Dynamic Modelling of Irregular Times, Prices and Volumes at High Frequencies

Abstract

This thesis undertakes an investigation into time series at high frequency. The three main channels of information in high frequency data - irregular time intervals (durations), prices and volumes - are all explored and modelled to improve current understanding, while accounting for the long memory property, a crucial stylised fact found in the literature. In doing so, we make use of the theory of point processes, econometric techniques such as Whittle estimation and Kalman Filter forecasting, and also sophisticated computing architecture including database systems and programming languages across multiple software environments.