Regulatory Alerts2023-06-29
Machine Learning in Algorithmic Trading: Application by Dutch Proprietary Trading Firms and Possible Risks
The Dutch Authority for the Financial Markets (AFM) published this report to analyze the actual use of machine learning in algorithmic trading among Dutch proprietary trading firms and to identify supervision-relevant risks. The study finds that machine learning is widely applied, primarily using supervised learning to predict prices based on order book data, with firms prioritizing model performance over explainability. The AFM highlights two critical risks: the lack of explainability complicating compliance with organizational requirements, and an increased potential for market manipulation due to unintended behaviors in reinforcement learning models.
Share