Hybridization of Bat Algorithm with XGBOOST Model for Precise Prediction of Stock Market Directions

Abstract

In recent days, prediction of stock market returns is generally treated as a forecasting problem. The implicit volatile nature of stock market across the world makes the prediction process highly challenging. As a result, prediction and diffusion modeling undermine a wide range of issues present in the stock market prediction. The minimization in prediction error will greatly minimize the investment risks. This paper presents a new method to determine the direction of stock market variations indicating gain and loss. A new machine learning ML based model is applied to predict the direction of stock market prices. The presented model undergoes preprocessing, feature extraction and classification. Initially, preprocessing takes place using exponential smoothing. Then, required features are extracted from the preprocessed dataset. Afterwards, an effective Bat algorithm (BA) with the XGBoost model called BA-XGB is applied for forecasting the stock prices in market. The proposed model predicts whether the stock values gets increased or decreased based on the price existing n days in advance. The presented model is experimented using Apple (APPL) and Facebook (FB) stocks. The obtained simulation outcome stated that the BA-XGB model has offered superior outcome by achieving a maximum accuracy of 96.42.