Fitting Multi-Layer Feed Forward Neural Network and Autoregressive Integrated Moving Average for Dhaka Stock Exchange Price Predicting

Abstract

The stock market plays a vital role in the economic development of any country. Stock market performance can be measured by the market capitalization ratio as well as many other factors. The primary purpose of this study is to predict the movement of the stock market based on the total market capitalization of the Dhaka Stock Exchange (DSE) using autoregressive integrated moving average (ARIMA) models as well as artificial neural networks (ANN). The data set covers monthly time series data of total market capitalization from November 2001 to December 2018. This study also shows the best model for forecasting the movement of DSE market capitalization. The ARIMA (2,1,2) model is chosen from among the several ARIMA model combinations. From several artificial neural networks (ANN) models as a modern tool, a three-layer feed-forward topology using a backpropagation algorithm with five nodes in the hidden layer, one lag, and a learning rate equal to 0.01 is selected as the best model. Finally, these selected two models are compared based on the Root-Mean-Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Theil’s U statistic. The results showed that the estimated error of ANN is less than the estimated error of the traditional method.