Financial (Ultra)High Frequency Data Modeling And Empirical Analysis

Financial markets operate efficiently in the opening time,thus high frequency data appear accordingly.High frequency data usually are thought to be data collected at minute or second interval,which contain plenty of market information.Traditional econometrical modeling is based on fixed interval data.That modeling method has drawback in practice use,because the microstructure features of data will be lost.This paper propose an alternative to fixed interval analysis and construct autoregressive conditional duration model based on Nikkei225 index futures in Singapore.This paper is composed of six chapters.In the first chapter,we introduced new development in econometrical area,which emphasized how high frequency data influnence econometrics.In the second chapter,we introduced financial market microstructure theory which based on inventory and game theory.In the third chapter,we first discussed features of high frequency data and then studied SV and JKL model describing the relation between price volatility and transaction volumn.In addition,allowing for game theory in market microstructure theory,we put forward the model which specified the dynamics of intraday bid and ask quotes.In the fourth chapter,we put forward basic ACD model and extended ACD models.In the next section we used ACD model to study Nikkei225 index futures in SIMEX with high frequency data.Empirical results indicated ACD model did a good job in describing features of uhf data.Finally,we concluded the content of this paper and put forward several directions in future study.