Indriyani, Silviya (2026) Peramalan Volatilitas Return Indeks Harga Saham Gabungan (IHSG) Dengan Pendekatan Hybrid GARCH-MIDAS-X Dan LSTM. Masters thesis, Institut Teknologi Sepuluh Nopember.
|
Text
6003241015-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (16MB) | Request a copy |
Abstract
Volatilitas pasar saham merupakan ukuran suatu resiko, sehingga volatilitas yang tinggi akan menyebabkan return yang bervariasi dan berakibat pada resiko yang besar juga. Model Generalized Autoregressive Conditional Heteroskedasticity (GARCH) merupakan metode yang digunakan untuk menganalisis volatilitas return saham. Tujuan penelitian ini adalah untuk memodelkan hybrid GARCH-MIDAS-X dan LSTM pada data volatilitas return IHSG. Model GARCH-MIDAS-X digunakan untuk memodelkan volatilitas jangka pendek dengan penambahan variabel eksogen sentimen berita pasar dan nilai tukar rupiah terhadap USD serta memodelkan volatilitas jangka panjang dengan variabel independent tingkat inflasi, suku bunga dan harga minyak mentah yang memiliki frekuensi yang berbeda. Data dibagi menjadi 70% in-sample dan 30% data out-sample. Pemodelan dilakukan dengan dua spesifikasi yaitu nilai selisih (Tipe I) dan nilai volatilitasnya (Tipe II), serta tiga spesifikasi lag pada model MIDAS yaitu lag 6 bulan, 12 bulan, dan 24 Bulan. Hasil penelitian menunjukkan model terbaik berdasarkan ukuran kebaikan AIC dan BIC pada model GARCH-MIDAS-X Tipe I dengan spesifikasi lag 24 bulan dan model GARCH-MIDAS-X Tipe II dengan spesifikasi lag 24 bulan. Peramalan dengan model hybrid GARCH-MIDAS-X dan LSTM pada data out-sample dilakukan dengan memprediksi 10 hari, 15 hari dan 20 hari ke depan. Hasil perbandingan model GARCH-MIDAS-X dengan model hybrid GARCH-MIDAS-X dan LSTM menunjukkan bahwa model hybrid GARCH-MIDAS-X dan LSTM Tipe II dengan spesifikasi lag 24 bulan memberikan kinerja peramalan yang lebih baik dibandingkan model lainnya terutama untuk peramalan dengan horizon yang lebih panjang.
===================================================================================================================================
Stock market volatility is a measure of risk; therefore, high volatility will cause returns to vary and consequently result in greater risk. The Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model is a method used to analyze stock return volatility. This study aims to develop a hybrid GARCH-MIDAS-X and Long Short-Term Memory (LSTM) model for the volatility of the Indonesia Composite Stock Price Index (IHSG) returns. The GARCH-MIDAS-X model is used to model short-term volatility by incorporating exogenous variables, namely market news sentiment and the Rupiah exchange rate against the US Dollar, as well as to model long-term volatility using independent variables of inflation rate, interest rate, and crude oil prices, which have different frequencies. The dataset is divided into 70% in-sample data and 30% out-of-sample data. The modelling was carried out using two specifications, namely the difference value (Type I) and the volatility value (Type II), as well as three MIDAS lag specifications, namely 6 months, 12 months, and 24 months. The modelling results show that the best model based on the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) is the Type I model with a 24-month lag and the Type II model with a 24-month lag. The hybrid model is subsequently used to generate out-of-sample forecast for 10, 15 and 20 trading days ahead. The comparison results demonstrate that the hybrid GARCH-MIDAS-X and LSTM Type II with 24-month lag model provides superior forecasting performance compared with the other models, particularly for forecast over longer prediction horizons.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Indeks Harga Saham Gabungan (IHSG), Volatilitas, GARCH-MIDAS-X, LSTM, IDX Composite Index, Volatility, GARCH-MIDAS-X, LSTM |
| Subjects: | Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis |
| Depositing User: | Silviya Indriyani |
| Date Deposited: | 05 Aug 2026 01:14 |
| Last Modified: | 05 Aug 2026 01:14 |
| URI: | http://repository.its.ac.id/id/eprint/143737 |
Actions (login required)
![]() |
View Item |
