关键词: 密度预测, LOG-ACD模型, SCD模型, MSACD模型

Abstract: Under the condition that the residual item follows Weibull,gamma and Burr distribution respectively and using high-frequency transaction data of two stocks including Pufa bank and G Zhonghai in Shanghai security market,we evaluate and compare the performances of LOGAutoregressive Conditional Duration(LOC-ACD) model,Stochastic Conditional Duration(SCD) model and Markov Switching Autoregressive Conditional Duration(MSACD) model of volume duration using density forecast method. We condude that most models capture the characteristic of cluster of volume duration. We also evidence that MSACD model outperforms LOGACD model and SCD model in abilities of both in-sample fitness and out-sample forecast.

Key words: density forecast LOGA CD model, SCD model, MSACD model