Volatility Research Of High-Frequency Data Under Shanghai-Hongkong Stock Connect Program

ARCH family model is an important branch of time-series studies,the study of volatility on financial data field is very important, it is often used as a measure of the size of the risk, which can have an impact in the field of risk management, and the high frequency data contains more market information compare to low frequency data.In the field of financial data study, research on stock data plays an incomparable role,and in November 17,2014,the start of Shanghai-Hongkong stock connect program takes a new situation to the mainland stock market.This paper is the use of ARCH model on high-frequency data of stock market under Shanghai-Hongkong stock connect program to model volatility to analyze.This paper selects high-frequency data of 1,5,15,30 and 60 minutes of the Hang Seng AH Index, and uses the ARCH family model to do the volatility research of Shanghai and Hongkong stock market under Shanghai-Hongkong stock connect program. This article first notices the introduction to the various volatility model and related test principles, then do the statistical analysis on the basic information of stock market with different frequency data,we find that the test data are heteroscedasticity after test.The sequence is assumed to obey a certain distribution,this paper aims to choose the optimal volatility model to analyze for different high-frequency data about rate of return with distribution of normal distribution,students t distribution and GED distribution,and to choose the optimal model according to that whether the model parameters is significant, the goodness of fit and SIC criterion.Through the analysis, the paper shows that the high-frequency data of the market have the property of cluster, fat tail with nonnormality, and leverage. The volatility caused by the "negative effect" is greater than "positive effect",and also the market is not fully equipped with self-stabilizing mechanism, it requires external intervention to weaken the impact of external volatility. most high-frequency data distribution is in favor of t distribution,with significantly higher frequency of the time, the model fitting results getting worse. Then use the optimal model for the Hang Seng AH Index prediction,we find that the lower frequency is, the better the forecast comes.Finally,use the best fit model,that is the ARMA-EGARCH(1,1,1)-GED model of 60 minutes to calculate the market value at risk VaR,the result shows that the fitting model evaluates the market risk effectively.