Comparison Of Alternative ACD Models Via Density And Interval Forecasts

Records of each transaction on the financial market data referred to as ultra-high frequency data In the financial field. Due to the ultra high frequency data recorded real-time financial market trading information for understanding the microstructure of financial markets provides the foundation and may therefore be ultra-high frequency data research hotspot in recent years the field of econometrics.Since Engle and other conditions for first time since reunification interval (ACD) model to describe the financial markets, changes in trade since the time interval, there have been economists from all angles measure to promote and improve their:Jasiak developed ARFIMA model FIACD (fractionaly interated ACD); Dufour and other proposed index model EACD (exponentialACD); Bauwens such as the establishment of the number of model log-ACD (logarithmic); Dufour and other proposed Box-Cox ACD model; zhang and other developed threshold model (threshold ACD); Hujer, etc.the development of the ACD model of variable structure MSACD (Markov chain regime switehing ACD); Drost, such as establishing semi-parametric model semiparametrie ACD; Beawens conditions such as the establishment of the random time interval model SCD (stochastic conditional duration).The existence of multiple models, the choice of the model appears difficuIn this paper a number of alternative autoregressive conditional duration (ACD) models are compared using a sample of data for three major companies traded on the Australian Stock Exchange. The comparison is performed by employing the methodology for evaluating density and interval forecasts, developed by Christoffersen and Diebold et al., respectively. Our main finding is that the generalized gamma and log-normal distributions for the error terms have similar performance and perform better than the exponential and Weibull distributions. Additionally, there seems to be no substantial difference between the standard ACD specification of Engle and Russel and the log-ACD specification of Bauwens and Giot.The main research work are summarized as follows: (1) Briefly describes various types of ACD models, a simple analysis research status.(2) Pproposed a prediction based on density and interval forecasts ACD model assessment method.(3) The use of Australian Stock Exchange (ASX) three stocks of high-frequency data, respectively, the density based on different ACD models for forecasting and prediction interval.(4) A detailed analysis and comparison of experimental results, and the pros and cons of different models to make an objective evaluation.