Emergent multifractality in a multi-asset three-state agent-based market model

This work introduces an Ising-inspired, multi-asset, three-state agent-based model designed to explain how multifractality and long-range nonlinear dependencies emerge in financial markets. Agents can take one of three states-buy, sell, or remain inactive-and their decisions depend on local neighbor interactions, global market sentiment, and cross-asset coupling between multiple financial instruments. The model reproduces key empirical market features, such as volatility clustering, heavy-tailed return distributions, long memory in volatility, and broad multifractal spectra, showing that these properties arise naturally from collective agent behavior rather than being imposed externally. A central result is that strong multifractality appears near a critical balance between individual autonomy and collective synchronization. When agents are too independent, the system is nearly monofractal; when they are too strongly coupled, the market "freezes" into overly stable states. The richest hierarchical structure emerges near the transition between these regimes, where the system alternates between metastable phases and turbulent rearrangements. Extending the model to multiple assets also generates realistic multiscale cross correlations between returns, suggesting that financial correlations arise mainly from direct interactions among investors and only secondarily from asset prices themselves. Overall, the study provides a mechanistic explanation for how complex multiscale organization in financial markets can emerge from simple microscopic interaction rules. From the perspective of applied soft computing, the proposed framework provides an interpretable and flexible computational intelligence approach for modeling complex adaptive financial systems, offering both predictive insight and a deeper understanding of emergent market behavior.