%0 Thesis
%9 Skripsi
%A Alifiya Aziza, NIM.: 22106010002
%B FAKULTAS SAINS DAN TEKNOLOGI
%D 2026
%F digilib:78885
%I UIN SUNAN KALIJAGA YOGYAKARTA
%K volatilitas; EWMA; LSTM; value at risk; Jakarta Islamic Index
%P 166
%T ESTIMASI RISIKO INDEKS SAHAM SYARIAH DENGAN PENDEKATAN HYBRID EXPONENTIALLY WEIGHTED MOVING AVERAGE (EWMA) DAN LONG SHORT-TERM MEMORY
%U https://digilib.uin-suka.ac.id/id/eprint/78885/
%X Volatility is one of the indicators of investment risk that reflects the degree of uncertainty in market movements. Financial return data generally exhibit volatility clustering and non-normal distribution characteristics, requiring methods that can accommodate these properties. This study aims to describe the implementation stages of the hybrid Exponentially Weighted Moving Average (EWMA)–Long Short-Term Memory (LSTM) model, evaluate its volatility forecasting performance, and estimate Value at Risk (VaR) for the Jakarta Islamic Index (JII) over the period from January 2016 to December 2025. The proposed model was developed through volatility estimation using EWMA, sequence generation, data normalization, LSTM model training, and one-day-ahead volatility forecasting using historical returns and EWMA-estimated volatility as input variables. The results show that the hybrid EWMA–LSTM model achieved an MSE of 0.000037, an RMSE of 0.006070, and an MAE of 0.003990. VaR estimation using the Cornish-Fisher Expansion correction yielded an average maximum daily potential loss of 3.70% of the investment value. The Kupiec backtesting results showed a violation ratio of 0.71%, indicating that the VaR estimates were consistent with the 99% confidence level. These findings indicate that the hybrid EWMA–LSTM model can be applied to volatility forecasting and market risk estimation for the Indonesian Islamic stock index.
%Z Dr. Mohammad Farhan Qudratullah, S.Si., M.Si.