Research On Recognition Of Stock Manipulation Behavior Based On Apriori Association Rule Algorithm

During the critical period of the financial industry's opening to the outside world,stock manipulation has become increasingly rampant,and it has gradually developed into a short-term and secretive.Stock manipulation will reduce the efficiency of stock market resource allocation,and will also harm the interests of small and medium investors.In addition,as can be seen from the penalty decision issued by the CSRC,the administrative penalty of the CSRC has a significant lag.This article studies the difficulty of current regulatory authorities to quickly and effectively identify market manipulation.Because of the high proportion of short-term manipulation,the high error of the model in the medium-term and long-term manipulation,and the low confidence of the model,the stock manipulation referred to in this article refers to short-term manipulation.First,on the basis of a comprehensive review of existing research,this article makes an in-depth analysis of the stock market manipulation behavior,and concludes that stock manipulation behavior will have abnormal closing prices,abnormal opening prices,abnormal returns,abnormal daily volatility,The characteristics of abnormal liquidity ratio.According to these characteristics,six indicators based on mathematical characteristics are summarized.Second,this article discusses the Apriori association rule algorithm in depth.On this basis,according to the Apriori association rule algorithm,this article finds the association between various indicators and stock manipulation.Thus,the stock manipulation association rule detection model in this paper was constructed,and a total of 9 strong association rules were found,and the confidences were all over 66.67%.Third,this paper proposes a method for identifying stock manipulation based on the comprehensive analysis of Apriori association rules and variable weight coefficients.This method re-continuousizes the discretized association rules and can directly calculate the stock manipulation police level.We believe that there is a high probability that stocks in Class A and Class B will be manipulated.Fourth,verify the accuracy of the identification method.The result is that the recognition rate of the model for manipulation events is 91.89%,and the overall recognition rate of the model is 84.89%.Compared with other scholars,the accuracy of the model is higher,which provides a new idea for the regulatory authorities to quickly identify stock manipulation behaviors.Fifth,on the basis of the research in this paper,it proposes countermanipulation of China's stock market from six aspects: real-time monitoring,restricting cancellation of orders,random delay of transactions,creating an investment environment,strengthening information disclosure,and improving the investor structure system.