Combining Prediction Models and Multiagent Systems for Automated Financial Trading

AI technologies have revolutionized financial investment by enhancing efficiency, accuracy, and accessibility. These technologies, encompassing algorithmic trading, robo-advisors, and AI-driven analytics, enable investors to execute complex trading strategies with minimal human intervention. Choosing an investment strategy and financial trading tool usually involves considering several key factors to align the investor’s needs with his/her financial goals, risk tolerance, profitability, and market outlook. One of the challenges with these approaches is that they cannot simultaneously deploy diverse investment strategies. As a consequence, they increase exposure to specific risks associated with a single strategy, potentially leading to significant losses if that strategy underperforms. Without diversification, the investment portfolio lacks balance and becomes more vulnerable to market volatility and unexpected economic shifts. In order to address these issues, automated collaborative investment systems may offer significant benefits such as enhancing market analysis and improving decision-making through agent collaboration and competition. Accordingly, this article proposes an automated investment multiagent strategy (IMAI) to combine intelligent agents techniques and recurrent neural networks. The approach aims at improving purchase and sale decisions of financial assets such as EUR/USD exchange rates, based on data from a digital Broker’s Platform and the next day’s price predictions. Thus, the method may expose investment capital to low risk, obtaining annual returns higher than those attained by using individual investment systems. The experiments showed better transaction performance of our IMAI approach with profitability of 99.73% compared to a single-agent model with 76.34%.