A General Framework Of Realized Volatility Modeling

This paper we will first propose a generalized concept in realized volatility modeling,we build a general framework of realized volatility modeling(Generalized heterogeneous autoregressive extended model of realized volatility,GHAR-E),we will point out GHAR-E modeling framework unified the field of realized volatility modeling.GHAR-E modeling framework will be the primitive model in the field of realized volatility modeling.We will propose a new model group which is GHAR-EJ-EXAMPLE based on GHAR-E modeling framework.We will define a general predict factors weighting method that given the original predictor sequence weighting to build the new predictor sequence based on GHAR-E modeling framewok.We significantly improve a non-parametric filter method’s algorithm by Fan and Yao(2003)62.The new algorithm that is much more efficient than old one.We study a several popular nonparametric jump detection methods including BNS(200425,2004b26,200627,2006b29)of the RB V,ABH(2011)9 of MinRV and MedRV,and CPR(2010)8 of TMPV models.We make lots of different inferences and conclusions.Ultimately determine the optimal sub model is CPR(2010)8TMPV based on general framework.CPR(2010)8 TMPV method can better identify days jump,with higher stability and predictive ability.Comparing with GHAR-EJ series model,GHAR-EJ-EXAMPLE series model has better fitting and prediction ability.The Example GHAR-EJ-EXAMPLE series model is one of feasibility modeling under GHAR-E modeling framework.GHAR-E modeling framework,directing the future feasibility modeling of realized modeling.