Examining Daily Closing Price Prediction of the NSE Index using an Optimized Artificial Neural Network: A Study of Stock Market

Abstract

Prediction of the stock market is considered a challenging task because of non-linear and speculative nature of data. Stock prediction means an attempt to forecast the future of a stock or any other financial instrument listed on any stock exchange. The success of stock prediction returns a significant profit for investors and daily traders. Deep learning models have proven to be a reliable option for developing successful prediction systems. In recent years, the use of hyperparameter optimization techniques for the creation of precise models has grown significantly. In this study, an attempt has been made to propose an optimized Artificial Neural Network for predicting the daily closing price of the NIFTY-50, an Index of the National Stock Exchange. For this purpose, the novel dataset has been generated using openly available financial data, and two technical indicators have been used to predict the intraday closing price of the index. To evaluate the performance of the proposed model, a standard percentage-based method known as MAPE, R², and RMSE has been applied. After evaluation of the model, lower values of MAPE and RMSE have been achieved which depicts that the model is efficiently predicting the stock closing price.