Experiments before conclusions.
Each experiment records its hypothesis, data contract, falsification criteria, maturity gate, and version lineage separately. Failed arms are not erased; they remain as frozen history.
H2_SPOOL_PIT_V3R2
Candidate-conditioned Korea → Binance lead-lag research with first-seen availability and point-in-time controls.
H2 PIT V2 Family
Historical consumer-first-seen coverage pilot. Preserved for diagnostics but not pooled into V3R2 confirmation.
Cross-Venue V1
Frozen after point-in-time and consumer-availability audit failures. Historical outcomes remain preserved.
Protocol B Rich-L2 V1
Historical feature/outcome arm retained with known source-event-age and same-second availability limitations.
Guard PASS / STALE. Review BLOCKED. Effect metrics remain blinded.
Latest verified snapshot: 207 fully matched non-quarantined decisions against a preregistered minimum of 300, with 4 quarantined decisions preserved outside the confirmatory cohort. Verified 2026-10-03 00:20 UTC.
Open H2 status →“It leads the entire market” and “it adds information within our candidate-conditioned sample” are different claims. MKTRC keeps them separate and never pools historically invalid or pilot arms into the active confirmatory arm.