A Study On The Return Patterns And Correlations Over Trading And Non-trading Periods

There have been numerous studies on return patterns and their correlations that are based on daily returns; however, only a few studies shed light on intraday returns and overnight returns. This paper studies the statistical characteristics of CSI300 Index and its future's intraday and overnight returns as well as close-to-close and open-to-open daily returns, and finds that intraday and overnight returns exhibit different statistical characteristics; therefore, this paper proposes separating daily return into intraday return and overnight return, and building models based on these two returns would improve the goodness of fit and the accuracy of prediction. Furthermore, there exists cross-correlation between intraday and overnight returns as well as between the two daily returns, but autocorrelation of the return series is negligible. And all the returns under study are stable and exhibit strong volatility clustering phenomenon. Based on these return characteristics, this paper employs BEKK-GARCH model and VAR model to build return estimation model based on intraday and overnight returns and based on two daily returns respectively. Comparing the goodness of fit and the accuracy of prediction, overall, the model based on the intraday and overnight returns exhibit higher goodness of fit and accuracy of prediction than the model based on daily returns, which verifies the proposal that the model based on intraday and overnight returns would improve the goodness of fit and the accuracy of prediction.Building model based on intraday and overnight returns, rather than based on daily returns, helps to better capture the characteristics of the return patterns. Additionally, comparing to the increasingly popular high frequency studies, the data for intraday and overnight return studies is more accessible and easier to compute. Therefore, the result of this paper is of great importance for further related studies as well as practical uses, such as transaction strategy design, portfolio construction and risk management etc.