QuantAgent: Price-Driven Multi-Agent LLMs for High-Frequency Trading
Fei Xiong, Xiang Zhang, Aosong Feng, Siqi Sun, Chenyu You
2025-09-15
Summary
This paper introduces QuantAgent, a new system that uses multiple Large Language Models (LLMs) working together to make decisions for high-frequency trading, which is a very fast-paced type of stock trading.
What's the problem?
Current LLM-based systems are good at understanding financial information and making long-term investment strategies, but they aren't fast or precise enough for high-frequency trading. This kind of trading needs quick reactions to short-term market signals like technical indicators and chart patterns, something traditional LLMs struggle with because they focus on broader, more conceptual understanding.
What's the solution?
The researchers created QuantAgent, which breaks down the trading process into four specialized LLM 'agents': one for analyzing indicators, one for recognizing patterns, one for identifying trends, and one for managing risk. Each agent focuses on a specific part of the market over a very short time frame, allowing the system to make rapid, informed decisions. They tested QuantAgent on things like Bitcoin and Nasdaq futures and found it performed better than existing computer programs and rule-based systems.
Why it matters?
This research shows that combining the reasoning abilities of LLMs with specific financial knowledge can create powerful tools for real-time trading. It opens the door to building trading systems that are not only effective but also easier to understand and track, which is important for trust and accountability in financial markets.
Abstract
Recent advances in Large Language Models (LLMs) have demonstrated impressive capabilities in financial reasoning and market understanding. Multi-agent LLM frameworks such as TradingAgent and FINMEM augment these models to long-horizon investment tasks, leveraging fundamental and sentiment-based inputs for strategic decision-making. However, such systems are ill-suited for the high-speed, precision-critical demands of High-Frequency Trading (HFT). HFT requires rapid, risk-aware decisions based on structured, short-horizon signals, including technical indicators, chart patterns, and trend-based features, distinct from the long-term semantic reasoning typical of traditional financial LLM applications. To this end, we introduce QuantAgent, the first multi-agent LLM framework explicitly designed for high-frequency algorithmic trading. The system decomposes trading into four specialized agents, Indicator, Pattern, Trend, and Risk, each equipped with domain-specific tools and structured reasoning capabilities to capture distinct aspects of market dynamics over short temporal windows. In zero-shot evaluations across ten financial instruments, including Bitcoin and Nasdaq futures, QuantAgent demonstrates superior performance in both predictive accuracy and cumulative return over 4-hour trading intervals, outperforming strong neural and rule-based baselines. Our findings suggest that combining structured financial priors with language-native reasoning unlocks new potential for traceable, real-time decision systems in high-frequency financial markets.