Outliers Detection For AH Index Of Cross-listing Securities Based On Wavelet

Common time series models are firstly probed into this paper,which analyses the basic properties and modeling methods for time series model and builds up model theory basis for empirical analysis.Secondly common outliers detection methods are introduced and the breakthrough method using wavelet is proposed for detecting outliers and time series analysis.According to the outliers detection method-wavelet transform method,this paper carries on a thorough explanation of the theory and the derivation of several common wavelet generating functions which are used to be selected as the empirical analysis part of wavelet generating function.Haar transform function is finally be chosen to assist the latter analysis on AH index.In order to facilitate subsequent empirical analysis,this paper in view of the A + H model interpretation of enterprise,further reveals factors which can influence AH premium,including the traditional four theory,the internal factors and external factors,which combines with the domestic and foreign literature summarizes,attempting to use AH premium index as an example to further discuss outliers detection applications.Finally,this paper introduces the empirical data,AH index for CITIC securities and Haitong Securities by the use of Wind software,from the date that two securities got thier IPO in Hong Kong,on the basis of pretreatment on the time series modeling and outliers detection.Through the detection several relatively obvious outliers are found.In order to analyze the cause of the outliers,this paper deeply studies the factors which can largely influence AH index,from the aspect of internal and external aspects of two securities since cross-listings events,and corresponds the outliers to big events,to discover the main cause of the production of outliers in AH inedx.