The Study Of Financial Volatility Based On High-Frequency Data

The study of high-frequency financial data is hot in financial field presently for it contains more market information, while the study of financial volatility is always a focus in this field. This paper studies financial volatility estimators and model based on high-frequency data. The key points and main achievements are listed as follows:(1)This paper makes comparison between realized volatility and realized bipower variation from the definition form, robustness and efficiency etc, then the conclusion is drawn that the realized bipower variation is better on definition, robustness and efficiency which is proved by theorem.(2)This paper gives a theorem that the more the power is, the more efficient the realized multipower variation is. The result of the theorem not only proves the efficiency of the multipower variation, but also provides a principle of how to select the power of the estimator. So it is important for application of the multipower variation.(3)In this paper, we improve the realized bipower variation and put forwards weighted realized bipower variation. Thus the realized bipower variation, realized volatility and weighted realized volatility become special examples when the parameters get special value. We also find the weighted realized bipower variation is not only robust but also thinks of the calendar effect. What's more we get a more efficient volatility estimator.(4)We provide a more succinct and convenient method of the optimal sampling frequency and then take the realized bipower variation and weighted realized volatility for example to explain how to get the optimal sampling frequency.(5)This paper put forwards the bias corrected realized bipower variation and the bias corrected weighted realized bipower variation. These two volatility estimators get rid of the bias of microstructure noise and so are unbiased estimators and can have higher sampling frequency to make measuring error smaller.The research is sponsored by National Natural Science Foundation of China: Research on the long-term equilibrium relationships of multivariate moment series and strategies to avoid dynamic financial risk.(NO.70471050).