Investigating the effects of AI trading robots in the financial market - a game theoretical approach

Abstract

This paper discusses the effects of increased proportion of investors using artificial intelligence algorithms to trade in a simulated stock market. The stock market model is modelled using a Markov game and agents’ trading logic is discussed to best mirror the real world. In particular, a Q-learning algorithm will be implemented for agents using artificial intelligence algorithms to trade. The results are discussed and compared across models to analyze the effects.