An Empirical Analysis Of Intraday Pattern At Shanghai Stock Market

As an important part of market micro-structure research, after the beginning of research on intraday pattern since the late 80's of the last century, the existence of intraday pattern have been confirmed in many countries ' stock market. Because the high frequency database of our country was not established until 2000, and also because of high frequency financial data available problems, the domestic research on intraday pattern started late and relatively few. Research on the stock market intraday pattern can reflect the degree of effectiveness of the market from the market micro-structure level, reference for investors ' choice of investment strategy and for speculators' arbitrage activities, and provide basis for the high frequency data modeling.For the problems of existing study only involved one aspect of the quality characteristics of stock market, and most study that relative to intraday pattern are just a simple representation of graph without significant test statistics, few research involved significant test statistics did not take into account the impact of other effects, which may lead to bias results. This paper takes Shanghai stock market as the research object and selects 97 stocks(excluding three incomplete data stocks) tick-by-tick data from November 6, 2014 to March 25, 2015 in Shanghai 100 index. After eliminating the interday effect's impact, this paper uses GMM test China's stock market intraday pattern based on 5 market variables, that is bid ask-spread, market depth, trading volume, return and volatility of return. Results are that the bid-ask spread, trading volume, return and volatility of return have significant intraday pattern but the depth of the market do not have.The causes of intraday pattern explained by foreign scholars is from inventory cost, information asymmetry and the different structure of market trading system, which are all factors outside variables. Analysis of low frequency data show that there is mutual influence between stock returns and trading volum e or other market variables, while analysis of high frequency data the mutual influence between the variables will become more prominent and important. Given this, this paper argues that the stock market intraday pattern not only affected by information as ymmetry or other factors outside variables but also affected by the mutual influence among the variables. By establishing multi-variable VAR(SVAR) model, and using the Granger causality test and variance decomposition method, this paper examines the mutua l influence among the variables. Results are that there are different degrees of two-way or one-way Granger causality among the variables and the change of variables are influenced by other variables in relatively large extent.