iteration 02 · provenance correction

Agnostic = algorithm-invariant, not asset-agnostic

A provenance audit (git authorship + live Telegram pull) shows the "agnostic = works across multiple assets" reading is an operator-side drift, not Terry's.

Who wrote each definition

definition / artifact	author	evidence
Agnostic = algorithm-invariant	Terry	Telegram msg id=428 & id=429, 2026-04-22 (verbatim)
RASEF (regime-aware strategy eval)	Terry	git: terry@eonlabs.com 2026-04-30 — but it has NO cross-asset KS test
CH_FEATURE_BANLIST (orthogonality filter)	Terry	git: terry@eonlabs.com 2026-04-27
Agnostic = "works across assets" / cross-asset KS	operator	3-AXIS-DISCIPLINE-TEMPLATE.md git: MD NASIM; FRAMEWORK.md = operator-drafted, awaiting review
"RASEF two-prong KS" attribution	Misattribution	the KS test is not in RASEF

The drift timeline

2026-04-25/27 (id=1029/1067): applied correctly — the Hurst "manifest says DFA, code computes R/S" mismatch filed as an agnostic failure ("two correct implementations should converge, but manifest and code diverge"). ✅
~2026-05-06+ (id=2782/3638/4415): language drifts to "asset-agnostic" / "agnostic across assets". ✗
2026-05-13 (id=3305): the monitor itself flags the framing as "partially off".
2026-05-21 (id=3860): notes the pass-thresholds "lived in the user's own FRAMEWORK.md draft".

Net: the cross-asset/multi-slice work is retained as the orthogonal substrate (exactly what orthogonality_probe.py already does). The agnostic axis becomes an algorithm-invariance oracle. Full provenance doc →

← prev next →