Introduction

QuantAgent is a multi-agent LLM system for high-frequency trading analysis, combining technical indicators, pattern recognition, and trend analysis through a LangGraph-orchestrated pipeline.

What is QuantAgent?

QuantAgent is a research system that applies large language models to high-frequency trading analysis. It orchestrates four specialized agents — Indicator, Pattern, Trend, and Decision — through a LangGraph pipeline. Each agent focuses on a distinct analytical task, and their outputs are synthesized by the Decision Agent into actionable trade directives.

The system is introduced in the paper “QuantAgent: Price-Driven Multi-Agent LLMs for High-Frequency Trading” (arXiv:2509.09995) by Fei Xiong, Xiang Zhang, Aosong Feng, Siqi Sun, and Chenyu You.

Key capabilities

Four specialized agents

Agent	Role
Indicator Agent	Computes RSI, MACD, Stochastic Oscillator, ROC, and Williams %R from raw OHLC candlestick data
Pattern Agent	Generates a K-line chart, identifies key highs and lows, and returns a plain-language description of the best-matching chart pattern
Trend Agent	Renders annotated K-line charts with fitted trend channels, then summarizes market direction, channel slope, and consolidation zones
Decision Agent	Synthesizes all agent outputs into a trade directive specifying LONG or SHORT position, entry/exit points, stop-loss thresholds, and rationale

Multi-provider LLM support

QuantAgent supports three LLM providers out of the box:

• OpenAI — default models: gpt-4o-mini (agents), gpt-4o (graph logic)
• Anthropic — default models: claude-haiku-4-5-20251001 (agents and graph)
• Qwen (DashScope) — default models: qwen3-max (agents), qwen3-vl-plus (graph)

Because the Pattern and Trend agents analyze visual charts, all providers require a vision-capable model.

Two interfaces

• Web interface — Flask application at http://127.0.0.1:5000 with real-time market data from Yahoo Finance, interactive asset and timeframe selection, and in-browser API key management
• Python API — Import TradingGraph directly for programmatic use in your own scripts or notebooks

Get started

Quick start

Create your environment, install dependencies, and run your first analysis in minutes.

Installation

Detailed setup for conda, pip, TA-Lib, and all three LLM providers.

Web interface

Use the Flask web app to analyze live market data with no code required.

Programmatic usage

Import TradingGraph to run analyses directly from Python.

QuantAgent is for educational and research purposes only. It is not intended to provide financial advice. Always conduct your own research and consult a qualified financial advisor before making investment decisions.