<![CDATA[The high volatility of the Indonesian capital market (JKSE) poses significant challenges for investors in decision-making. Classical linear forecasting methods are often inadequate due to their inability to capture heteroskedasticity and asymmetric responses in financial data. This study aims to systematically compare and select the best hybrid model combining ARIMA with various GARCH specifications (Standard GARCH, EGARCH, GJR-GARCH, and TGARCH) under different error distributions, and to forecast JKSE price movements for the next three months. Using weekly closing price data from November 2022 to November 2025, the evaluation based on the lowest Akaike Information Criterion (AIC) identifies the ARIMA (0,0,4)-GJR-GARCH model with a Normal distribution as the most robust specification. This finding confirms the presence of a leverage effect in the Indonesian stock market, where volatility responds more intensely to negative shocks than to positive ones. The model demonstrates high accuracy with an RMSE of 2.090. The forecast projects a short-term correction followed by a gradual upward trend reaching the 8,603.62 level by March 2026, accompanied by widening confidence intervals that indicate increasing uncertainty risks in the future.]]>
Copyrights © 2026
