A Review Of Time Duration Models Based On Estimation Functions Theory

High-frequency data analysis has important reference value for financial market evaluation and can help investors and investment institutions make effective judgments on the financial market.As a variable generated in the modern financial market,time duration has the characteristics of high-frequency data,which can describe the dynamic changes of traders’ behavior and trading events.The autoregressive conditional duration(ACD)model is a model established based on the characteristics of time duration variables,modeling time duration and its conditional expectations,using the idea of the Garch model,establishing a linear regression model,and expanding the traditional econometric economy The model is limited based on a fixed time interval,which expands new ideas for high-frequency data and ultra-high-frequency data analysis.At present,the ACD model and its extended form have become one of the important tools for financial market analysis,and have received extensive attention from scholars.The collation and research of the model is of great significance.This article mainly reviews the research progress on the time duration model.It sorts out and summarizes the research situation of time duration model,sorts out the main types of ACD model and its extended model,introduces more traditional model construction methods,and analyzes the main characteristics of several types of time duration model extended models.At the same time,this article summarizes the research status of the estimation function theory(EF),considering the advantages of the estimation function theory,and applying the estimation function method to estimate the parameters of the time du3 ration model.The theoretical proof shows that the use of the estimation function theory is effective for the parameter estimation of the time duration model.The time duration model is mainly based on the conditional autoregressive model.The overall development of the model is three stages,namely the linear ACD model,the nonlinear ACD model,and the semi-parametric and non-parametric ACD model.Models at different stages have different characteristics.This article describes representative model forms.The empirical research of a large number of documents shows that the ACD model and its extended form have excellent applicability to the financial field at home and abroad and some other fields.At the same time,by introducing the basic content of the EF method and its most important extended form,the joint estimation function(CEF),it is found that the CEF method can obtain slightly better than pseudo maximum likelihood estimators,and has strong applicability and is suitable for most long-term Period models and time series models.