QuantGPT

v1.0.0Finance & Fintechstable

Agent-driven alpha factory — LLM autonomously designs, backtests, and submits factors to WorldQuant BRAIN

a-sharesai-agentalphabacktestingclaude
Share:
252
Stars
0
Downloads
0
Weekly
0/5

What is QuantGPT?

QuantGPT is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to agent-driven alpha factory — llm autonomously designs, backtests, and submits factors to worldquant brain

Agent-driven alpha factory — LLM autonomously designs, backtests, and submits factors to WorldQuant BRAIN

This server falls under the Finance & Fintech category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • Agent-driven alpha factory — LLM autonomously designs, backt

Use Cases

Autonomously design and backtest trading factors.
Submit factors to WorldQuant BRAIN for evaluation.
Build quantitative finance strategies with AI.
Miasyster

Maintainer

LicenseMIT
Languagepython
Versionv1.0.0
UpdatedMay 22, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx quantgpt

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use QuantGPT

QuantGPT is an agent-driven quantitative finance platform that uses Claude (or another LLM) as an autonomous factor designer: it designs alpha factors, backtests them against Chinese A-share markets using free data sources (baostock, akshare), scores them across multiple dimensions, runs anti-overfit statistical tests, and can optionally submit passing factors directly to WorldQuant BRAIN for evaluation. It exposes 15 MCP tools covering the full factor research pipeline from expression validation through rolling walk-forward verification, giving quant researchers an AI co-pilot that can iterate on factors at LLM speed. Teams doing systematic factor investing who want to compress the research cycle from days to minutes will find it a significant productivity multiplier.

Prerequisites

  • Python 3.10 or higher and pip installed
  • DEEPSEEK_API_KEY for LLM-powered factor generation (obtain from the DeepSeek platform)
  • Optional: PostgreSQL if you want a persistent database instead of the default SQLite
  • Optional: WorldQuant BRAIN account credentials (QUANTGPT_CLOUD_EMAIL and QUANTGPT_CLOUD_PASSWORD) for factor submission
  • Claude Desktop or Claude Code as the MCP client
1

Clone the repository and run the setup

Clone QuantGPT and use the Makefile to create a virtual environment, install dependencies, and generate the .env configuration file.

git clone https://github.com/Miasyster/QuantGPT.git
cd QuantGPT
make setup
2

Configure your API keys in the .env file

Edit the generated .env file to add your DeepSeek API key and optional WorldQuant BRAIN credentials.

DEEPSEEK_API_KEY=sk-your-deepseek-key-here
[email protected]
QUANTGPT_CLOUD_PASSWORD=your-wq-brain-password
3

Start the QuantGPT server

Launch the QuantGPT REST and MCP server. It will start on port 8003 with a React dashboard at the same address.

make run
4

Add the MCP server to your Claude config

Register QuantGPT as an MCP server in your claude_desktop_config.json or .mcp.json for Claude Code.

{
  "mcpServers": {
    "quantgpt": {
      "command": "python",
      "args": ["-m", "quantgpt"],
      "env": {
        "DEEPSEEK_API_KEY": "sk-your-deepseek-key-here"
      }
    }
  }
}
5

Restart Claude and run a test backtest

After restarting Claude, ask it to design and backtest a simple momentum factor to confirm all 15 MCP tools are accessible.

QuantGPT Examples

Client configuration

Claude Desktop configuration for QuantGPT with the DeepSeek API key for LLM-powered factor generation.

{
  "mcpServers": {
    "quantgpt": {
      "command": "python",
      "args": ["-m", "quantgpt"],
      "env": {
        "DEEPSEEK_API_KEY": "sk-your-deepseek-key-here",
        "QUANTGPT_CLOUD_EMAIL": "[email protected]",
        "QUANTGPT_CLOUD_PASSWORD": "your-wq-brain-password"
      }
    }
  }
}

Prompts to try

Example prompts covering the factor design, backtest, scoring, and submission workflow.

- "Design a momentum factor for CSI 300 stocks and backtest it over the last 3 years."
- "Validate this alpha expression: rank(close / ts_mean(close, 20)) and check its syntax."
- "Score my latest factor across multiple dimensions and give it a letter grade."
- "Run anti-overfit tests on factor ID 42 to check its statistical robustness."
- "Submit the top-scoring factor to WorldQuant BRAIN for evaluation."

Troubleshooting QuantGPT

The server fails to start with a missing DEEPSEEK_API_KEY error.

Ensure the .env file in the QuantGPT directory contains DEEPSEEK_API_KEY, or pass it via the env block in the MCP config JSON. The key must be active on the DeepSeek platform.

Backtest results return empty data or market data fetch errors.

QuantGPT uses baostock and akshare for free Chinese A-share data. Ensure you have internet access and these packages are installed. Run `pip install baostock akshare` inside the virtual environment if they are missing.

WorldQuant BRAIN submission fails with authentication error.

Register at quant-gpt.com to obtain valid QUANTGPT_CLOUD_EMAIL and QUANTGPT_CLOUD_PASSWORD credentials. BRAIN submission is optional — the local backtesting and scoring tools work without it.

Frequently Asked Questions about QuantGPT

What is QuantGPT?

QuantGPT is a Model Context Protocol (MCP) server that agent-driven alpha factory — llm autonomously designs, backtests, and submits factors to worldquant brain It connects AI assistants to external tools and data sources through a standardized interface.

How do I install QuantGPT?

Follow the installation instructions on the QuantGPT GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with QuantGPT?

QuantGPT works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is QuantGPT free to use?

Yes, QuantGPT is open source and available under the MIT license. You can use it freely in both personal and commercial projects.

Browse More Finance & Fintech MCP Servers

Explore all finance & fintech servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "quantgpt": { "command": "npx", "args": ["-y", "quantgpt"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use QuantGPT?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides