spb/anomaly-atlas Public License

1---2project: anomaly-atlas3document: F001 confidence4author: Simon-Pierre Boucher5contact: contact@spboucher.ai6data_source: hfmarketdata.io7created: 2026-08-128status: reviewed9---1011Level: 21213# Confidence — F001 (negative finding)1415**Level 2 — robust** (corrected & OOS & robust). Evidence per layer:1617* **Corrected**: the claim is made against the FULL searched universe (37218  signed rules), with BH-FDR + White RC + Hansen SPA (500 stationary19  bootstraps) + DSR; the 22-hypothesis budget was declared before scanning20  (research_gaps.md) and never expanded.21* **Artifact-adjusted**: scans ran net of measured nulls (taxonomy T1-T7:22  bounce/EDGE variance-consistent MA(1) null, both-fresh synchronization,23  permuted calendar); detectors passed 29 synthetic gate tests first.24* **Out-of-sample**: the 2016-2021 validation split, frozen before any scan25  and opened exactly once (expH), REPLICATES the negative: zero net26  survivors at full half-spread; the lone 0.25x survivor fails its27  pre-registered sub-period clause; ES->SPY flips sign; the daily reversal28  family collapses (-0.143 -> +0.023 VR30 excess).29* **Robust**: holds across 4 timeframes, 2 train sub-periods, 2 validation30  sub-periods, liquidity buckets, futures splice variants, and both close31  conventions where applicable.3233**Why not Level 3**: Level 3 is reserved for confirmation on the sealed34holdout (2022->), which this project deliberately did not spend on an35already-settled negative. Costs are estimated (EDGE), not quoted — the36cost-real qualifier is model-based.37