A MACHINE LEARNING FRAMEWORK FOR FORECASTING TOP GAINERS IN THE NIFTY50 INDEX

This research introduces a machine learning framework for forecasting top gainers in the NIFTY50 index. The framework refreshes five years of historical stock data daily from Yahoo Finance and applies interpolation,normalization, and feature engineering (SMA, EMA, RSI, MACD). Six different models were considered namely Ridge Regression, Decision Tree, Random Forest, XGBoost, LightGBM, & LSTM are evaluated using MAE, RMSE, and R2. Ridge Regression provided strong baselines (NTPC with R2 0.965), ensembles delivered stability under volatility, and LSTM captured sequential trends with sufficient data. A consensus approach merged per-model rankings into a final top-5 picklist (IOC.NS, ONGC.NS, NTPC.NS, CIPLA.NS, BPCL.NS). Results demonstrate that combining diverse models offers more reliable stock selection than relying on a single regression baseline.