Wara, Shindi Shella May (2021) Estimasi Value at Risk dengan Model Hybrid-SVR-GARCH-KDE untuk Optimasi Portfolio Saham LQ45. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
06211950010014-Master_Thesis.pdf - Accepted Version Download (1MB) | Preview |
Abstract
Saham merupakan salah satu instrumen finansial yang memiliki variasi yang tinggi dari waktu ke waktu. Investasi pada saham memiliki keuntungan dan risiko. Salah satu cara untuk mengetahui risiko dari suatu saham adalah dengan estimasi Value at Risk. Namun kasus Value at Risk cenderung memiliki variasi yang berfluktuatif dari waktu ke waktu dan susah untuk dimodelkan karena diduga bersifat non linear. Untuk menangkap unsur heteroskedastisitas dilakukan permodelan dengan GARCH, dimana heterogenitas tidak hanya dipengaruhi data waktu sebelumnya, melainkan dipengaruhi oleh variansi dari data sebelumnya. Sedangkan identifikasi model non linear bisa diselesaikan dengan metode machine learning, salah satunya dengan Support Vector Regression yang sensitif terhadap kasus overfitting. Untuk menghasilkan model yang optimal diperkuat dengan menerapkan Kernel Density Estimation. Dengan kombinasi tersebut didapatkan metode Hybrid-SVR-GARCH-KDE. Metode diterapkan dalam dua kajian, yaitu kajian simulasi dan terapan. Hasil dari kajian simulasi menunjukkan bahwa metode Hybrid-SVR-GARCH-KDE baik digunakan untuk melakukan estimasi Value at Risk karena dengan replikasi 50 kali, selisih persentase kumulatif replikasi yang berada di daerah tolak estimasi dengan taraf signifikansi bernilai kecil. Pada kajian terapan dilakukan pada data return pada harga saham LQ45 periode Januari 2018 sampai Maret 2021 yang memiliki PEB dan PBV terkecil dan diterapkan juga pada return hasil optimasi portfolio dari saham LQ45 yang sesuai kriteria sebelumnya. Dari hasil estimasi Value at Risk pada kajian terapan, dihitung jumlah loss dan didapatkan kesimpulan jumlah loss berada di daerah tolak estimasi.
=====================================================================================================
Stock is one of the financial instruments that has high variation from time to time. Stock investment has both profits and risks. One way to determine the risk of a stock is to estimate its Value at Risk. However, Value at Risk cases tend to have fluctuating variations over time and are difficult to model because they are hypothesized to be non-linear. To capture the heteroscedasticity element, modeling with GARCH was carried out, where heterogeneity was not only influenced by the previous data, but also by the variance of the previous data. Meanwhile, the identification of non-linear models can be solved using machine learning methods, one of them is the Support Vector Regression which is sensitive to overfitting cases. To produce an optimal model, it is strengthened by appliying Kernel Density Estimation. By using this combination, the SVR-GARCH-KDE hybrid method is obtained. This method is applied in two studies, namely simulation study and applied study. The results of the simulation study show that the Hybrid-SVR-GARCH-KDE method is good to use for estimating Value at Risk because with 50 replications, the difference in cumulative percentage of replications in the estimated rejection area with a small significance level. An applied study was conducted on return data on LQ45 stock prices for the period January 2018 to March 2021 which had the smallest PEB and PBV and was also applied to returns from portfolio optimization results from LQ45 stocks that matched the previous criteria. From the results of the Value at Risk estimation in the applied study, the amount of loss is calculated and it is concluded that the amount of loss is in the area of rejection of estimation.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | GARCH, Kernel, LQ45, SVR, Value at Risk |
| Subjects: | H Social Sciences > HA Statistics H Social Sciences > HA Statistics > HA30.3 Time-series analysis H Social Sciences > HG Finance > HG4529.5 Portfolio management Q Science > Q Science (General) Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis |
| Depositing User: | Shindi Shella May Wara |
| Date Deposited: | 09 Sep 2021 08:22 |
| Last Modified: | 12 Jun 2024 07:01 |
| URI: | http://repository.its.ac.id/id/eprint/91925 |
Actions (login required)
 |
View Item |