Hoping someone can help me with the two questions below.
1) What is the kernel SimuHawkes() is expecting? I have fitted HawkesEM on 90 percentile daily log loss returns for some stocks. This gives me the kernel and a baseline, but plugging in this kernel in SimuHawkes does not seem to work. Is it possible to simulate from a fitted (non-parametric) kernel? I would like to simulate based on the HawkesEM/HawkesConditionalLaw fit if possible, as they make no assumption on the kernel shape.
2) How does tick treat the time between events. I noticed especially with HawkesConditionalLaw that the fit (norms) can vary extremely if I divide/multiply all event times by the same constant factor x. Should I tweak one of the parameters when working with daily data rather than in the Bacry paper where they use tick data?
Hi,
Hoping someone can help me with the two questions below. 1) What is the kernel SimuHawkes() is expecting? I have fitted HawkesEM on 90 percentile daily log loss returns for some stocks. This gives me the kernel and a baseline, but plugging in this kernel in SimuHawkes does not seem to work. Is it possible to simulate from a fitted (non-parametric) kernel? I would like to simulate based on the HawkesEM/HawkesConditionalLaw fit if possible, as they make no assumption on the kernel shape. 2) How does tick treat the time between events. I noticed especially with HawkesConditionalLaw that the fit (norms) can vary extremely if I divide/multiply all event times by the same constant factor x. Should I tweak one of the parameters when working with daily data rather than in the Bacry paper where they use tick data?
Thanks in advance!