Conviction Score Clustering

Conviction Score Clustering aggregates multiple overlapping signals, such as Form 4 filings, transaction patterns, temporal alignment of trades, and market microstructure anomalies, into discrete conviction tiers. Each cluster represents a confidence level, where tier membership is determined by the density and consistency of supporting evidence. This approach mitigates false positives inherent in single-signal detection by requiring corroborating factors to reach elevated conviction thresholds.

In insider trading detection, clustering enables regulators and compliance teams to prioritize investigation resources toward clusters with the highest conviction scores, where violations are most probable. Clustering also reduces computational overhead by replacing continuous scoring distributions with categorical conviction bands, facilitating real-time decision-making on suspicious transactions and pre-clearance approvals.