ARIMA-Based Time Series Analysis Method for Gold Price Forecasting

Authors

DOI:

https://doi.org/10.33021/41371685

Abstract

Abstrak — Gold is often seen as a key sign of how well an investment is doing because it helps protect against rising prices and uncertain economic times. However, because they change a lot, it's important to predict them accurately in order to make good decisions. This research tries to predict the price of gold in Indonesia by using the ARIMA model, which looks at monthly data from January 2020 up to April 2026. The Box-Jenkins approach was used, which involved checking for stationarity, identifying the model with ACF and PACF, estimating the parameters, and doing checks to make sure the model works well. Several ARIMA models were tested using the Akaike Information Criterion (AIC) and different measures of forecasting accuracy like MSE, RMSE, MAE, and MAPE. The results show that even though the ARIMA(1,1,2) model had the smallest error value (MAPE = 2.8252%), it did not meet the required conditions for the residuals. So, the ARIMA(0,1,2) model was chosen as the best one because it satisfies all the statistical requirements and still provides a high level of accuracy, with a MAPE of 2.9971%. The model can show the overall movement of gold prices, but it doesn't do as well when there are quick and big jumps in price. This study finds that the ARIMA(0,1,2) model works well for predicting gold prices in the short term and can help with making investment choices when the market is fairly steady.

Keywords: Forecasting, Gold Price, ARIMA,MAPE

Downloads

Published

2026-09-28

How to Cite

ARIMA-Based Time Series Analysis Method for Gold Price Forecasting. (2026). Journal of Actuarial, Finance, and Risk Management, 5(1). https://doi.org/10.33021/41371685