MarketSenseAI 2.0 is an enhanced framework for holistic stock analysis that leverages Large Language Models (LLMs) to integrate financial news, historical prices, company fundamentals, and macroeconomic data. The paper introduces two key improvements: a Chain-of-Agents (CoA) approach for fundamental analysis that processes SEC filings and earnings call transcripts, and a Retrieval-Augmented Generation (RAG) module for macroeconomic analysis that uses semantic chunking and Hypothetical Dense Embeddings (HyDE) to handle diverse institutional reports. These enhancements enable more accurate and context-aware investment signals.

Empirical evaluation on S&P 100 stocks over two years (2023-2024) shows MarketSenseAI achieving cumulative returns of 125.9% compared to the index return of 73.5%, while maintaining comparable risk profiles. Further validation on S&P 500 stocks during 2024 demonstrates scalability, delivering a 33.8% higher Sortino ratio than the market. The framework's modular architecture and transparent decision-making process make it suitable for both retail and institutional investors, offering a significant advancement in applying LLM technology to financial analysis.