Construction And Correlation Research Of Realized GARCH Copula Model Based On Generalized Realized Measure

In recent years,the application of high-frequency data is more and more extensive.Based on the research status,this paper studies the classic GARCH model through empirical analysis,and confirms that the model is suitable for the research of financial time series.In addition,this paper introduces the realized GARCH model,and extends it to the case of thick tailed distribution,making the residuals obey t distribution and Skew-T distribution respectively.In the process of comparing the measurement of tail risk effect,this paper analyzes the model robustness according to the Va R effect.Next,we introduce the generalized realized measure into the realized GARCH model based on skewed-t distribution for volatility estimation and Va R prediction,and use MCS test,unconditional coverage test,conditional coverage test,independence test and loss function to compare the Va R prediction effect of the realized GARCH model with different realized measures.In addition,this paper also constructs the realized GARCH copula based on the generalized realized measure,defines the edge distribution function as the realized GARCH based on the generalized realized measure,and describes the structure of each financial time series through the binary copula function.After empirical analysis of the high-frequency data of Shenzhen Composite Index and hang seng index,this paper finds that compared with the classical GARCH and the realized GARCH based on T and normal distribution,the realized GARCH based on Skew-T distribution has obvious advantages in risk measurement;in this paper,the realized GARCH based on Skew-T distribution has obvious advantages In GARCH model,the generalized realized measure is introduced to replace the traditional realized volatility measure,so as to improve the prediction ability of the future market Va R,among which the RVAR realized measure has the best effect in the prediction ability;the t-copula function has the strongest prediction ability in the binary coupla function;in the measurement correlation,the generalizedrealized measure based on the RVAR and the generalized realized measure based on the t-copula function The realized GARCH copula of the present measure also has certain robustness.In addition,this paper also finds that hang seng index and Shenzhen composite index have obvious correlation,which means that there is a correlation between Shenzhen financial market and Hong Kong financial market in a sense.