QuantAgent

Price-driven multi-agent LLMs for high-frequency trading — combining technical indicators, pattern recognition, and trend analysis to generate actionable trade decisions.

QuantAgent is an open-source multi-agent trading analysis system that coordinates four specialized LLM agents to analyze market data and produce LONG/SHORT trade decisions. It supports OpenAI, Anthropic (Claude), and Qwen as LLM backends, and provides both a web interface and a Python API.

Quick Start

Run your first market analysis in under 5 minutes

Installation

Set up the environment and install dependencies

Multi-Agent Architecture

Understand how the four agents collaborate

Programmatic Usage

Integrate QuantAgent directly into your Python code

How it works

QuantAgent orchestrates four specialized agents in sequence using LangGraph:

1. Indicator Agent

   Computes RSI, MACD, Stochastic Oscillator, Williams %R, and Rate of Change from OHLCV data using TA-Lib, then summarizes momentum and oscillator signals.

2. Pattern Agent

   Generates a candlestick chart image and passes it to a vision-capable LLM that identifies chart formations such as flags, triangles, and head-and-shoulders patterns.

3. Trend Agent

   Fits support and resistance trendlines to recent price action and produces a channel analysis with directional bias.

4. Decision Agent

   Synthesizes all three reports into a final LONG or SHORT directive, including entry/exit rationale and a risk-reward ratio between 1.2 and 1.8.

Explore the docs

Web interface guide

Use the Flask UI to analyze assets without writing code

LLM providers

Configure OpenAI, Anthropic, or Qwen as your backend

Configuration reference

Full list of configuration parameters

TradingGraph API

Complete Python API reference

QuantAgent requires a vision-capable LLM (e.g., GPT-4o, Claude Haiku, or Qwen VL) because the Pattern and Trend agents analyze chart images. Text-only models will not work for those agents.