Study On Parameter Estimation Method Of Stochastic Volatility Model

The volatility of financial time series is a very important characteristic.It reflects the risk of financial market,so people always pay special attention to it.Many scholars focus on risk estimation in the estimation and prediction of volatility.To solve this problem,scholars put forward two basic models: autoregressive conditional heteroscedasticity model and stochastic volatility model.These two methods can describe the characteristics of fluctuation with time well,and then can calculate the volatility more accurately,so it is widely used.However,in contrast,the stochastic volatility model is less limited,can describe the volatility more widely,and more excellent,but its parameter estimation is more complex,so it was not popular at the time.With the improvement of computer technology,the problem of large amount of computation has been solved,and the stochastic volatility model has been used by people.Stochastic volatility model is an important model of volatility quantitative research in financial market,and its parameter estimation is a hot topic in this field over the past ten years.This paper focuses on the parameter estimation methods of two different stochastic volatility models of Markov Montecalo method and experiential characteristic function.The estimation of Markoff Montecalo's method is more precise,but the computation is huge and the efficiency is low.The prediction accuracy of the empirical eigenfunction method is slightly lower,but the computation is small and it is more efficient.Considering the fluctuation of the process may produce structural change,this paper adopt the method of wavelet decomposition,the decomposition of the data into high frequency and low frequency signal,then the estimation method of the model parameters are estimated by using the two parameters to get high frequency and low frequency SV model SV model,the structure of two different frequency data in theory should be there are different.Then add the two models linearly,and get the new model after integration.The new model parameters should be more reasonable and the prediction progress is more accurate.This paper selects the Shanghai Composite Index from November 3,2011 to December 31,2017 the 1500 trading days of the closing price of the empirical data,using the MCMC method and the empirical characteristic function method to estimate the ARSV(1)model,the results showed that the two methods estimate the volatility persistence parameters are similar,but the MCMC estimation method for long time,and the estimated time of the empirical characteristic function in short,these two methods are then used to estimate the model predicted that MCMC method to estimate the model with higher estimation accuracy,finally combined with the wavelet analysis method to get the frequency of the SV model,through the above two kinds of parameter estimation methods have found better prediction model,prediction SV model analysis method combined with MCMC features or experience estimated by wavelet function method,then draws the conclusion,wavelet analysis can improve the frequency estimation of SV model parameter estimation Result?...