Iteration 29 — seqICP Crypto Persistence Role EXCLUDED SCOPED KILL · CANDIDATE ALIVE

In plain words: the label-free test took the same two-part crypto exam the last two admitted instruments took: does its reading help predict which features stay orthogonal next regime? The run itself was clean — regime by regime, its invariance flag lights up for a small, sensible set of features, and one (bar_ehlers_increment_asymmetry) stays lit across every late-era regime, quietly corroborating last week's identifier yet again. But the exam result was a miss with an instructive shape: the features the flag lights up for are almost all outside the certified prediction panel's universe — and the one inside it isn't orthogonal at the relevant moments. On the held-out test set the flag never lit at all: a constant reading can't predict anything. One exam part narrowly missed; the other was unmeasurable by construction. Verdict, by the pre-registered rule: the crypto persistence role is excluded — a clean, scoped kill. The candidate itself stays alive: its real claim (working where regime labels are wrong or absent) is decided next, on forex, head-to-head with the reigning identifier.