A Novel Approach for Stock Market Prediction Using Feature Enhanced Learning Model (FELM)

Abstract- Stock market prediction is an important task that is required by investors, traders and financial institutions in order to make informed decisions. Traditional models have the following problems like volatility, historical data inputs are dependent on features, model cannot adjust to real-time market changes. A novel feature-enhanced learning model, a Feature Enhanced Learning Model (FELM), is proposed which integrates advanced feature engineering with a hybrid learning method based on Extreme Gradient Boosting (XGBoost). Using technical indicators and sentiment analysis as well as macroeconomic data as inputs, the model gains greater accuracy which is around 99%-100% when compared with previous traditional models like ARIMA and LSTM.