AI Agent swarm for Stock trading simulation (github.com)

A new project has emerged from NEO, showcasing a distributed multi-agent swarm for stock trading simulation, designed primarily for educational and research purposes. This innovative system employs 10 specialized agents that communicate via an asynchronous message bus, enabling effective coordination for trading decisions. The simulation, which uses S&P 500 data over 250 trading days, demonstrates a total return of +4.62% on a simulated capital of $1 million, with a high order approval rate of 86.9%. This highlights NEO's potential for practical applications in algorithmic trading and financial analytics without offering investment advice. The architecture features three analyst agents providing BUY/SELL signals, four trader agents managing individual portfolios, and two risk managers ensuring compliance with stop-loss parameters. Notably, the system's modular design allows for extensive customization, such as incorporating machine learning models for signal generation or adapting it for live trading with broker APIs. With a technical foundation built on Python and Docker, this project not only serves as a robust educational tool but also opens doors for further innovations in AI-driven trading methodologies within the AI/ML community.