Typical Characteristics Of High Frequency Data Of The Stock Market Empirical Research

Return and its volatility of financial assets are important concepts in financial economics. The right description of stock return and its volatility would have great influence on the exactness of portfolio selection, the validity of risk management and the rationality of options pricing. Using the high-frequency data of Shanghai Synthesis Index and Shenzhen Composition Index, this paper aims to do empirical analysis on their returns and volatilities. The main contents are as follows,1. This paper estimates the thickness of tail Index of high-frequency returns of China stock market. Firstly, it describes the concepts of heavy tail and tail index and the approaches to estimating the tail index. Then, the paper utilizes two-step sub-sample bootstrap method to estimate the tail index of china stock market. The empirical results are that tail index estimators of Shenzhen stock market are generally larger than two but smaller than three, and vary with sampling frequencies, i.e., shorter the time interval, smaller the indices, indicating heavier tails. It could be reflected that there is high probability for extremal events to take place in Chinese stock market and the market is not efficient enough to absorb big external shocks. In addition, through the empirical analysis on high-frequency returns of six stocks, it could be concluded that the tail index estimation of super stocks are larger than those of smaller ones, demonstrating that smaller stocks are more fragile to extremal events.2.This paper compares three methods to filter the intraday periodicity of volatility. There exists significant intraday pattern for high-frequency volatility of China stock market. If other stylized facts of the volatility are to be correctly studied, intraday pattern must be treated with. This paper introduces three popular deseasonality methods-? -scale, FFF (Flexible Fourier Form) and Multi-resolution of wavelet analysis. After comparison, this paper draws the conclusion that ? -scale is not applicable to filter seasonality of China stock market while Multi-resolution Analysis (MRA) of wavelet analysis is good at separating short-term seasonality from long-term trend. MRA preserves more information from original series than FFF does. If sampling interval is longer than 30 minutes, FFF doesn't work well on the intraday deseasonality.3.Through lead-lag correlation analysis, it is found that coarse volatility predicts fine volatility, implying that China stock market might be composed with participants with different time horizons. And further, this paper identifies and classifies different components of the market through spectral analysis. China stock market has short-term traders as 20 min, 40 min, 2 hours, 2 days and 3days, medium-term traders as 5 days, 7 days, 15 days, one and half months and long-term traders as two months, four months and 9 months.