Adaptive market anomaly detection (AMAD): Enhancing minimum spanning tree stability in financial networks

This paper introduces the adaptive market anomaly detection (AMAD) transformation, which enhances minimum-spanning tree stability in financial networks by adaptively dampening extreme market movements while preserving essential return information. Empirical validation across multiple market regimes demonstrates that AMAD-preprocessed MSTs exhibit greater edge persistence, improved structural consistency, and superior risk-adjusted portfolio performance compared to MSTs constructed using raw returns. • AMAD transform enhances MST stability by 36% during market stress with 11% improvement overall. • Cross-market validation confirms consistent improvements across US and European markets. • Theoretically-motivated parameter selection maintains robust performance across configurations. • Network stability translates to superior risk-adjusted performance (36% Sharpe ratio improvement). • Computationally efficient approach offers practical advantages over complex alternatives.