Machine Learning for Stock Market Volatility Prediction Using LSTM and XGBoost: An Indian Market Perspective

The time series behavior of stock price patterns raises a significant challenge in financial analytics: stock market volatility forecasting. This paper proposes a hybrid LSTM-XGBoost model that utilizes the sequence learning capability of LSTM and the ensemble boosting method of XGBoost in its capability to efficiently predict stock market trends. The model incorporates technical indicators like Moving Averages, RSI and MACD as training inputs. The indicators are applied to historical stock prices obtained from the NSE and BSE, respectively. The proposed methodology attains an experimental rate of accuracy of 99.23%, surpassing the individual LSTM, CNN-LSTM, GRU, and XGBoost. The model produces lower RMSE (0.045) and lower MAPE (1.23%) than other systems. The efficiency of the hybrid model as a monetary prediction tool has been established by these outcomes, which have now allowed us to achieve sound investment plans and reduced risk.