Weekly Stock Market Index prediction of Pakistan Stock Exchange and Dhaka Stock Exchanges. The Demonstration of Machine Learning Algorithms

Abstract

Stock market is an index of a country financial performance. Evaluating stock markets and predicting their performance using machine learning algorithms is mostly preferred in finance these days. We use stock market data of Pakistan Stock Exchange and Dhaka Stock exchange applying numerous machine learning algorithms i.e. Smoothing Spline Algorithm, Group Mean Data Handling, Neural Net Work Auto Regression and Support Vector Machine to predict the stock market performance and to highlight the forecasting accuracy of these models. We collect data of stock market indices form the first week of July 2011 to the last week closing indices on March 2022. Our study findings confirm that in case of Pakistan the Smoothing Spline Algorithm has better predicting accuracy as compared to the other counterpart machine learning algorithms on the basis of all three error matrices while in case of Bangladesh the Support Vector Machine is preferred model. The results are very meaningful for the investors in these stock exchanges.