expansion

The research question

Why this question matters

What was tested

Eligible sample and period

Main findings

Conditions where the pattern strengthened

Conditions where the pattern weakened or reversed

Stability and sensitivity

Using this research responsibly

Use this research for

Do not use this research for

Limitations

Methodology summary

Related studies

Turn this study into a routine

The volatility clock: when NQ actually moves

The research question

Why this question matters

What was tested

Eligible sample and period

Main findings

Conditions where the pattern strengthened

Conditions where the pattern weakened or reversed

Stability and sensitivity

Using this research responsibly

Use this research for

Do not use this research for

Limitations

Methodology summary