Stochastic Volatility Modeling based on Doubly Truncated Cauchy Distribution and Bayesian Estimation for Chinese Stock Market

作     者：Cai-feng WANG Cong XIE Zi-yu MA Hui-min ZHAO Cai-feng WANG;Cong XIE;Zi-yu MA;Hui-min ZHAO

作者机构：School of Mathematics and PhysicsUniversity of Science and Technology BeijingBeijing 100083China School of BusinessSun Yat-sen UniversityGuangzhou 510275China 

出 版 物：《Acta Mathematicae Applicatae Sinica》 (应用数学学报（英文版）)

年 卷 期：2023年第39卷第4期

页      面：791-807页

核心收录：

学科分类：12[管理学] 02[经济学] 0202[经济学-应用经济学] 020208[经济学-统计学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 020204[经济学-金融学（含∶保险学）] 07[理学] 0714[理学-统计学（可授理学、经济学学位）] 070103[理学-概率论与数理统计] 0701[理学-数学] 

基　　金：supported by the Open Fund of State Key Laboratory of New Metal Materials Beijing University of Science and Technology (No.2022Z-18) 

主　　题：stochastic volatility model doubly truncated Cauchy distribution Bayesian estimation Markov Chain Monte Carlo method deviance information criterion 

摘      要：In order to measure the uncertainty of financial asset returns in the stock market, this paper presents a new model, called SV-dt C model, a stochastic volatility(SV) model assuming that the stock return has a doubly truncated Cauchy distribution, which takes into account the high peak and fat tail of the empirical distribution simultaneously. Under the Bayesian framework, a prior and posterior analysis for the parameters is made and Markov Chain Monte Carlo(MCMC) is used for computing the posterior estimates of the model parameters and forecasting in the empirical application of Shanghai Stock Exchange Composite Index(SSECI) with respect to the proposed SV-dt C model and two classic SV-N(SV model with Normal distribution)and SV-T(SV model with Student-t distribution) models. The empirical analysis shows that the proposed SV-dt C model has better performance by model checking, including independence test(Projection correlation test), Kolmogorov-Smirnov test(K-S test) and Q-Q plot. Additionally, deviance information criterion(DIC) also shows that the proposed model has a significant improvement in model fit over the others.