ESTIMASI CVAR PADA PORTOFOLIO SAHAM MENGGUNAKAN METODE GJR-EVT DENGAN PENDEKATAN D-VINE COPULA

Abstract

Risk measure using Conditional Value at Risk can be calculate if values that exceeds the p-quantile is known in VaR. The models used to accommodate characteristics of the stock portfolio in this research are EVT-GARCH-D-vine copula and EVT-GJR-D-vine copula so the performance of these two models can be compared. A comparison of the performance of the EVT-GARCH-D-vine copula and EVT-GJR-D-vine copula models can be seen from the Kupiec test backtesting process. Exceeded value Kupiec Test on CVaR 99% is 2, CVaR 95% is 6, and CVaR 90% is 13 for AR(1)-GARCH-t(1,1)-GPD and CVaR 99% is 3, CVaR 95% is 7, and CVaR 90% is 13 for AR(1)-GJR-t(1,1)-GPD. The Kupiec test describes the estimated risk value of CVaR running well with the value of the entire model above the significant level of ? = 0.05 so as to provide a conclusion of risk estimates considered feasible.