<![CDATA[This study aims to compare the forecasting performance of Holt’s Exponential Smoothing and the ARIMA–ARCH/GARCH models in predicting the stock return volatility of Bank Negara Indonesia (BNI). Accurate forecasting of financial time series is essential for investors, policymakers, and market analysts, particularly in emerging markets such as Indonesia, where volatility levels tend to fluctuate due to global and domestic economic conditions. The data used in this study consist of weekly closing prices of BNI stock from January 2020 to August 2025, which were transformed into weekly stock returns. The analysis began with descriptive statistics to examine the trend and volatility behavior of the return series. Holt’s Exponential Smoothing was employed to capture the level and trend components of the data. Meanwhile, the ARIMA–ARCH/GARCH modelling approach was applied to address conditional heteroskedasticity and volatility clustering, which are typical features of financial return data. Model diagnostics, including parameter significance, stationarity tests, and white-noise assessments, were conducted to ensure the suitability of the models. The forecasting accuracy of both models was evaluated using RMSE criteria. The results indicate that the ARIMA ([4],0,0)–ARCH (2) model provides the most accurate predictions, reflected by its lower RMSE value compared to Holt’s Exponential Smoothing. This finding demonstrates that volatility-sensitive models outperform trend-based smoothing methods when applied to financial data characterized by fluctuating variance. Overall, this study highlights the importance of selecting forecasting methods that align with the statistical behavior of financial time series. The findings offer valuable insights for investors, financial analysts, and economic policymakers seeking to improve forecasting accuracy and strengthen risk management strategies in dynamic market environments.]]>
Copyrights © 2026
