Enhancing Stock Market Forecasting with ARIMA and Artificial Neural Networks

Forecasting the stock market has been a topic of great interest with researchers exploring different methods to predict the future price action of stock prices. There are well-established statistical methods of forecasting that can be applied to the stock prices to try and predict the future prices of listed stocks with the highest accuracy. This study sought to predict the price action in the stock market using two forecasting methods and compare the results to determine which of the two has the highest forecasting accuracy. Specifically, the study investigated the forecasting accuracy of Artificial Neural Networks (ANNs) that use ARIMA forecasts as one of the input features and Auto-Regressive Integrated Moving Average (ARIMA) and did a comparison to determine which method had the best accuracy in forecasting future stock prices and trends. In this context, the stock price is the closing price of the stock (CLOSE), which is the last price that the stock was traded during the regular trading day. The results indicated that the proposed Artificial Neural Networks model (ANN) had small Root Mean Square Error therefore having more accurate forecasts than the ARIMA model.