CAJ | 학술논문

采用4种Backtesting检验方法，检验22个常态和时变投资组合动态VaR预测模型的风险预测精度，发现GJR-GPD-TV-Copula具有最高的投资组合风险预测精度，GIR-GPD-Copula的拟合、密度预测和组合风险预测精度都要高于GJR-SKST-Copula，且Copula模型的组合风险预测精度分别与拟合精度和密度预测精度存在较弱的正相关关系．
채용4충Backtesting검험방법，검험22개상태화시변투자조합동태VaR예측모형적풍험예측정도，발현GJR-GPD-TV-Copula구유최고적투자조합풍험예측정도，GIR-GPD-Copula적의합、밀도예측화조합풍험예측정도도요고우GJR-SKST-Copula，차Copula모형적조합풍험예측정도분별여의합정도화밀도예측정도존재교약적정상관관계．
Backtests the accuracy of risk prediction of 22 portfolio dynamic VaR forecast models with four methods. The results show that, first, the GJR-GPD-TV-Copula is the highest in portfolio risk prediction accuracy; second, the fitting, density forecasting and portfolio risk prediction accuracy of GJR-GPD-Copula are higher than those of the GJR-SKST-Copula; third, the portfolio risk prediction accuracy is weakly positively correlated with the fitting accuracy and density prediction accuracy of the Copula model.