Clusters · MAY 2026
Decomposed thesis structures.
A cluster is an anchor question and the child events whose joint resolution defines a regime. Each child scores a channel-specific factor — rates, labor, inflation, credit, growth, policy — and the anchor probability is recomposed from the children. Clusters publish their full child set, model-by-model probabilities, and resolution schedule.
Clusters compound in value as their child events resolve. A cluster page surfaces the anchor probability, the per-child probabilities and 4-model breakdown, and the resolution schedule. Calibration of anchor and child events is tracked alongside the cross-event archive on the methodology page.