Point Processes with Coincidences and High-Frequency Finance

• Activity Number: 396
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Victor Solo*+ and Ahmed Pasha
• Companies: University of New South Wales and University of New South Wales
• Keywords: point process ; hawkes model ; high frequency data

Abstract

Point processes are an emerging tool in financial econometrics. High frequency financial data provides trade times to a milli-second (ms) time scale and point process models are the right way to describe this kind of data without losing information. However almost all point process models assume no-simultaneity (the technical word is orderliness) i.e. in a small time interval, with probability near 1, no more than 0 or 1 events can occur i.e. events cannot be coincidental i.e. events cannot be simultaneous. This is true e.g. of the Poisson process and the Hawkes process which is the main current tool of analysis. But even with ms data, because of the large number of financial markets and the huge range of assets being traded, scores and even hundreds of events can occur within a single ms i.e. essentially simultaneously. We discuss construction and fitting of models that allow simultaneity i.e. coincidences and illustrate with simulation and real data.