Hybrid LASSO-Quantile Regression dan Support Vector Regression (LASSO-QR-SVR) untuk Pemodelan Conditional Value-at-Risk Return Saham Perbankan di Indonesia

A, Ahmad Hilal. (2025) Hybrid LASSO-Quantile Regression dan Support Vector Regression (LASSO-QR-SVR) untuk Pemodelan Conditional Value-at-Risk Return Saham Perbankan di Indonesia. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6003232007-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Fluktuasi harga saham perbankan dalam beberapa tahun terakhir, terutama akibat pandemi COVID-19, menunjukkan pentingnya analisis risiko yang komprehensif untuk membantu investor dalam pengambilan keputusan. Metode Value-at-Risk (VaR) dan Conditional Value-at-Risk (CoVaR) digunakan untuk mengestimasi risiko, dengan CoVaR dianggap lebih efektif dalam menangkap risiko sistemik dan efek limpahan (spillover effect) antar perusahaan keuangan. Penelitian ini bertujuan untuk menganalisis risiko sistemik pada saham perbankan di Indonesia dengan menggunakan metode Hybrid LASSO-Quantile Regression dan Support Vector Regression (LASSO-QR-SVR) yang melibatkan variabel makroekonomi. Hybrid LASSO-QR-SVR digunakan untuk meningkatkan akurasi estimasi risiko, menggabungkan keunggulan seleksi variabel LASSO, fleksibilitas Quantile Regression (QR), dan kemampuan Support Vector Regression (SVR) dalam menangani data non-linear. Data yang digunakan mencakup harga saham perbankan dan variabel makroekonomi dari Juni 2018 hingga Januari 2025. Hasil penelitian menunjukkan bahwa estimasi nilai risiko menggunakan VaR-POT dan CoVaR LASSO-QR-SVR mengalami penurunan seiring meningkatnya level kuantil yang digunakan. Saham ARTO, BBHI, BTPS, dan BRIS memiliki risiko kerugian yang tinggi, sehingga perlu dipertimbangkan untuk melakukan investasi. Saham ARTO dan BBHI memiliki nilai varians terbesar pada semua kuantil, menunjukkan tingkat fluktuasi yang tinggi dan pergerakan yang kurang stabil. Dari segi keakuratan CoVaR LASSO-QR-SVR memiliki selisih terkecil antara Expected Shortfall dan kuantil, dengan semua model akurat di kuantil 1%, 5%, dan 10%. Berdasarkan kupiec test (POF test), CoVaR LASSO-QR-SVR juga menunjukkan validitas yang tinggi dengan semua model valid pada kuantil 1%, 5%, dan 10%. Hasil ini menunjukkan bahwa CoVaR LASSO-QR-SVR lebih unggul dalam akurasi dan validitas.
====================================================================================================================================================================================
Fluctuations in banking stock prices in recent years, particularly due to the COVID-19 pandemic, highlight the importance of comprehensive risk analysis to assist investors in decision-making. The Value-at-Risk (VaR) and Conditional Value-at-Risk (CoVaR) methods are used to estimate risk, with CoVaR considered more effective in capturing systemic risk and spillover effects among financial institutions. This study aims to analyze systemic risk in Indonesian banking stocks using the Hybrid LASSO-Quantile Regression and Support Vector Regression (LASSO-QR-SVR) method, incorporating macroeconomic variables. The Hybrid LASSO-QR-SVR approach is employed to enhance risk estimation accuracy by combining the strengths of LASSO variable selection, the flexibility of Quantile Regression (QR), and the capability of Support Vector Regression (SVR) in handling non-linear data. The data used includes banking stock prices and macroeconomic variables from June 2018 to January 2025. The result findings indicate that the estimated risk values using the VaR-POT and CoVaR LASSO-QR-SVR methods tend to decrease as the quantile level increases. Stocks such as ARTO, BBHI, BTPS, and BRIS exhibit high loss risk, suggesting that investment decisions involving these stocks should be made with caution. ARTO and BBHI stocks show the highest variance across all quantiles, indicating high volatility and less stable price movements. In terms of accuracy, the CoVaR LASSO-QR-SVR model demonstrates the smallest difference between Expected Shortfall and the corresponding quantile, with all models performing accurately at the 1%, 5%, and 10% quantiles. Based on the Kupiec test (POF test), the CoVaR LASSO-QR-SVR model also exhibits strong validity, with all models being valid at the 1%, 5%, and 10% quantiles. These results suggest that CoVaR LASSO-QR-SVR outperforms other models in terms of both accuracy and validity.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	CoVaR, Hybrid LASSO-QR-SVR, Risiko Sistemik, Saham Perbankan, VaR,CoVaR, Hybrid LASSO-QR-SVR, Systemic Risk, Banking Stocks, VaR.
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis H Social Sciences > HG Finance > HG4529 Investment analysis H Social Sciences > HG Finance > HG4910 Investments H Social Sciences > HG Finance > HG4915 Stocks--Prices
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis
Depositing User:	Ahmad Hilal. A
Date Deposited:	04 Aug 2025 05:07
Last Modified:	04 Aug 2025 05:07
URI:	http://repository.its.ac.id/id/eprint/125490

Actions (login required)

View Item