Stocksage AI: An Intelligent Explainable Deep Learning Framework for Stock Market Prediction and Portfolio Analysis

The stock market is characterized by high volatility and complex nonlinear behavior, making accurate price prediction a challenging task. Traditional statistical models often fail to capture long-term temporal dependencies present in financial time-series data. This paper presents StockSage AI, an intelligent stock market analysis platform that integrates Long Short-Term Memory (LSTM) neural networks, Linear Regression, technical analysis indicators, and Explainable Artificial Intelligence (XAI) to assist retail investors in making informed decisions. The proposed system collects real-time stock data, performs preprocessing and feature extraction using technical indicators such as Relative Strength Index (RSI), Moving Average Convergence Divergence (MACD), and Simple Moving Average (SMA), and predicts future stock prices using an optimized LSTM model. For newly listed stocks with limited historical data, a Linear Regression model is employed as a fallback predictor. The platform further incorporates a Large Language Model to generate human-readable explanations of prediction results and market trends. In addition, an integrated portfolio management module calculates Short-Term Capital Gains (STCG) and Long-Term Capital Gains (LTCG) taxes according to Indian taxation rules. Experimental evaluation demonstrates that the proposed LSTM model achieves superior prediction accuracy compared with conventional regression models while maintaining low inference latency and providing transparent AI-assisted financial insights. The proposed framework offers an efficient, scalable, and user-friendly decision support system for intelligent stock market analysis.