Honesty doctrine. Every candidate anomaly is an artifact until proven otherwise; in-sample results are never findings; past statistical regularity does not imply future returns. This is research on statistical properties of market data — not investment advice, not a trading system.

Research documents

Research log

Append-only, auditable record of every question, experiment, result, and decision.

Data source profile

Phase 0.5 — what hfmarketdata.io actually provides, established empirically.

State of the art

Phase 2 — per-anomaly map with epistemic status: robust / decayed / disputed / likely-artifact.

Artifact taxonomy

The living catalogue of dataset artifacts that masquerade as anomalies (Q4).

Research gaps

Phase 3 — ≥20 testable hypotheses, each with a constructable artifact null.

Candidate ranking

Phase 4 — 10-axis scoring and selected prototype candidates.

Validation methodology

Phase 9 — the pre-registered protocol behind every atlas entry.

placeholder

Bibliography

Every consulted source with URL and access date.

Novelty check

Phase 11 — assume not novel until evidence says otherwise (written last).

placeholder

Phase 1 literature notes

calendar effects

Phase 1 literature notes — 51 lines.

8 sources

mean reversion leadlag

Phase 1 literature notes — 45 lines.

10 sources

microstructure artifacts

Phase 1 literature notes — 47 lines.

8 sources

multiple testing snooping

Phase 1 literature notes — 41 lines.

15 sources

timeseries methodology

Phase 1 literature notes — 45 lines.

6 sources

transaction costs

Phase 1 literature notes — 41 lines.

11 sources

The project charter itself is public: read the full research charter.