A comparative study of time-series models for forecasting the indonesian gold price

Abstract

Although the value improves from year to year, gold can also suffer from significant price drops sometimes. For example in 2013, due to a severe decline in inflation rate all around the world, the price of gold plummeted either in Indonesia or other countries. Therefore, to make an investment decision and minimize the risks, sufficient information about price fluctuation of gold is highly needed. One of the common approaches for predicting gold price is time-series analysis. The ultimate goal of this research is to determine the most appropriate model of time series to predict gold price in Indonesia. The time series forecasting methods which were compared through simulations are Brown's Double Exponential Smoothing, Holt's Double Exponential Smoothing, and Fuzzy Time Series Markov Chain. The data used in this paper is the daily gold price in Indonesia that was observed from May to July 2020. According to the examination carried out with the Durbin Watson test, it was revealed that the data was dependent on time thus the time series analysis was suitable to apply. The testing with standard and relative statistical measurement showed that the Fuzzy Time Series Markov Chain was the best compared to other tested models, with the value of RMSE= 3269.022 and MAPE= 0.3%.