<![CDATA[This study evaluates the sensitivity of the Single Exponential Smoothing (SES) parameter in forecasting the monthly end-of-month closing prices of PT Astra International Tbk (ASII) during a period covering the COVID-19 market disruption and subsequent recovery, rather than merely applying forecasting software. The dataset comprises 72 monthly observations from January 2020 to December 2025. Data from January 2020 to December 2024 were used for model training, while January–December 2025 was reserved for out-of-sample testing. SES was initialized with and evaluated using α values of 0.1, 0.3, 0.5, 0.7, and 0.9. Forecasting accuracy was assessed through rolling one-step-ahead forecasts. The recalculated results show that α = 0.9 achieved the lowest out-of-sample errors, with a MAPE of 5.63%, MAD of Rp301.62, and MSD of Rp115,141.97. These values represent test-set accuracy, not in-sample model fit. The high α enabled the model to respond rapidly to recent price changes, but also increased its sensitivity to short-term fluctuations. After refitting the selected model using all observations through December 2025, the forecasts for January, February, and March 2026 were each Rp6,700. This constant forecast reflects the level-only structure of SES, which does not explicitly extrapolate trend or seasonality. The accuracy obtained can serve as a transparent baseline for short-term investment analysis. However, it should not be the sole basis for investment decisions, as it does not account for market volatility, fundamental information, structural changes, or other risk factors.]]>
Copyrights © 2026
