The Research Of Financial Assets Jump Characteristic Base On High-frequency Data

Accurately modeling and prediction of return on assets volatility have become one of the important topics of finance. Most traditional volatility modeling bases on the analysis of low frequency data, which tends to miss a lot of market information, leads to volatility estimation bias. High frequency data fully reflects instantaneous volatility of asset prices, in a timely manner to capture the financial market information. In addition, the high frequency data also includes the rich information in a day, which makes it possible to study the financial market volatility characteristic and market microstructure.There are two directions to model asset returns by considering jump:the jump diffusion process and the infinite activity process. In jump diffusion process, jump occurs at random point, belonging to the finite activity. In jump Infinite activity process, there are infinite jumps in the asset price process, including finite big jumps and possible infinite small jumps, which show infinite activity. In the research of infinite activity jump process, identifying jump activity degree and measuring the activity of jump to build volatility model with activity index are the topic in the first place. For this reason, this paper has carried on the corresponding research, the main work and conclusions are listed as follows:1) This paper introduces the recent research on financial assets jump, and analysis the asset price process, in order to provide the theoretical background for the further study.2) Based on the jumping test statistics, this paper proves the existence of jump from the empirical angle, and finds the infinite jump activity phenomenon. By improving the time-varying intercept, this paper estimates the jump activity index of financial assets price, which is estimated by changing sampling frequency and estimate methods respectively. The empirical results show that the financial asset price jump activity index has relative stability and continuity, which also proves the asymmetry in the angle of jump activity.3) This paper estimates the jump activity index series to build the HAR-RV-CJ-JA model in the structure of HAR model. The empirical analysis result shows that it has significant effect to consider jump activity index when modeling the volatility model. This result illustrates the complementary role of considering small jump to the model.4) By using the quadratic variation, this paper estimates the relative magnitude of the continuous component, the total jump component, the big jump and the small jump respectively. The result shows that the total volatility is mainly controlled by the continuous part, jump part also has the effect of a certain extent; the jump part is relevant to the economic or financial impact in the same period.