Multi-Agent Financial Analysis System: From Real Time Data Fetching to Expert Advisory

Abstract- This paper suggests an automated financial analysis multi- agent system based on modular rule-based architecture
implemented in Python. The design involves independent agents—DataFetcherAgent, NewsAgent, and FinancialExpertAgent—managed by a central CoordinatorAgent [1] [2]. Stock data is fetched using the yfinance library, and sentiment is emulated to replicate actual world news polarity. The framework uses deterministic principles based on financial metrics including price-to-earnings (P/E) ratio, volume, and price change to output explainable investment suggestions. Transparent and extensible in nature, the architecture eschews third-party cloud APIs and allows for complete execution. The output is a structured investment advice: type of recommendation (BUY/SELL/HOLD), risk, confidence score, and sentiment summary. This framework acts as a light and interpretable base for subsequent investigations in agent-based systems, financial intelligence, and rule-based AI and can be further extended to include autonomous agents, natural language processing, machine learning, or LLM-based
advisory systems [3][4].

Keywords: Multi-agent systems, Financial analysis, Autonomous agents, Sentiment analysis, Investment advisory, Risk Management, Recommendations, LangChain, Agent2Agent, Model Context Protocol