Title: On Hawkes processes
Authors: Matthias Kirchner - ETH RiskLab Switzerland (Switzerland) [presenting]
Paul Embrechts - ETH Zurich (Switzerland)
Abstract: Event streams are shown to have become an increasingly important data category. The mathematical counterparts of empirical event stream data are point processes on the real line. Hawkes point processes form a most successful model class. We motivate and explain the model. In finance and insurance, one typically considers multitype event stream data. e.g., different orders in limit order books, price jumps of different stocks, or credit defaults in different industries. We illustrate how in these kinds of Hawkes process applications, graphical description of data and models turns out to be useful.