QuantAgent is a sophisticated multi-agent system designed for high-frequency trading analysis, leveraging LLMs to process price-driven data. It targets engineers, researchers, and power users seeking to automate and enhance trading strategy development by integrating technical indicators, chart pattern recognition, and trend analysis. The system offers both a user-friendly web interface and programmatic access, providing actionable trade directives derived from comprehensive market analysis.

How It Works

QuantAgent employs a multi-agent architecture orchestrated via LangGraph. Key agents include the Indicator Agent for computing technical metrics (RSI, MACD), the Pattern Agent for identifying chart formations, and the Trend Agent for quantifying market direction using annotated charts. A Decision Agent synthesizes these outputs to formulate specific LONG/SHORT trade directives, including entry/exit points and stop-loss levels. A core innovation is its reliance on LLMs capable of image input, enabling visual analysis of price charts for pattern recognition and trend assessment, which is crucial for its analytical approach.

Quick Start & Requirements

1. Create and activate a Conda environment: conda create -n quantagents python=3.11 followed by conda activate quantagents.
2. Install dependencies: pip install -r requirements.txt. For TA-Lib issues, conda install -c conda-forge ta-lib is recommended.
3. Set LLM API keys (OpenAI, Anthropic, Qwen) as environment variables or configure via the web interface.

• Prerequisites: Python 3.11, LLM API keys, and crucially, an LLM that supports image input for chart analysis.
• Usage: Run the web interface with python web_interface.py. The application will be accessible at http://127.0.0.1:5000.
• Links: Official quick-start instructions are embedded within the README.

Highlighted Details

• Multi-Agent System: Features distinct agents for technical indicators, pattern recognition, trend analysis, and trade decision-making.
• Web Interface: A Flask-based application providing real-time market data (via Yahoo Finance), interactive asset selection (stocks, crypto, commodities, indices), multi-timeframe analysis (1m to 1d), dynamic chart generation, and API key management.
• Programmatic Access: Offers a TradingGraph module for integration into custom Python code, allowing direct invocation of the analysis pipeline.
• LLM Flexibility: Supports multiple LLM providers including OpenAI, Anthropic (Claude), and Qwen, with configurable model choices and parameters.

Maintenance & Community

The project lists authors from several prominent universities (Stony Brook, CMU, UBC, Yale, Fudan). Acknowledgements highlight key libraries like LangGraph, OpenAI, Anthropic, Qwen, and yfinance. No specific community channels (Discord/Slack) or roadmap links are provided.

Licensing & Compatibility

The project is licensed under the MIT License. This permissive license allows for broad compatibility, including commercial use and integration within closed-source projects.

Limitations & Caveats

This software is intended for educational and research purposes only and does not constitute financial advice. It requires LLMs capable of processing image inputs for its core analysis functions. Data fetching relies on Yahoo Finance, which may have limitations for certain symbols. Users dealing with large datasets might need to adjust analysis windows or timeframes to manage memory constraints.