btcl

Gap Up, Gap Down, or In Range — What Happens Next? (12,390 Sessions Analyzed) Every trading day begins with a single print on the tape — the RTH session open. That first price relative to the previous session’s range sets the tone for everything that follows. But how exactly does opening type — gap up, gap down, or in-range — influence intraday behavior? We analyzed 12,390 trading days across four major futures contracts to answer these questions: How often does each open type occur, and is the distribution stable? Which previous-session levels are most likely to be touched based on opening type? Does opening within vs outside the Value Area change the day’s character? When price touches one Value Area boundary, how often does it reach the VPOC? How does gap size affect the probability of revisiting the prior range? The results reveal that opening type is not merely descriptive — it carries measurable statistical weight. In-range opens show 71-73% probability of touching the previous close, while a touch of the prior VAH or VAL leads to 81–84% probability of also reaching the VPOC. Data and Methodology This research covers minute-level RTH (09:30-16:00 ET) data for four CME futures contracts: #NQ (Nasdaq-100) — 3,116 days #ES (S&P 500) — 3,042 days #YM (Dow Jones) — 3,116 days #RTY (Russell 2000) — 3,116 days The dataset spans from February 2014 through March 2026 — over 12 years of market history covering multiple regimes including low-volatility bull markets, COVID-era extremes, and the 2022-2023 rate hiking cycle. Classification Definitions Open Type is determined by comparing the RTH open price to the previous session’s high and low: Gap Up — RTH open above previous session high Gap Down — RTH open below previous session low In Range — RTH open between previous session low and high (inclusive) Touch means the level fell within the current day’s RTH traded range (session low ≤ level ≤ session high). No specific sequence or timing is required — only that the level was traded through at some point during the session. Value Area sub-classification applies only to In Range days and uses the previous session’s TPO-derived Value Area High (VAH), Value Area Low (VAL), and Volume Point of Control (VPOC).

Limitations & Future Work No clean OOS holdout yet. All probabilities are full-sample. Walk-forward validation with rolling windows and a strict 2025–2026 holdout (train ≤ 2024) is the next step. News-day effects unmodeled. NFP, FOMC, CPI release days are included in the sample; their effects are diluted in the aggregate. A separate news-filtered analysis could surface event-specific cascade asymmetries. Day-of-week / gap-direction slices were dropped in this study due to weak theoretical basis and sample dilution (5x split for DoW). Subsequent work could re-examine if larger samples become available. Volume regime not yet conditioned. High-volatility vs low-volatility days likely produce different cascade speeds; pre/post-August 2024 regime shifts deserve separate inspection. Bilateral break ordering and second-side timing warrant their own paper. We have the per-session data captured but reserved a focused analysis for later publication.

Cross-Instrument Validation If the Asia cascade reflects a real liquidity-based mechanism, the same structure should appear in adjacent index futures with at most a uniform scaling factor. The marginal ladder confirms this: Marginal extension probabilities cluster within ~2pp across all four instruments — extraordinarily tight. The 10% and 50% ext levels match within 1.1pp; the 25% and 100% ext levels diverge by up to 1.9–2.2pp. The largest cross-instrument divergence is in the bilateral break rate, where YM and RTY trade ~3pp higher than NQ/ES. We attribute this to lower liquidity in YM and RTY contracts, which produces wider sweep behavior on average. The structural conclusion is unchanged: the Asia cascade is a property of CME index futures liquidity, not an instrument-specific anomaly.

Asia Range Cascade: A 12-Year Backtest on CME Index Futures 93–95% of sessions break the Asia range (20:00 ET → 02:00 ET) before 08:00 ET across all four CME index futures (n = 12,372). 96–97% of broken sessions revert back inside the Asia range — mean reversion is near-universal. The “failed break paradox”: when the first break extends less than 10% of Asia range (peak ext < 0.10), the cascade through retrace levels is stronger, not weaker — 88% reach 25% retrace, 86% the mid given 25%, and 98% break the opposite side once 100% retrace touches. Cohort × time-decay × stalled buckets produce ~84 distinct conditional probability surfaces per instrument — a state-aware framework rather than a single static stat. Cross-instrument marginal probabilities are within ±2pp NQ/ES/YM/RTY (76.4 vs 76.5% at 25% ext, etc.) — the structure is index-wide, not NQ-specific.

#NQ_F on Non-Farm Payrolls (NFP) — what 71 events show (Jun 2020 → today): Volatility: • ATR expands 3.3× post-release • Range vs normal day: 1.23× • Median delivery from 08:30 release: 60 pts in 30m, 67 pts in 1h, 142 pts in 2h, 202 pts by close Event Initial Balance (08:30 → 09:30 ET): • Range median: 135 pts • IB High touched: 70% • IB Low touched: 62% • Both extremes touched: 39% • Neither touched: 7% • First-break side: IBH 58% / IBL 42% • Median extension beyond IB: upside 79 pts / downside 118 pts Classic Initial Balance (09:30 → 10:30 ET): • Range median: 158 pts • IBH 63% · IBL 50% · Both 22% · Neither 9% Surprise → direction match: • 47% across 70 events • Statistically indistinguishable from a coin flip Direction split: • 52% LONG · 48% SHORT (near-balanced) Asymmetry: • Downside extension beyond Event IB Low (~118 pts median) is roughly 50% larger than upside extension beyond IB High (~79 pts) $NQ

Limitations Touch probability measures whether a level was traded through at any point during the session. It does not capture sequence, timing, or whether the touch represented a reversal or continuation. Value Area data requires TPO profiles, which may not be available for all trading days in the early portion of the dataset. Gap classification uses the previous RTH high/low, not the overnight high/low. Overnight context may provide additional nuance. All statistics are unconditional within their category. Further filtering by volatility regime, trend context, or market internals could reveal sub-patterns. Past statistical patterns do not guarantee future results. Market microstructure evolves over time. Key Takeaways 60% of sessions open in-range, making the previous close and VPOC the most frequently relevant reference levels. In-range opens touch the previous close 71-73% of the time — the highest single-level touch probability in unconditional analysis. The VAH/VAL → VPOC conditional chain is 81–84% across all instruments — the strongest and most stable pattern in 12 years of data. Value Area classification matters: opening in-value yields 82-85% VPOC touch; opening above/below value yields 71-80% nearest boundary touch. Gap size degrades all probabilities: small gaps behave like in-range days, while large gaps (>1%) render prior levels largely inaccessible. IB break direction shows no strong edge based on open type — individual break rates range from 53% to 67% across all conditions.

IB Break Direction Does the opening type predict which way the Initial Balance (first hour, 09:30-10:30 ET) will be broken? The result is clear: IB break direction shows no strong edge based on open type alone. Both sides are broken roughly 53-67% of the time regardless of how the session opened. Open type alone does not predict which direction.

Gap Up, Gap Down, or In Range — What Happens Next? (12,390 Sessions Analyzed) Every trading day begins with a single print on the tape — the RTH session open. That first price relative to the previous session’s range sets the tone for everything that follows. But how exactly does opening type — gap up, gap down, or in-range — influence intraday behavior? We analyzed 12,390 trading days across four major futures contracts to answer these questions: How often does each open type occur, and is the distribution stable? Which previous-session levels are most likely to be touched based on opening type? Does opening within vs outside the Value Area change the day’s character? When price touches one Value Area boundary, how often does it reach the VPOC? How does gap size affect the probability of revisiting the prior range? The results reveal that opening type is not merely descriptive — it carries measurable statistical weight. In-range opens show 71-73% probability of touching the previous close, while a touch of the prior VAH or VAL leads to 81–84% probability of also reaching the VPOC. Data and Methodology This research covers minute-level RTH (09:30-16:00 ET) data for four CME futures contracts: #NQ (Nasdaq-100) — 3,116 days #ES (S&P 500) — 3,042 days #YM (Dow Jones) — 3,116 days #RTY (Russell 2000) — 3,116 days The dataset spans from February 2014 through March 2026 — over 12 years of market history covering multiple regimes including low-volatility bull markets, COVID-era extremes, and the 2022-2023 rate hiking cycle. Classification Definitions Open Type is determined by comparing the RTH open price to the previous session’s high and low: Gap Up — RTH open above previous session high Gap Down — RTH open below previous session low In Range — RTH open between previous session low and high (inclusive) Touch means the level fell within the current day’s RTH traded range (session low ≤ level ≤ session high). No specific sequence or timing is required — only that the level was traded through at some point during the session. Value Area sub-classification applies only to In Range days and uses the previous session’s TPO-derived Value Area High (VAH), Value Area Low (VAL), and Volume Point of Control (VPOC).