Forecasting Stock Price using ARMA Model

Abstract

Forecasting is the process of making predictions based on the historical data. In this paper, we took
the daily opening stock prices of Maxis Berhad from Jan 2010 to Dec 2017 to analyze and forecast
the opening stock prices from Jan 2018 to Dec 2019. Before the modelling part, we examined the
stationarity of the time series data. The data were found to be non-stationary and some
transformation procedures were implemented onto the data such as differencing and log
transformations. After that, the transformed data were modeled with Autoregressive Moving
Average (ARMA) models through Eviews software. ARMA model is the combination of AR(p)
and MA(q) models. In this study, we examined ARMA models of order p+q up to 5 order. Then,
we did the Global and Coefficients tests to produce the selected models. The selected models will
then be inspected based on standard error, r squared and some criteria to obtain the best model.
The best model is used to derive the predicted time series data. The predicted time series data is
then detransformed and compared with the real daily opening stock prices of Maxis Berhad from
Jan 2018 to Dec 2019. Finally, the predicted daily opening stock prices were shown to be having
high accuracy with the Mean Absolute Percentage Error (MAPE) of 1.41%.