Signal Confluence Matrix

Cross-frame agreement classifier with independence-weighting. Distinguishes genuine confluence (orthogonal evidence agreeing) from echo (correlated inputs producing apparent agreement).

Honesty contract: Convergence is informational only when sources are conditionally independent (Geman-Hwang 1983). When prescience-tier and mosaic-verdict share inputs, "agreement" is partly redundancy not corroboration. We surface independence_score = 1 − Jaccard(prescience-inputs, mosaic-inputs) on every row so you can size positions based on real signal, not echo.
Filter All Strong only Divergent only Actionable (Buy/Sell)
# Ticker State Conviction Indep (2-axis) Prescience Mosaic Intraday Indep (3-axis) Disposition
Methodology — signal-confluence-2026-05-09-v2 + confluence-render-foresight-2026-06-02-v3

Citations (inspired-by): Geman-Hwang 1983 Annals.Stat 11(2):401 (cond.indep + compounding) · Madigan-Raftery 1994 J.Am.Statist.Assoc 89:1535 (BMA) · Kish 1965 Survey Sampling §8.2 (effective N) · Diether-Malloy-Scherbina 2002 J.Finance 57:2113 (disagreement-and-returns) · Pearl 1988 Probabilistic Reasoning §3 · López de Prado 2018 Adv. Financial ML §8 (deflated metrics).