Man Group Develops AI Assistant for Quantitative Research
Man Group PLC has introduced the Alpha Assistant, an AI system designed to accelerate the development of quantitative strategies in trend-following. The system helps quantitative researchers by integrating generic AI models with the firm's proprietary data sources and tools.
Investment firm Man Group PLC has developed a new AI assistant, dubbed Alpha Assistant, aimed at enhancing the efficiency of quantitative research, particularly in the design of trend-following strategies. The new system seeks to overcome the limitations of large language models (LLMs) in quantitative research environments.
Off-the-shelf LLMs have shown utility in conversational AI but lack the contextual awareness and integration capabilities necessary for effective quantitative research. Man Group's Alpha Assistant functions as an "agentic AI", capable of accessing proprietary data, writing code using internal libraries, and executing specialized analytical tools through natural language commands.
"Agentic AI systems can function as autonomous toolkits that bridge the gap between generic LLM responses and fully integrated, actionable research output," stated Man Group in a release. The Alpha Assistant is designed to reduce the bottleneck in the implementation phase of the research workflow, which often consumes the most time. Researchers can define analytical objectives in natural language, and the assistant handles code development and debugging.
The system's core is an 'orchestrator' agent that operates with system prompts containing the firm's core principles and conventions. It maintains awareness of the internal code base and has access to modular tools, enabling it to orchestrate complex analytical tasks from natural language prompts. The goal is to free up quantitative researchers' time to focus on evaluating output and formulating new hypotheses, rather than on code development and refinement.
Man Group presented the Alpha Assistant as part of a two-part series exploring agentic AI applications in trend-following strategy design. The initial part outlines the capabilities of agentic AI and contrasts them with the limitations of standard LLMs in a quantitative research context.