Can Reinforcement Learning Efficiently Discover Price Manipulation?
Can AI discover and exploit price manipulation before regulators catch on?
Researchers compared two approaches to finding price manipulation opportunities in financial markets: a traditional method that assumes it knows how prices work, and an artificial intelligence agent that learns patterns from raw data. For moderately volatile markets, the AI agent discovered profitable manipulation strategies using limited training data and actually outperformed the traditional method, even though the traditional method started with correct assumptions about how markets function.
Financial regulators need to understand whether AI systems could discover market manipulation faster than humans can detect and prevent it. This work shows that AI agents can indeed find exploitation strategies that evade detection, suggesting exchanges and regulators must develop better surveillance tools before deploying their own AI systems in trading. The findings also reveal a blind spot in traditional market models: when real-world data is noisy, AI's flexibility can beat expert knowledge—a warning that deploying unsupervised learning in finance without safeguards could create new vulnerabilities.