Gold continues to be an important financial asset because of its ability to preserve value during periods of economic uncertainty and inflation. The post-COVID environment created significant fluctuations in commodity and financial markets, increasing the need for reliable gold price forecasting techniques. This study analyses daily gold prices from 1 April 2021 to 30 March 2026 using the Box-Jenkins ARIMA approach. Stationarity was examined through the Augmented Dickey-Fuller test, which showed that the original series was non-stationary and became stationary after first differencing, indicating an integration order of I (1). Patterns in the autocorrelation and partial autocorrelation functions suggested the estimation of ARIMA(1,1,0), ARIMA(0,1,1), and ARIMA(1,1,1) models. Comparative evaluation based on AIC, SIC, and adjusted R² identified ARIMA(1,1,1) as the most suitable model. Both the autoregressive and moving-average coefficients were statistically significant, and diagnostic checking confirmed that the residuals were approximately random. Forecasts produced for 31 March 2026 to 30 September 2026 indicated a moderate upward movement in gold prices during the forecast horizon. Although the model captured the general direction of price changes, the forecasting error measures revealed limitations in obtaining highly accurate point forecasts. The study concludes that ARIMA models are useful for identifying short-term trends in gold prices, but their predictive performance may be improved by incorporating volatility measures and hybrid forecasting techniques in future research..