The Comparison And Analysis Of Nonparametric Estimation Methods Of SV Model

Quantifying the risk in the financial trading market has a good importance for the investment,and this risk is due to the changes of the price.How to predict the risk accurately of the financial trading market is not only the concerns of the investors but also a hot research topic for scholars.Stochastic volatility models have been applied widely,because it could depict the characteristics of stock price movement in financial market.The article uses the two-step estimation procedure method which was proposed by the seniors,this method combine the realized volatitity estimation that is model free with the Markov diffusion model.In the first step,we estimate the unobserved instantaneous volatility process by the methods which are nonparametric.In the second step,we estimate the unobserved diffusion process by N-W estimation method and substitute the latent volatility process with the estimated volatility process in the first step.The main contribution of the article is that when we calculate the instantaneous volatility in the first step,we use the method of Fourier transform and the method of kernel smoothing respectively,then compare and analyze the two estimation methods.The first estimation method is not based on the realized volatility estimation,but on the computation of the Fourier coefficients of the diffusion process.The method is fully model free and nonparametric.The method is fully model free and just assumed that the observations is a semimartingale.Because the conditions mentioned above are simple,so this method is suited to analysing the financial market and it is particularly used to calculate the volatility of high frequency time series.In the next step,we compare the sthrengths and weaknesses of the two different estimation methods by computer simulation.According to the results,we know that the estimation errors of the Fourier transform method is lower than that of the kernel smoothing method,and so is the estimation result of the drift term and the diffusion term of the stochastic volatility model.Finally,the SSE 50 Index between 2019 and 2020 is used for empirical analysis.