基于经验似然贝叶斯计算方法在随机波动模型中的应用  

Sv Model Estimation Basedon Bayesian Computation with Empirical Likelihood

作　　者：周双酉 钱夕元[1] ZHOU Shuang-you;QIAN Xi-yuan(School of Science,East China University of Science and Technology,Shanghai 200237,China)

机构地区：[1]华东理工大学理学院,上海200237

出　　处：《数学的实践与认识》2020年第6期8-15,共8页Mathematics in Practice and Theory

基　　金：国家高科技研究发展计划(“863计划”)(2015AA20107)。

摘　　要：基于经验似然的贝叶斯计算方法(Bayesian Computation with Empirical Likelihood,BCel)对两种随机波动模型SV-N、SV-T模型进行参数估计,数值实验验证了该方法的可行性和有效性,与传统的基于马尔科夫链蒙特卡罗方法进行了对比研究.最后将SV-T模型应用到上证指数上,利用BCel算法得出该模型的参数估计结果.We innovatively employ the Bayesian computation with empirical likelihood(BCel)to estimate the parameters of two kinds of stochastic volatility models:SV-N,SV-T.Extensive simulation study results show that the proposed method is feasible and effective.Compared with the traditional method based on Markov Chain Monte Carlo,BCel is more efficient.Finally,the proposed method is applied to model the log returns of Shanghai securities composite index with SV-T model.

关 键 词：SV模型 近似贝叶斯算法 经验似然 

分 类 号：O212.8[理学—概率论与数理统计] F832.51[理学—数学]