Real Time Stock Market Analysis Using Agentic AI

Abstract

Modern financial markets have produced vast amounts of data, the complexity of which is way beyond the ability of traditional statistical methods to handle. This paper is a stock-market analysis platform in real time, which combines both deep learning forecasting and an autonomous AI agent. The system will retrieve openhigh-low-volume live data and historical Open-High-Low-Volume (OHLCV) data in the Alpha Vantage API and augments it with feature-engineering operations to generate four technical indicators: Relative Strength Index (RSI), Average True Range (ATR), Moving Average Convergence Divergence (MACD), and Money Flow Index (MFI). In this study, the feature-augmented sequences are used to train a two-layer Long Short-Term Memory (LSTM) network using the TensorFlow. Another unique addition of the work is that an Ollama-driven AI agent is integrated, which interacts with the predictive engine via the Model Context Protocol (MCP), allowing fully conversational and natural-language-driven stock queries. The entire pipeline gets surfaced using Flask web interface that can be accessed by technical and non technical user. Analysis of the intraday data of Apple Inc. (AAPL) provides an intraday predictive fidelity of 1.85, the Mean Absolute Error of 1.42, and the coefficient of the square of the regression, R2, of 0.92. It has a modular and extendable architecture that provides a practical base to next-generation AI-based financial decisionsupport systems.