Realized Volatility Estimating And Option Pricing In The Presence Of Market Microstructure Noise

In the epoch of big data,the information contained in the data has increasingly become a type of efficient material resource.At present,high-frequency trading is still rarely witnessed in mainland China,whereas in Hong Kong is it under development,and in western countries is it in the ascendant.Hence it is foreseeable that high-frequency trading will be the trend in the future.But compared with low frequency data,high frequency data have some specialties,which contain market microstructure noise that can deviate the process of asset pricing,so the models and methods on low frequency data are no longer unconditionally applicable.Based on China 50ETF,this paper studies the market microstructure noise of high-frequency data in two aspects of realized volatility estimating and option pricing.In terms of market microstructure,this paper first reviews its theoretical framework,including definition,function and research scope;then explains its role in price discovery process via models;finally,reveals the existence of market microstructure noise.As a matter of fact,these microstructure noises originates from a series of market frictions in practice,such as trading commissions,information asymmetry,tax implications,minimum quote size,minimum price change,trading hours,trading discontinuity etc..The theory of market microstructure noise was initially proposed in 2004 and hence is a relatively new field in the research of market microstructure.Due to market microstructure noise,realized volatility on high-frequency data,one of estimators of integrated volatility,is no longer unbiased,consistent,and efficient.In light of non-parametric models of realized volatility,it can be proved that the noise in realized volatility approaches to infinity when increasing intraday data frequency and the sample size,thus making it instead a noise estimator.To eliminate the noise,this paper considers two types of realized volatility estimators--Two Scales Realized Volatility and Realized Kernel-Based Estimator.This paper completes an empirical research on China 50ETF data aiming to achieve the following three goals:(i)to explore microstructure features of China equity market;(ii)to eliminate microstructure noise in the realized volatility;(iii)to compare the accuracy of option pricing based on the realized volatility before and after noise reduction.Three main conclusions are yielded:(i)the microstructure noise of China spot market,is generally obey white noise process;(?)Two Scales Realized Volatility and Realized Kernel-Based Estimator can reduce the noise significantly,and are closer to the integrated volatility than standard realized volatility;(?)according to the result of option pricing,different realized volatilities has different effects on various types of options,and the accuracy of option pricing using volatilities with noise reduction is much better than those without.The innovation of this paper is the application of realized volatility after noise reduction.At present,domestic researchers keep studying on how to reduce the noise and almost no one has explored the effects of noise reduction on asset pricing.Secondly,for selecting parameters in non-parametric method,a detailed illustration is provided,which ensures the accuracy,scientificity and repeatability.Thirdly,the research related to Realized Kernel-Based Estimator is initially conducted in this paper,as compared with Two Scales Realized Volatility,such estimator has been seldom studied domestically.This paper makes the comparison of these two estimators and provides the resources for successive researchers.