Inside Day
compression precedes expansion — but not in a predictable directionAn inside day is one where the session’s high is at or below the prior day’s high AND the low is at or above the prior day’s low — the entire bar fits inside the previous bar’s range. Per Brooks, this represents balance: neither buyers nor sellers were willing to push beyond yesterday’s extremes. The market is compressing. This study asks: what happens next? Across 1,762 inside days on 15 liquid US large-caps and ETFs (Jan 2022 – May 2026), the answer is unambiguous on the magnitude question and a coin flip on the direction question.
Next session’s range exceeds inside day’s range on nearly 3 in 4 inside days.
On any random day (non-inside), next day’s range exceeds prior day’s range less than half the time.
1.35× range expansion in the next session (0.91 ATR vs 0.67 ATR inside day mean).
Inside-day vs. baseline expansion rate
50% reference = coin-flip. Difference = +26.6 pp (CI95 entirely above zero). 15 tickers · 2022–2026. n=5,000 bootstrap.
What IS predictable: magnitude
An inside day reliably signals that the next session will have a larger range than the inside day itself. The expansion rate is 72.9% across 1,762 events — 15/15 tickers show expansion rates between 69% and 77%, with the weakest name (AAPL at 69.0%) still 21 pp above its own baseline. The mean next-day range is 1.35×the inside day’s range in ATR units (0.91 vs 0.67 ATR). Compression → expansion is structural, not luck.
What is NOT predictable: direction
Whether the next open is above or below the inside day’s midpoint does NOT predict the close direction of the next session. Directional carry rate: 49.7% (CI95 [47.3%, 52.0%] — straddles 50%). This is statistically indistinguishable from a coin flip. Compression resolves into expansion, but the market keeps the direction coin hidden until the session unfolds. You know THAT tomorrow will move; you don’t know WHICH WAY.
Per-ticker detail
Sorted by inside-day expansion rate. All 15 tickers show expansion rates between 69–77%. The directional carry column shows noise throughout — no ticker achieves CI-clear directional edge.
| Ticker | N inside | Expand % | Baseline | Edge | Dir carry |
|---|---|---|---|---|---|
| AMZN | 115 | 77.4% | 46.9% | +30.5 pp | 53.0% |
| AMD | 135 | 77.0% | 46.1% | +30.9 pp | 49.6% |
| XLF | 106 | 76.4% | 46.0% | +30.4 pp | 50.0% |
| AVGO | 121 | 76.0% | 45.8% | +30.2 pp | 52.9% |
| SPY | 102 | 75.5% | 45.3% | +30.2 pp | 49.0% |
| XLK | 109 | 74.3% | 45.4% | +28.9 pp | 45.9% |
| GOOG | 132 | 74.2% | 46.0% | +28.2 pp | 44.7% |
| QQQ | 110 | 71.8% | 47.2% | +24.6 pp | 48.2% |
| NVDA | 128 | 71.1% | 46.7% | +24.4 pp | 50.0% |
| IWM | 106 | 70.8% | 46.2% | +24.6 pp | 40.6% |
| TSLA | 131 | 70.2% | 45.5% | +24.7 pp | 55.0% |
| META | 107 | 70.1% | 46.4% | +23.7 pp | 48.6% |
| MSFT | 129 | 69.8% | 45.8% | +24.0 pp | 53.5% |
| JPM | 118 | 69.5% | 47.0% | +22.5 pp | 49.2% |
| AAPL | 113 | 69.0% | 47.8% | +21.2 pp | 53.1% |
What this means for trading
- Inside days are range-expansion setups, not directional ones.The 72.9% expansion rate means that after an inside day, the next session is very likely to offer a larger range than the inside day itself. Position sizing, expected-move calculations, and options pricing can all be informed by this. The direction, however, must come from the next day’s own price action — not from the inside day’s midpoint.
- Wait for the break before trading direction.The 49.7% directional carry failure means pre-positioning based on midpoint location is coin-flip territory. On an inside day, do NOT bias your next-day trade direction based on the midpoint. Instead, watch for the first breakout above or below the inside day’s H/L — that break is the actual signal.
- Inside days flag tomorrow as a “potential trend day” candidate. A 1.35× average range expansion after inside days means these setups frequently generate directional sessions. Combined with the Day Types study, inside days feeding into trend days is the sequence to watch.
- The baseline gap is striking: range expansion is NOT the default.On a random day, the next day’s range only exceeds today’s range 46.3% of the time — range CONTRACTION is the slight default in this universe. Inside days flip this forcefully to 72.9%. The market does not normally expand; it normally contracts. Inside days break the pattern.
How this fits the other studies
- Gap Fill Rate: The 46.3% baseline range-expansion rate (slightly below 50%) echoes the gap-fill finding: at the daily scale, prices naturally tend to compress rather than expand from session to session. Just as small ATR-normalized gaps fill 75% of the time (mean reversion at the open), daily ranges tend to contract (reversion in daily volatility). Inside days are the extreme of this compression, and they resolve the other way — into expansion.
- First-Hour Fade:Inside days that break OUT of their range at the open of the next session are producing exactly the kind of first-hour extension that the First-Hour Fade targets. If an inside day is followed by a gap or strong opening drive, the first retest of that morning’s extreme is the fade opportunity — the same mechanism operating on the intraday timeframe, seeded by the prior day’s compression.
- Day Types: A session following an inside day is disproportionately likely to be a trend day or a breakout-and-close day — sessions where the Day Types classifier sees one clear directional move rather than a range day or reversal. Compression on Day N feeds expansion and directionality on Day N+1.
Pre-registered verdicts
H13(c)’s failure is the most useful result: it rules out a common trader intuition (“if the next day opens above the midpoint, it’s going up”) and confirms the Brooks framing — an inside bar is a two-sided setup. The compression tells you the magnitude; the market reveals the direction only after the fact.
Honest caveats
- “Expansion” is defined as next-day range > inside-day range — a binary outcome. The 1.35× mean ratio suggests the expansion is meaningful in size, but some expansions will be marginal (just barely larger). A minimum-expansion filter (e.g., next range > 1.5× inside range) would sharpen the signal at the cost of fewer events.
- Consecutive inside days are not excluded. A 2-day or 3-day inside pattern (each day inside the last) would naturally produce larger eventual expansion; the current study treats each day independently. A streak-length analysis is a natural extension.
- Universe is the same 15 mega/large-cap US names as the prior studies — all highly liquid. The effect may differ for small-caps or individual names with low float where inside days can form on low volume and carry less information.
- Daily bar OHLC is used. True intraday path matters for whether “expansion” is tradeable: the expansion might occur in the first 30 minutes (tradeable) or only in the final auction (not). An intraday path study would refine the finding.
Methodology
1,762 inside-day events + 14,498 baseline events on 15 names (2022-01-03 – 2026-05-30) via Massive/Polygon daily bars. Inside day: strict containment (today.H ≤ prev.H AND today.L ≥ prev.L). Expansion: next-day range (H-L) > inside-day range. ATR normalization uses Wilder ATR(14) strictly pre-signal. Directional carry: open vs inside-day midpoint → predicts close direction. Bootstrap CI (n=5,000, seed=17). Engine: scripts/ml/backtest_inside_day.py. Run 2026-06-15.
Related: Gap Fill Rate · First-Hour Fade · Day Types