Iteration 94 — H-086 Extremogram EXCLUDED KILL #33 · THE TAIL-INVARIANCE PATTERN NAMED · EXAM KILL

In plain words: the second tail-structure candidate died the exact same death as the first — and two identical deaths make a discovery. The extremogram reads how extreme events CLUSTER in time (one big bar following another). Its instrument credentials were impeccable: fed shuffled versions of the same data — same values, all time-ordering destroyed — the real clustering reading beat the shuffled one in every regime (by up to double), proving it reads genuine temporal clustering, not just how fat the tails are; this is precisely the test the recurrence candidate failed weeks ago. But between regimes: the clustering intensity in the covid crash and the calm mid-cycle differ by 0.11 against a noise floor of 0.65. Regime-invariant, again. The pattern now has a name on the discovery register: on this substrate, TAIL structure is the constant — crash-tail co-movement excess (~0.25) and extreme-event clustering (~1.3) are the same everywhere — while BULK structure (whole distributions, predictive advantage, verdict flips) is what regimes actually change. A prediction is registered in advance: the next pair in the queue are also tail meters, and the pattern says they die the same way. If they do, the characterization layer closes with a genuine scientific finding and zero instruments — which is an honest outcome, not a failure.