Day-type base rates
Every 5-minute session in the look-alike corpus — 186,211 sessions across 117 tickers, Oct 2019 → present — labeled mechanically by the same Brooks day-type classifier that drives the Day Tape rail (a parity-locked port of scripts/daytape.py). These are base rates: how often each tape shape actually occurs, and how the opening gap shifts the mix. They describe the past — they are not a forecast.
Overall mix
| day type | sessions | share | closed above open | median open→close |
|---|---|---|---|---|
| TD+ bull trend day | 32,257 | 17.3% | 100% | 1.70% |
| TD- bear trend day | 27,693 | 14.9% | 0% | -1.85% |
| TFO+ bull trend-from-the-open | 26,739 | 14.4% | 100% | 0.87% |
| TFO- bear trend-from-the-open | 26,446 | 14.2% | 0% | -0.92% |
| TR trading range day | 32,890 | 17.7% | 48% | 0.00% |
| TTR+ bull trending trading range | 20,260 | 10.9% | 99% | 0.36% |
| TTR- bear trending trading range | 19,926 | 10.7% | 0% | -0.38% |
“Closed above open” is definitionally high for bull shapes and low for bear shapes — the label encodes the direction. The interesting column is the share: two-sided tape (TR + TTR) is ~39% of all sessions — most days are not trend days.
Day character by type — the physics numbers
| day type | reversion half-life (med, IQR) | no measurable reversion | laminarity (med) | perm. entropy (med) | Hurst (med) |
|---|---|---|---|---|---|
| TD+ bull trend day | 47m(28–83m) | 7% | 0.839 | 0.88 | 0.54 |
| TD- bear trend day | 49m(29–87m) | 9% | 0.837 | 0.891 | 0.534 |
| TFO+ bull trend-from-the-open | 32m(20–54m) | 1% | 0.812 | 0.892 | 0.54 |
| TFO- bear trend-from-the-open | 32m(20–54m) | 1% | 0.813 | 0.897 | 0.542 |
| TR trading range day | 29m(19–45m) | 0% | 0.799 | 0.898 | 0.544 |
| TTR+ bull trending trading range | 39m(24–66m) | 3% | 0.814 | 0.894 | 0.543 |
| TTR- bear trending trading range | 39m(24–67m) | 3% | 0.815 | 0.896 | 0.537 |
Dynamical-systems measurements over 180,591 labeled sessions (117 tickers): how many minutes a stretch away from the mean takes to decay (Ornstein–Uhlenbeck half-life), how often a day trends so cleanly reversion is unmeasurable, how much time the tape spends stuck in one neighborhood (RQA laminarity), how orderly the bar sequence is (permutation entropy), and trend persistence (DFA Hurst). The half-life column is the separator — trend days’ stretches decay slowest and are the only types where reversion regularly fails to register; ranges snap back fastest. Hurst, honestly, barely separates the types at this bar size — a measured null, reported rather than hidden. Character describes a finished session; it forecasts nothing.
Mix by asset class
| asset class | sessions | TD+ | TD- | TFO+ | TFO- | TR | TTR+ | TTR- |
|---|---|---|---|---|---|---|---|---|
| crypto | 28,716 | 14% | 13% | 11% | 12% | 22% | 14% | 14% |
| equity | 135,920 | 18% | 15% | 15% | 15% | 17% | 10% | 10% |
| fx | 21,575 | 17% | 15% | 13% | 13% | 19% | 12% | 12% |
Opening gap × day type — equities only
Gap measured open vs prior session close, in basis points. FX/crypto excluded (continuous markets barely gap). Row percentages: of the sessions that opened with this gap, how the day resolved.
| opening gap | sessions | TD+ | TD- | TFO+ | TFO- | TR | TTR+ | TTR- | closed above open |
|---|---|---|---|---|---|---|---|---|---|
| big gap-down (≤ −75 bp) | 24,870 | 18% | 15% | 16% | 15% | 17% | 10% | 10% | 52% |
| gap-down (−75 to −25 bp) | 21,335 | 18% | 15% | 16% | 14% | 17% | 11% | 10% | 52% |
| flat open (±25 bp) | 37,388 | 18% | 15% | 15% | 15% | 17% | 10% | 10% | 52% |
| gap-up (25 to 75 bp) | 25,561 | 18% | 16% | 15% | 15% | 17% | 10% | 10% | 50% |
| big gap-up (≥ 75 bp) | 26,667 | 19% | 16% | 15% | 15% | 16% | 9% | 10% | 51% |
Teal cells run ≥1.25× their all-corpus baseline — hover for the comparison.
The six canonical day shapes — canon v1, frozen 2026-07-02
The site's canonical day-type vocabulary. All 186,135 sessions clustered by shape (the matcher's own DTW geometry), k chosen by sweep, exemplars frozen — every past and future day is labeled by its nearest frozen exemplar, so these labels never shift as the corpus grows. Each card is a real session. Shape-space is a continuum (the sweep showed no “natural” cluster count), so this vocabulary is deliberately coarse: families, with the finer grain below and sub-day structure handled by the pattern layer.
- the high of the day is in before the first hour ends 90% vs 45% all days
- closes below its open 90% vs 49% all days
- the low of the day comes after the first hour 93% vs 54% all days
- closes in the bottom quarter of its range 53% vs 25% all days
- almost never closes in the top quarter of its range 2% vs 29% all days
- deep pullbacks through half the range are less common here 56% vs 69% all days
- the low of the day is in before the first hour ends 85% vs 46% all days
- closes above its open 87% vs 51% all days
- the high of the day comes after the first hour 89% vs 55% all days
- closes in the top quarter of its range 63% vs 29% all days
- almost never closes in the bottom quarter of its range 2% vs 25% all days
- the afternoon rarely fights the morning 35% vs 50% all days
- almost never closes in the bottom quarter of its range 4% vs 25% all days
- closes in the top quarter of its range 50% vs 29% all days
- at least one pullback retraces more than half the range 86% vs 69% all days
- closes above its open 65% vs 51% all days
- almost never closes in the top quarter of its range 5% vs 29% all days
- closes in the bottom quarter of its range 48% vs 25% all days
- closes below its open 68% vs 49% all days
- at least one pullback retraces more than half the range 83% vs 69% all days
- the low of the day comes after the first hour 67% vs 54% all days
- the low of the day is in before the first hour ends 83% vs 46% all days
- the high of the day comes after the first hour 84% vs 55% all days
- closes above its open 72% vs 51% all days
- the afternoon runs against the morning 70% vs 50% all days
- closes in the middle half of its range 65% vs 46% all days
- almost never closes in the bottom quarter of its range 12% vs 25% all days
- ranges wider than usual for the symbol (≥ 1.25× its median) 100% vs 32% all days
- closes in the bottom quarter of its range 84% vs 25% all days
- rarely closes mid-range 14% vs 46% all days
- at least one pullback retraces more than half the range 100% vs 69% all days
- almost never closes in the top quarter of its range 2% vs 29% all days
- closes below its open 66% vs 49% all days
Do shapes have memory? No — and that's worth knowing
Across 186,030consecutive-session transitions, tomorrow's shape mix after ANY shape is statistically indistinguishable from the base mix — the strongest lift anywhere in the matrix is 1.03×. “Trend day begets trend day” does not survive measurement: at the day-shape level, the market deals each session fresh. Rows are yesterday's shape; cells are today's mix (lift vs base).
| after ↓ | bear-drive | bull-drive | bull-grind | bear-fade | bull-fade |
|---|---|---|---|---|---|
| bear-drive | 27.5% (0.98×) | 25.4% (1.01×) | 18.2% (0.99×) | 15.1% (1×) | 13.9% (1.03×) |
| bull-drive | 28.4% (1.02×) | 24.8% (0.98×) | 18.5% (1.01×) | 14.7% (0.98×) | 13.7% (1.01×) |
| bull-grind | 27.9% (1×) | 25.5% (1.01×) | 17.8% (0.97×) | 15.4% (1.02×) | 13.4% (0.99×) |
| bear-fade | 27.7% (0.99×) | 25.5% (1.01×) | 18.8% (1.03×) | 15% (1×) | 13% (0.96×) |
| bull-fade | 28.4% (1.02×) | 25% (0.99×) | 18.3% (1×) | 15% (1×) | 13.2% (0.98×) |
Finer grain — the same clustering at k=16
Everything above ports Al Brooks' hand-written rules. This goes the other way: all 186,135 sessions clustered purely by shape (the same z-normed multi-channel DTW geometry the Look-Alikes matcher scores), k = 16. Each card is one empirical day kind: its real exemplar session, population share, how those days finished, and which Brooks labels its members carry. Descriptions are generated from the cluster's own aggregates — the data named these, not us.