MCPInstall
equity-research-agent logo

equity-research-agent

by CarlosFinEngg 2 stars CarlosFinEngg/equity-research-agent

A multi-agent Python system for comprehensive stock analysis and equity research reporting

View on GitHub
READMEmain

Equity Research LLM Agent

A multi-agent Python system for comprehensive stock analysis and equity research reporting.

Overview

equity-research-agent leverages multiple specialized sub-agents to perform:

  • Fundamental analysis (financial statements, ratios, trends)
  • Technical analysis (price data, indicators)
  • Fund flow analysis (capital inflow/outflow, liquidity)
  • Policy analysis (government/regulatory news)

The main agent coordinates these sub-agents and synthesizes their findings into a unified research report.

Features

  • Modular agent-based architecture
  • Integrates API, Google Search, and MCP tools for data gathering
  • Generates detailed, structured equity research reports
  • Flexible report output formats (PDF/HTML) with full Chinese language support
  • Send reports via email to a specified recipient or distribution group

Project Structure

pyproject.toml
uv.lock
stock_analysis_agent/
    agent.py
    config.py
    .env
    ...
    sub_agents/
        fundamental_agent/
        fund_agent/
        policy_agent/
        technical_agent/
mcp/

Requirements

  • Developed and tested on Linux. Performance on Windows is not garanteed.
  • Python 3.11+
  • Dependencies as listed in pyproject.toml, including:

Setup on Linux

1. Environment Rebuilding

To recreate the project's virtual environment using uv, follow these steps:

i. Install uv (if not already installed)

pip install uv

ii. Clone the repository and navigate into it

git clone https://github.com/CarlosFinEngg/equity-research-agent.git
cd equity-research-agent

iii. Sync the environment to match pyproject.toml and uv.lock

uv sync

iv. Activate the virtual environment

source .venv/bin/activate

2. MCP Configuration

For the setup of MCP servers, please refer to the following:

If you use local MCP servers:

  • clone the MCP repo to this project's root folder (parallel with stock_analysis_agent)
  • rebuild the venv within the MCP's folder

If you use hosted MCP servers:

  • add you API key to .env

Lastly, configure the parameters of StdioServerParameters() in agents according to the specific requirement of MCPs.

Usage

  1. Add a .env file in stock_analysis_agent/ and configure environment variables (e.g. LLM and MCP host platform API key) in as needed, for example:
GOOGLE_GENAI_USE_VERTEXAI=FALSE
GOOGLE_API_KEY=...
OPENAI_API_KEY=...
SMITHERY_API_KEY=...

  1. Configure LLM models in stock_analysis_agent/config.py

  2. Activate venv

source .venv/bin/activate
  1. Then run the agent in ADK WEB:
adk web

or run in Terminal:

adk run stock_analysis_agent

The analysis report will be generated in both Markdown format and your chosen output format (PDF/HTML). The reports can be found in the reports directory with the following naming convention:

reports/
    equity_research_report_<company_name>_<ticker>_<YYYYMMDD>.md
    equity_research_report_<company_name>_<ticker>_<YYYYMMDD>.html  # or .pdf

License

MIT

Disclaimer

This project is for educational and informational purposes only. Analyses are generated by AI and does not constitute financial advice. Please conduct your own research or consult a qualified advisor before making investment decisions.