Modelowanie procesu składania zleceń na niemieckim rynku energii elektrycznej 
intraday z wykorzystaniem procesów Hawkesa
[Modeling market order arrivals on the intraday power market for deliveries 
in Germany with the Hawkes process]

Nikolaus Graf von Luckner (House of Energy Markets and Finance, 
University Duisburg-Essen, Essen, D)


We analyze market order arrivals on the intraday power market for deliveries 
in Germany. In addition to distinguishing between buy and sell market orders, 
we also consider partitions such as market orders which cause the mid price 
and bid-ask spread to change and those which do not. The incorporation of such 
partitions gives rise to a multivariate setting with individual counting 
processes for each partition. We focus on whether Hawkes processes with 
time-varying baseline intensities and exponential kernels are suitable for 
modeling these counting processes. For this process class we provide 
analytical expressions for the conditional expectations of the intensities and 
introduce a new simulation method. In our empirical part we assess the 
goodness-of-fit and the out-of-sample forecasting performance. Our results 
indicate that the models where each component's intensity depends at least on 
its own history are the most promising ones.