Mosaic Knowledge Graph

Algorithmic Mosaic Theory — synthesizing seemingly unrelated public information into material conclusions about asset value. Nodes are data sources or assets. Edges are empirical correlations with IC, lead-time, and recency.

How to read this: click a node to see correlated assets and source citations. Click an edge to see correlation history, lead-time window, and the empirical IC. Edges with |IC| < 0.03 or p > 0.05 are archived (not drawn) per anti-overfit guards.

Filters

Graph Statistics

Methodology. Each edge represents an empirically calibrated relationship between an alternative-data source (foot-traffic, port congestion, GitHub activity, etc.) and a specific asset or asset cluster. The IC (information coefficient) is the Spearman rank-correlation of the source signal vs forward asset return at the indicated lead-time window. Sample sizes and citations are surfaced on click. Random-data null-hypothesis test runs nightly via lane-16 backtest harness; edges that fail (|IC| < 0.03 or p > 0.05) are archived and require double-confirm to permanently delete (per feedback_nexus_kill_failing_experiments.md). Charter compliance: feedback_scientific_rigor_portfolio_wide.md, feedback_silent_failures.md, feedback_no_defaults.md.