What the study supports
LF/HF sequencing may materially influence later directional expansion probabilities. It is most useful as an expansion-regime classifier.
Intraday Path-Dependency Framework
09:30–10:10 LF/HF No-Quartile Probability Framework
A pure conditional-probability classification model evaluating whether early-session high/low sequencing contains information about later intraday expansion. This is not an execution system and does not define entries, exits, stops, targets, or position sizing.
Opens the LF/HF no-quartile Python script, not the IB or late-day script.
Result Snapshot
83.39%
P(post-10:10 formation high break | LF), sample N = 578.
Executive Finding
The 09:30–10:10 LF/HF framework measures whether the order of early-session high and low formation changes the later intraday probability distribution. LF sessions showed elevated upside expansion probabilities, while HF sessions showed elevated downside expansion probabilities. The strongest asymmetry appeared in post-formation directional extension rather than terminal close location.
Core Definitions
| State | Definition | Interpretation |
|---|---|---|
| LF | First occurrence of the 09:30–10:10 low occurs before the first occurrence of the 09:30–10:10 high. | Low forms first; upside expansion state. |
| HF | First occurrence of the 09:30–10:10 high occurs before the first occurrence of the 09:30–10:10 low. | High forms first; downside expansion state. |
| SAME_BAR | Both first extremes occur on the same row/bar. | Ambiguous sequencing; separated from LF/HF. |
Empirical Statistical Results
Conditional probability outputs.
| Condition | Outcome | Sample Size | Probability |
|---|---|---|---|
| LF | Post-10:10 formation high break | 578 | 83.39% |
| LF | Post-10:30 IB high break | 578 | 74.91% |
| LF | Close above IB midpoint | 578 | 65.22% |
| HF | Post-10:10 formation low break | 577 | 79.38% |
| HF | Post-10:30 IB low break | 577 | 69.15% |
| HF | Close below IB midpoint | 577 | 54.94% |
Path Dependency
Same range, different sequence, different distribution.
Two sessions may print identical highs, lows, and ranges while forming those extremes in different orders. This framework isolates whether that sequencing itself contains conditional information. The model intentionally excludes quartile filters, prior-day location filters, VWAP filters, overnight inventory filters, volatility regime filters, and macro-event filters to preserve a cleaner base-rate framework.
No EntriesNo StopsNo TargetsNo PnL LogicProbability Classification Only
Strategic Interpretation
What it does not prove
The framework does not establish a standalone trading edge, optimal execution, or future persistence. Rolling and out-of-sample validation are still required.
Use Case
Use as an intraday regime label that informs risk expectations after the 09:30–10:10 formation window.
Next Research Layer
Test conditional behavior by volatility regime, prior-day location, IB range size, and rolling stability without contaminating the no-quartile base model.