In a paper published on March 14, 2026, researcher Hui Gong presents a groundbreaking framework for understanding the role of agentic AI systems in financial markets. The study proposes a four-layer architecture—comprising data perception, reasoning engines, strategy generation, and execution with control—designed to model how autonomous or semi-autonomous AI agents participate in market workflows.(arxiv.org)
Central to the paper is the Agentic Financial Market Model (AFMM), a stylized agent-based representation that links design parameters—such as autonomy depth, heterogeneity, execution coupling, infrastructure concentration, and supervisory observability—to market-level outcomes including efficiency, liquidity resilience, volatility, and systemic risk. This model provides a structured way to assess how different agent architectures might influence market behavior.(arxiv.org)
The paper also includes an empirical application using event studies of AI-agent capability disclosures and heterogeneous market repricing. The findings suggest that the systemic implications of AI in finance depend less on model intelligence alone and more on how agent architectures are distributed, coupled, and governed across institutions. In the near term, Gong argues, the most plausible equilibrium is one of bounded autonomy—where AI agents act as supervised co-pilots, monitoring systems, and constrained execution modules embedded within human decision processes.(arxiv.org)
This development is significant for both industry and regulators. It offers a rigorous framework for evaluating the trade-offs between automation and oversight, and underscores the importance of governance structures in mitigating systemic risk. As financial institutions increasingly deploy agentic AI, Gong’s AFMM could become a foundational tool for designing safer, more resilient market systems.