Bayesian Estimation of Non-Gausian Time Series with Applicaitons to Transaction Data

A general Bayesian Markov Chain Monte Carlo methodology is utilized for conducting an analysis of the intensity process of stock market data. The sampling scheme employed is a hybrid of the Gibbs and Metropolis Hastings algorithms. Both duration and count data time series approaches are utilized to model trading intensity. Regression effects are incorporated in the model so that market microstructure hypothesis can be tested. The specific analysis is undertaken on Australian stock market data.