Enhancing PETRONAS share price forecasts: a hybrid Holt integrated moving average

Abstract

Understanding the variations in PETRONAS share price over time is important for improving the forecast accuracy of PETRONAS share prices to provide stakeholders with reliable analyses for future market predictions. Therefore, the main objective of this study is to improve the accuracy of PETRONAS share price by utilizing a hybrid Holt method with the moving average (MA) from the Box-Jenkins model. Holt's method will address linear trends for non-stationary data, while MA will analyze residual aspects of the data. This combination transforms non-stationary data into stationary by removing noise and averaging out fluctuations. The secondary data used in this study consists of daily observation from bursa Malaysia, the official national stock exchange of Malaysia, covering the period from January 3, 2000, to October 2, 2023. The study encompasses both low and high share price scenarios. The models’ performance was compared using various error metrics across different training and testing splits. The findings highlight that the proposed hybrid [Holt–MA] model called Holt integrated moving average (HIMA) improves the accuracy of forecasting model with the smallest errors for both daily low and high share price. The HIMA model demonstrates significant potential, particularly in reducing residuals and improving prediction accuracy.