Multi-Agent Strategy Enhancement Based on Qlib Model Inference Results

Qlib is an excellent project that provides a path for retail investors to engage in quantitative trading. However, after the model infers and selects the top-k stocks, retail investors often face two challenges: on one hand, they typically lack sufficient capital to purchase all of the top-k stocks; on the other hand, the top-k selection may contain some underperforming stocks. To mitigate risk, a suitable investment strategy for retail investors is needed.

Building on Qlib's outputs, We have developed a set of LLM-based multi-agent strategies to assist with decision-making. This system incorporates additional data sources and employs eight specialized agents to analyze the gathered information: a Macro Analyst, an Industry Policy Analyst, a Valuation Analyst, a Fundamental Analyst, a Technical Analyst, a Sentiment Analyst, a Risk Manager, and an Independent Supervisor. Depending on individual preferences, investors can choose different investment masters (Benjamin Graham, Warren Buffett, Philip Fisher, Peter Lynch, John Templeton, etc.) to help analyze and reach final decisions. Project link: https://github.com/freenowill/stock-fish

The issue names no Qlib files, tests, or entry points. Start by reviewing Qlib's model inference output and the linked stock-fish project; an agreed integration boundary, data sources, agent behavior, and acceptance criteria are needed before implementation can be considered done.