The Study Of Volatility And Jumps Based On High-frequency Financial Data

Researchers and practitioners alike now to study high-frequency financial datafor it contains much more market information than low-frequency data, while how tomeasure the volatility accurately income has been one of the hot research issues infinancial field. So how to estimator volatility based on high-frequency data becomesmore important.This paper mainly introduce two different ideas of pre-averaging methodproposed by Podolskij and Vetter and Jacod et al. To eliminate the impact ofmicrostructure noise, they choose non-overlapping and overlapping blocks to estimatevolatility rely on high-frequency data. They also gives the proof of first orderasymptotic distribution. Because of the reason that the pre-averaging returns arepossibly heteroskedastic, this motivates the application of the Wild Bootstrap. Whenstudy about jumps, we introuduce the consistent estimator of integrated volatilitywhile jumps existed. We also introduce test statistics of jumps.The main achievements are listed as follows：The first chapter is an introduction. In this chapter we introduces the backgroundand the research survey of volatility and jumps. The second chapter introduces thepre-averaged realized volatility and their asymptotic theory. We also study the optimalchoice of the sample frequency. The third one combines the wild Bootstrap methodwith the pre-averaged realized volatility. And construct the percentile interval forintegrated volatility in this context. Chapter four compare the finite sampleperformance of the Bootstrap with what Podolskij and Vetter proposed through MonteCarlo experiments. And use empirical work to illustrate the validity. The last chapteris a conclusion of this paper and points out our shortage and future research direction.Innovation points of our paper are that we get a comprehensive understanding ofthe pre-averaged realized volatility. And use the real trading data for empiricalanalysis.