Forecasting Monthly Spot Gold Prices of XAU/USD Based on the ARIMA Model

Abstract

Gold is widely recognized as a representative safe-haven asset in the global financial system, and the accurate modeling of its price fluctuations is of considerable importance for macroeconomic risk management and asset allocation. This study draws on monthly closing prices of spot gold (XAU/USD) spanning April 2016 to March 2026, using this decade-long window to build an ARIMA time-series framework aimed at tracing how gold prices have evolved over time. The empirical procedure applies a logarithmic transformation, first-order differencing, and the Augmented Dickey–Fuller (ADF) test to assess stationarity. Model selection follows the Akaike Information Criterion (AIC), which points to an ARIMA (1,1,0) specification with drift as the best-fitting option among the candidates considered. The estimation results indicate that, after controlling for stochastic disturbances, the spot gold price series displays a statistically significant upward drift over the long run. Projections for the next 24 months imply that the central tendency of gold prices could rise toward USD 6,800 per ounce. But as the forecasting horizon increases, widening confidence bands also indicate increasing volatility risks under macroeconomic uncertainties.