Iteration 16 — Forex k_v Hardened: The M3 Concern Defused ADMIT HARDENED

What happened, in plain words: when the regime counter earned its forex certificate (iteration 13), one number was flagged for follow-up: its readings correlate strongly (−0.90) with a much simpler quantity — each feature's average redundancy — uncomfortably close to the "it's just a repackaging" kill line of 0.95. This iteration ran the tough version of that test: hand the forecasting panel the simple quantity explicitly, then ask whether the counter still improves forecasts on top of it. It does — clearly (+0.016, entirely positive interval, against a null band an order of magnitude smaller). The counter knows things the simple average doesn't. The certificate stands, now on firmer ground, and the follow-up note is closed. (The literature sweep stayed blocked — search errored a 4th time; nothing unverified admitted.)