AN ADVERSARIAL AUTONOMOUS RISK GOVERNANCE SYSTEM FOR HIGH-FRICTION EMERGING EQUITY MARKETS UNDER MULTI-DAY SETTLEMENT AND PRICE CORRIDORS

Abstract

Conventional deep reinforcement learning (DRL) algorithms and high‑frequency trading models in quantitative finance are built on idealized assumptions of market clearing — namely continuous settlement (T+0) and unlimited counterparty liquidity. Yet, when these systems are deployed in emerging equity markets, where settlement delays of multiple days (T+2.5) and strict daily floor/ceiling price bands are unavoidable, they tend to collapse in a systematic way. This paper introduces Agent‑2.5 JueShen a deterministic Autonomous Risk Governance Operating System (RGOS) designed to counter both human behavioral biases and the execution traps inherent in algorithmic trading.