Comparing the Forecasting Abilities of Stock Returns of Electricity Sector Stocks in the Stock Exchange of Thailand Using ARIMA and ARIMAX Models

Abstract

This research aims to study and compare the forecasting ability of stock returns for the power plant energy sector listed on the Stock Exchange of Thailand (SET) over a period from January 2018 to December 2023, a total of 72 months. This study employed the ARIMA and ARIMAX models to forecast the stock prices of 10 companies in the power plant energy sector: BGRIM, GPSC, GULF, ACC, CKP, EA, GUNKUL, NOVA, SOLAR, and SSP. The forecast accuracy was evaluated using the Mean Absolute Deviation (MAD) and Mean Absolute Percentage Error (MAPE) as criteria to identify the model with the least error. The results demonstrate that the ARIMAX model, which incorporates independent variables such as global crude oil prices, the Thai baht exchange rate, and the stock market index, significantly enhances forecasting accuracy. In most cases, ARIMAX outperformed the ARIMA model, yielding lower MAD and MAPE values. For example, for SOLAR, ARIMA produced a MAPE of 12.365%, while ARIMAX achieved a lower MAPE of 9.749%. Similarly, for GPSC, the MAD decreased from 4.596 (ARIMA) to 3.420 (ARIMAX). The results indicate that incorporating relevant independent variables can significantly enhance the accuracy and efficiency of the forecasting model.