Multi-strategy integration for actionable trading agents

Trading agents are very useful for developing and back-testing quality trading strategies to support smart trading actions in the market. However, the existing trading agent research mainly focuses on simple and simulated strategies. As a result, there exists a big gap between academia and business when the developed trading agents are deployed in the real life. Therefore, the actionable capability of developed trading agents is often very limited. In this paper, we introduce approaches for optimizing and integrating multiple classes of strategies for trading agents. Five categories of trading strategies, including 36 types of trading strategies are trained and tested. A strategy integration and optimization approach is proposed to identify golden trading strategy in each category, and finally recommend positions associated with these golden strategies to trading agents. Test in five international markets on ten years of data respectively has shown that the final strategies recommended to trading agents can lead to high benefits while low costs. Concurrent execution of positions recommended by all golden strategies can greatly enhance performance.