Research
Validation: multivariate test
A preregistered out-of-sample test of whether price, volume, volatility regime and lagged context improve directional forecasts.
Historical similarity is not a prediction or trading signal. These are descriptions of what happened after comparable periods in the past.
2 hypotheses survived correction.
Of 216 preregistered hypotheses, 2 survived a 10% false-discovery-rate correction and 1 survived at 5%. The result is reported as research, not as a strategy.
Technical summary
| Measure | Value |
|---|---|
| Candidate combinations | 153 |
| Valid combinations | 108 |
| Preregistered hypotheses | 306 |
| Tested hypotheses | 216 |
| Discoveries at 10% FDR | 2 |
| Discoveries at 5% FDR | 1 |
| Holm-corrected discoveries | 1 |
Generated 2026-10-05 08:05 UTC · 2026-08-13.1 · raw JSON
How to read this
Every combination of market, pattern length and forward period is a separate test. Running many tests produces some low p-values by chance, so the result that matters is what survives a correction for multiple testing — that is the number reported above.