Type of Publication: Article in Journal

Modeling the Serial Structure of the Hawkes Process Parameters for Market Order Arrivals on the German Intraday Power Market

Author(s):
Blasberg, A.; Graf von Luckner, N.; Kiesel, R.
Title of Journal:
16th International Conference on the European Energy Market (EEM)
Publication Date:
2019
pages:
1-6
Keywords:
Data models, Forecasting, Analytical models, Economics, Predictive models, Finance, Time series analysis, Intraday Power Market, Hawkes Process, VAR, Forecasting Performance, Trading Strategy
Digital Object Identifier (DOI):
doi:10.1109/EEM.2019.8916326
Link to complete version:
https://ieeexplore.ieee.org/document/8916326
Citation:
Download BibTeX

Abstract

Existing research indicates that on the intraday market for power deliveries in Germany market orders tend to arrive in clusters. To capture such clustering, point processes with an intensity depending on past events, so-called Hawkes processes, appear to be promising. We consider the question whether there is a temporal structure prevalent in the parameters of Hawkes processes estimated for adjacent delivery hours. First we model a diurnal seasonality pattern found in the data and provide an economic intepretation for it. For the remaining decomposed series, we then propose simple (vector) autoregressive models to describe the serial structure. To evaluate our model we conduct a forecasting study. Testing against a benchmark model and a model without any serial structure, we find evidence for our proposed model. Our study reveals that capturing the serial structure in the parameters proves to be useful in understanding the underlying market microstructure.