SPB Git

spb/anomaly-atlas Public License

Systematic discovery & rigorous validation of statistical anomalies in open HF market data (hfmarketdata.io) — pre-registered, artifact-null-driven, fully reproducible. Live atlas: www.anomaly-atlas.io

quant-researchmarket-microstructureanomaliesreproducibilityduckdbapple-silicon
Python 61.4% JavaScript 28.7% CSS 8.6% Shell 0.7% Makefile 0.5%
1.3 KB
View source Raw Blame History

# project: anomaly-atlas document: F002 confidence author: Simon-Pierre Boucher contact: contact@spboucher.ai data_source: hfmarketdata.io created: 2026-08-12 status: reviewed

Level: 2

# Confidence — F002 (artifact demonstration)

Level 2 — robust. This is a finding-type C/D entry (artifact demonstration / methodological), published as a negative result about a would-be anomaly.

  • Corrected: the effect itself survives Hansen SPA — which is the demonstration: statistical correction cannot detect mechanism.
  • Artifact mechanism established: measured in expB (+0.065) and expD (+0.132) with print-synchronization applied — the lead persists at fresh_fraction 0.996 with artifact-share-from-print-gaps ~0, isolating CONTENT staleness (Fisher 1966) as the mechanism; the synthetic gate demonstrates the same signature on planted staleness.
  • Out-of-sample: replicates on the validation split (+0.080) — the artifact is structural, unlike every "alpha" in the same pipeline.
  • Robust: present in two independent windows plus validation, at both raw and synchronized treatments.

Why not Level 3: holdout untouched (sealed); magnitude varies with index-constituent liquidity regime, so the number (not the mechanism) should be re-measured per period.