Comparison on estimating Malaysia gold price via nonlinear prediction method and Box–Jenkins model

Abstract

Gold investment worldwide has grown dramatically in the last few years. To be a gold investors, people itself at least should know the basic knowledge of the current gold price. Besides, they should have the best model as a benchmark tool in order to make the decision whether they are supposed to buy or sell the gold. Therefore, to forecast the gold price in Malaysia for the first quarter of year 2017, the time series data of gold price in Malaysia since 4th January until 30th December 2016 were used for this study. The purpose of this study is to determine the possible chaotic behavior of gold in term of price changes using forecasting methods which are Nonlinear prediction (NLP) method and Box-Jenkins method. NLP method consists of the reconstruction of phase space and local linear approximation approach. Meanwhile for Box-Jenkins method, three different ARIMA models with first order differencing have been chosen based on the plots of Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF). The Bayesian Information Criterion (BIC), Mean Square Error (MSE), and Mean Absolute Percentage Error (MAPE) for each model were compared for identifying the best model that suit the time series data of gold price in Malaysia. Out of the three Box-Jenkins model that have been applied to the data series, it was found that the ARIMA (1,1,0) was the best model and it suit the time series model of the gold price trend in Malaysia. A comparison of prediction performance between NLP and Box-Jenkins models based on MAPE was employed and as a result, NLP shows the better prediction performance. Thus, this paper provides a summary of how price of gold can benefit to the investors and also provide a better view of the current gold prices movement and gold investment in Malaysia specifically.