MountainLion: A Multi-Modal LLM-Based Agent System for Interpretable and Adaptive Financial Trading

2025-07-13

arXiv.org

AI Summary

Cryptocurrency trading faces challenges including difficulty in integrating heterogeneous multi-source data, poor model interpretability, and insufficient adaptability of trading strategies to dynamic market conditions. To address these, this paper proposes an intelligent trading agent system based on multimodal large language models (MLLMs). It introduces a novel cross-modal understanding framework that jointly processes news text, candlestick charts, and signal plots, and designs a multi-agent collaborative architecture augmented with a centralized reflective module to enable real-time analysis, dynamic strategy optimization, and historical policy introspection. The system generates high-quality, interpretable natural-language reports and supports interactive user-driven strategy refinement. Empirical evaluation demonstrates a 23.6% improvement in annualized return and enhanced robustness against macroeconomic shocks and capital flow volatility. The integration of multimodal joint modeling and reflective reinforcement learning effectively balances decision transparency and performance, establishing a new paradigm for human–AI collaborative financial decision-making.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Multiagent Learning
• Cognitive Modeling & Cognitive Systems: Agent Architectures

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Cryptocurrency trading is a challenging task requiring the integration of heterogeneous data from multiple modalities. Traditional deep learning and reinforcement learning approaches typically demand large training datasets and encode diverse inputs into numerical representations, often at the cost of interpretability. Recent progress in large language model (LLM)-based agents has demonstrated the capacity to process multi-modal data and support complex investment decision-making. Building on these advances, we present extbf{MountainLion}, a multi-modal, multi-agent system for financial trading that coordinates specialized LLM-based agents to interpret financial data and generate investment strategies. MountainLion processes textual news, candlestick charts, and trading signal charts to produce high-quality financial reports, while also enabling modification of reports and investment recommendations through data-driven user interaction and question answering. A central reflection module analyzes historical trading signals and outcomes to continuously refine decision processes, and the system is capable of real-time report analysis, summarization, and dynamic adjustment of investment strategies. Empirical results confirm that MountainLion systematically enriches technical price triggers with contextual macroeconomic and capital flow signals, providing a more interpretable, robust, and actionable investment framework that improves returns and strengthens investor confidence.

Problem

• Integrating multi-modal data for cryptocurrency trading decisions
• Overcoming interpretability loss in traditional deep learning approaches
• Enabling adaptive investment strategies through real-time analysis and refinement

Innovation

• Multi-modal LLM agents process text and charts
• Reflection module refines decisions using historical data
• Real-time report analysis with dynamic strategy adjustment