Estimasi Risiko Portofolio Saham Dengan Credible Value At Risk Melalui Pendekatan Copula-GARCH

Marsela, Hana (2025) Estimasi Risiko Portofolio Saham Dengan Credible Value At Risk Melalui Pendekatan Copula-GARCH. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5006211034-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Investasi saham merupakan kegiatan mengalokasikan sejumlah dana pada perusahaan yang menawarkan potensi return tinggi. Namun, semakin tinggi return yang diperoleh, semakin tinggi pula risiko yang harus ditanggung. Salah satu cara untuk meminimalkan risiko tersebut adalah dengan pembentukan portofolio saham. Penelitian ini bertujuan untuk mengestimasi risiko portofolio saham menggunakan pendekatan ARIMA GARCH guna mengatasi autokorelasi dan heteroskedastisitas. Selanjutnya, pemodelan struktur dependensi antar saham menggunakan Copula, yang fleksibel terhadap berbagai asumsi distribusi. Risiko kemudian diukur menggunakan Value at Risk (VaR), yang dilanjutkan dengan Credible Value at Risk (CredVaR), yaitu estimasi risiko yang menggabungkan informasi risiko individu dan kelompok melalui teori kredibilitas. Data yang digunakan pada penelitian ini adalah data harga closing harian dari dua saham perusahaan sub sektor perbankan yang terdaftar di IDX30 dengan market kapitalisasi terbesar yaitu PT Bank Central Asia Tbk. (BBCA) dan PT Bank Rakyat Indonesia (Persero) Tbk. (BBRI) selama periode 1 Januari 2022 hingga 1 Januari 2025 yang diperoleh melalui website investing, dengan variabel penelitian yang digunakan adalah return saham. Hasil penelitian menunjukkan bahwa model terbaik untuk saham BBCA adalah ARIMA(1,0,1)–GARCH(0,1) dan untuk saham BBRI adalah ARIMA(0,0,1)-GARCH(1,1). Residual dari model GARCH didapatkan tidak berdistribusi normal, kemudian ditransformasikan ke domain U [0,1] dan dimodelkan menggunakan pendekatan Copula Archimedean. Berdasarkan nilai log-likelihood terbesar, Copula Clayton dipilih sebagai model terbaik untuk memodelkan struktur dependensi antara saham BBCA dan BBRI yang berarti mampu menangkap lower tail dependence. Selanjutnya, dilakukan simulasi data Copula berdasarkan nilai parameter Copula Clayton yaitu θc= 0,8459902 yang kemudian ditransformasikan kembali ke domain return untuk menghasilkan simulasi data return yang akan digunakan dalam perhitungan estimasi risiko portofolio saham dengan bobot masing-masing saham sebesar 50%. Hasil estimasi menunjukkan bahwa nilai VaR pada tingkat kepercayaan 90%, 95%, dan 99% masing-masing adalah sebesar -1,745%, -2,134%, dan -3,216%, sedangkan nilai CredVaR pada tingkat kepercayaan yang sama masing-masing sebesar -1,748%, -2,149%, dan -3,459%. Menunjukkan bahwa Credible Value at Risk memberikan estimasi risiko yang lebih konservatif dibandingkan Value at Risk, terutama pada kondisi pasar ekstrem.
========================================================================================================================
=================
Stock investment is the activity of allocating a certain amount of funds to companies that offer high return potential. However, the higher the potential return, the greater the associated risk. One way to minimize such risk is through the construction of a stock portfolio. This study aims to estimate portfolio risk using the ARIMA-GARCH approach to address issues of autocorrelation and heteroskedasticity. Subsequently, the dependency structure between stocks is modeled using the Copula approach, which is flexible with respect to various distributional assumptions. Risk is then measured using Value at Risk (VaR), followed by Credible Value at Risk (CredVaR), which incorporates both individual and collective risk information through credibility theory. The data used in this study consist of daily closing prices of two stocks from the banking sub-sector listed on the IDX30 with the largest market capitalizations, namely PT Bank Central Asia Tbk. (BBCA) and PT Bank Rakyat Indonesia (Persero) Tbk. (BBRI), covering the period from January 1, 2022, to January 1, 2025, obtained from the Investing website. The research variable used is stock return. The results show that the best-fitting model for BBCA stock is ARIMA(1,0,1)–GARCH(0,1), while for BBRI stock, it is ARIMA(0,0,1)–GARCH(1,1). The residual from the GARCH models do not follow a normal distribution; hence, they were transformed into the [0,1] domain and modeled using the Archimedean Copula approach. Based on the highest log-likelihood value, the Clayton Copula was selected as the best model to capture the dependency structure between BBCA and BBRI stocks, indicating its ability to capture lower tail dependence. Subsequently, a Copula-based data simulation was conducted using the Clayton Copula parameter θ_c= 0,8459902, which was then transformed back into the return domain to generate simulated return data. These were used to estimate the portfolio risk, assuming equal portfolio weights of 50% for each stock. The estimation results show that the VaR values at 90%, 95%, and 99% confidence levels are -1.745%, -2.134%, and -3.216%, respectively, while the corresponding CredVaR values are -1.748%, -2.149%, and -3.459%. These findings indicate that Credible Value at Risk provides a more conservative risk estimate compared to Value at Risk, especially under extreme market conditions.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	ARIMA, Credible Value at Risk, Copula, GARCH, Portofolio Saham ARIMA, Credible Value at Risk, Copula, GARCH, Stock Portfolio
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis H Social Sciences > HG Finance > HG4529.5 Portfolio management
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Hana Marsela
Date Deposited:	14 Jul 2025 00:48
Last Modified:	14 Jul 2025 00:48
URI:	http://repository.its.ac.id/id/eprint/119551

Actions (login required)

View Item