💰💰 DeepFund 🔥🔥

Will LLMs Be Professional At Fund Investment?

This project serves as an ideal solution to the above key question. We evaluate the trading capability of LLM across various financial markets given a unified environment. The LLM shall ingest external information, drive a multi-agent system, and make trading decisions. The LLM performance will be presented in a trading arena view across various dimensions.

🏁 Car-racing is for illustration only. LLM performance is subject to the real market.
‼️ This project is for research purposes only, it DOES NOT TRADE.

Updates

• 2025-10: Check out the latest news about DeepFund on QuantML(link) and SyncedTech(机器之心,link). Besides, our project won the Best Open-source Award (最佳开源项目) in AI Agent 2025 competition (link).
• 2025-09: Our project paper is accepted by NeurIPS 2025 as a poster.
• 2025-08: Our vision paper is accepted by IJCAI 2025 FinLLM Workshop with Oral Paper Award.
• 2025-06: DeepFund & Arena Leaderboard is available for demonstration.
• 2025-05: Our project paper is released on arXiv.

Contact Us

This project aims to be aligned with TRL 1 –> 9 progression, self-rated as 4 (What is TRL?), from exploratory research to production-ready systems, so we welcome any collaboration. If you are interested, reach out by email or scan the WeChat QR code below.

General Inquiries Mr. Changlun Li 📮 cli942[at]connect.hkust-gz.edu.cn 💬 WeChat QR Code

Research Collaborations Prof. Yuyu Luo 📮 yuyuluo[at]hkust-gz.edu.cn

Setup Environment

1. Choose an environment manager:

   • If you want to use Anaconda: Install Conda (if not already installed): Go to anaconda.com/download.
   • If you want to use uv: Install uv (if not already installed): Go to uv installation guide.

2. Clone the repository:

git clone https://github.com/HKUSTDial/DeepFund.git
cd DeepFund

3. Create a virtual env from the env configuration file:

   • If you are using Anaconda:

     conda env create -f environment.yml # if using Conda

   • If you are using uv:

     uv sync # detects pyproject.toml, create an venv at project root, and install the dependencies
     source .venv/bin/activate # or .venv\Scripts\activate for windows to activate the venv

4. Set up environment variables:

# Create .env file for your API keys (OpenAI, DeepSeek, etc.)
cp .env.example .env

Connect to Database

To better track the system performance, DeepFund uses a database to timely monitor the trading status. Besides, it also stores the LLM reasoning results for future analysis and traceback.

Option 1: Use Supabase

DeepFund connects to Supabase by default.

• Supabase is a PostgreSQL-compatible Cloud Database
• You can create a free account on Supabase website.
• Refer to src/database/supabase_setup.sql to create the tables.
• Update the SUPABASE_URL and SUPABASE_KEY in .env file.

Option 2: Use SQLite

SQLite is a lightweight database that stores data locally.

• Run the following command to create a sqlite database in the path

cd src
python database/sqlite_setup.py

• You may need to install VSCode Extension SQLite Viewer to explore the database.
• Path: src/assets/deepfund.db

Relation Diagram

DeepFund system gets supported by four elementary tables:

• Config: store user-defined configurations
• Portfolio: record the portfolio updates
• Decision: record the trading decisions from managers
• Signal: record the signals generated from analysts

The ERD is generated by Supabase - DB Schema Visualizer.

Running the System

Enter the src directory and run the main.py file with configuration:

cd src
python main.py --config xxx.yaml --trading-date YYYY-MM-DD [--local-db]

trading-date coordinates the trading date for the system. It can be set to historical trading date till the last trading date. As the portfolio is updated daily, client must use it in chronological order to replay the trading history. Switch to local DB by adding --local-db option in the command line.

Configurations

Configs are saved in src/config. Below is a config template:

# Deep Fund Configuration
exp_name: "my_unique_exp_name"

# Trading settings
tickers:
  - ticker_a
  - ticker_b

# Analysts to run, refer to graph.constants.py
planner_mode: true/false
workflow_analysts:
  - analyst_a
  - analyst_b
  - analyst_c

# LLM model settings, refer to llm/inference.py
llm:
  provider: "provider_name" 
  model: "model_name"

Planner Mode

We use planner_mode configs to switch the mode:

• True: Planner agent orchestrates which analysts to run from workflow_analysts.
• False: All workflow analysts are running in parallel without orchestration.

Remarks

• exp_name is unique identifier for each experiment. You shall use another one for different experiments when configs are changed.
• Specify --local-db flag to use SQLite. Otherwise, DeepFund connects to Supabase by default.

Project Structure

deepfund/
├── src/
│   ├── main.py                   # Main entry point
│   ├── agents/                   # Agent build and registry
│   ├── apis/                     # APIs for external financial data
│   ├── config/                   # Configuration files
│   ├── database/                 # Database setup and helper
│   ├── example/                  # Expected output
│   ├── graph/                    # Workflow, prompt, and schema
│   ├── llm/                      # LLM providers
│   ├── util/                     # Utility functions and helpers
├── environment.yml               # For Conda
├── README.md                     # Project documentation
├── ...

📚 Additional Resources

Technical Guide - Detailed information for developers including:

• Analyst breakdown and system architecture
• System dependencies and API requirements
• Advanced usage: Adding new analysts and LLM providers

Acknowledgements

The project gets inspiration from the following projects:

• AI Hedge Fund, An AI Hedge Fund Team
• LangGraph, Tutorial on Workflows and Agents
• OpenManus, An open-source framework for building general AI agents
• Supabase, The Open Source Firebase Alternative
• Cursor AI, The AI Code Editor

Star History

Citation

If you find it useful, please cite it as follows:

• Project Paper

@misc{li2025timetravel,
      title={Time Travel is Cheating: Going Live with DeepFund for Real-Time Fund Investment Benchmarking}, 
      author={Changlun Li and Yao Shi and Chen Wang and Qiqi Duan and Runke Ruan and Weijie Huang and Haonan Long and Lijun Huang and Nan Tang and Yuyu Luo},
      year={2025},
      eprint={2505.11065},
      archivePrefix={arXiv},
      primaryClass={cs.CE},
      url={https://arxiv.org/abs/2505.11065}, 
}

• Vision Paper

@misc{li2025deepfund,
      title={DeepFund: Will LLM be Professional at Fund Investment? A Live Arena Perspective}, 
      author={Changlun Li and Yao Shi and Yuyu Luo and Nan Tang},
      year={2025},
      eprint={2503.18313},
      archivePrefix={arXiv},
      primaryClass={cs.MA},
      url={https://arxiv.org/abs/2503.18313}, 
}