Modeling And Applied Research Of Ultral-high Frequency Time Sequence Based On Heterogeneous Market Hypothesis

Stock market fluctuation and associated features are the key variables which arealways be employed by scholars both at home and abroad to research pricing offinancial derivative instruments, effective portfolio choice and financial riskmanagement, also become a research hotspot. For the past few years, along with thecomputer and communication technology development, the cost of data recordingand storing can be significantly reduced, therefore, high frequency financial dataincreasingly become objects of studying the fluctuation of price of financial assets,meanwhile, it also sets off a wave of studying financial high-frequency timesequence. Since Autoregressive Conditional Duration be put forward by Engle andRussell(1998) and Realized Volatility be proposed by Anderson andBollerslev(1998), modeling based on the data of the high frequency days has madesignificant development.Firstly, based on the understanding of the characteristics about the high-frequencyfinancial time series, this paper deeply explores the modeling problems on thefinancial high-frequency/ultra-high-frequency time series, and makes a detailintroduction on the HAR-RV model based on high frequency data and the ACDmodel based on ultra high frequency data both from the theoretical deduction and theparameters estimation aspects.Secondly, this paper analyzes definition and basic statistical characteristics ofhigh-frequency time sequence, and pay much attention on summarizing basicstatistical characteristics of high frequency data, then,1minute data of Shanghaicomposite index is employed to empirical analysis of basic statistical properties ofhigh frequency time sequence in Chinese stock market, our study shows that highfrequency data of Chinese stock market display obvious “leptokurtosis and fat tail”Non-normal characteristics,“days effect” of high frequency data are also verysignificant.Finally, according to the heterogeneous market hypothesis, the HAR-BACD-V model based on the ultra high frequency time series is constructed and applied inthe Chinese stock market to make empirical analysis. The empirical results furtherconfirm the Chinese stock market traders are heterogeneous, and different frequencytransaction investors’ trading behaviors have different impact on China stock marketfluctuations, which display that the short-term traders have the largest influence tothe stock market fluctuations, the medium term traders next, and the long-termtraders affect minimum. In addition, there exist the obvious negative effects betweenthe microscopic factor trading volume and trading duration, which further proves theEasley and O’Hara’s (1992) market microstructure hypothesis from the empiricalsight. Simultaneously, we found trading volume have a strong positive influence onthe stock market return and volatility, which also shows that China stock marketinformation dissemination are basically consistent with the Copeland’s (1976)continuous information arrival hypothesis.