A Comparison of Forecasting Models of Stock Price using ARIMA and GVM Models

Abstract

The objective of this study is to construct the appropriate forecasting model for the BBL Stock Price via the use of Box-Jenkins method and Grey models. The daily average data, which were collected during July 2022 to June 2023 of 240 days were divided into 2 datasets. The first dataset, which consisted of 222 months from July 2022 to May 2023 was used for constructing the forecasting models via the use of 3 statistical methods, namely, Box-Jenkins, GM(1,1) and  Grey Verhulst Model (GVM) methods. The second dataset, which consisted of 18 days from June 2023 was used for comparing the accuracy of the forecasting model. The criteria of the lowest mean absolute percentage error (MAPE) and root mean squared error (RMSE) were used for comparing the accuracy of the forecasting model. Research findings indicated that for all forecasting methods that were studied, the most accurate method was the GM(1,1) method. This forecasting method was 0.724 percent forecasting error (MAPE = 0.724) or was forecasting error of 1.463 (RMSE = 1.463). The comparison of the accuracy of the forecasting model for both criteria provided the same result. Therefore, it made more reliable that the GM (1,1) method was the most suitable method for this time series.