search

X-DataInitiative / tick

Module for statistical learning, with a particular emphasis on time-dependent modelling
https://x-datainitiative.github.io/tick/
BSD 3-Clause "New" or "Revised" License
484 stars 105 forks source link

Latency for parametric kernels #454

Open MarcosCarreira opened 4 years ago

MarcosCarreira commented 4 years ago

For financial data, it would be useful to specify a latency, such that HawkesKernelExp(intensity, decay, latency=0)would be equal to $\phi\left(t\right)=\alpha\cdot\beta\cdot e^{-\beta\cdot (t-t{0})}$ for $t>t{0}$ since the triggering of a new event cannot happen faster than a certain latency (from order book to participant + processing time + from participant to order book); or also because of the granularity of the time stamps.

MarcosCarreira commented 4 years ago

A clear example is attached:

NeedForLatency.pdf

MarcosCarreira commented 3 years ago

Addressed in the paper "Exponential Kernels with Latency in Hawkes Processes: Applications in Finance" https://arxiv.org/abs/2101.06348