A0511
Title: Neural Hawkes: Non-parametric estimation in high dimension and causality analysis in cryptocurrency markets
Authors: Ioane Muni Toke - CentraleSupelec (France) [presenting]
Timothee Fabre - SUN ZU Lab (France)
Abstract: A novel approach to marked Hawkes kernel inference is proposed, which is named the moment-based neural Hawkes estimation method. Hawkes processes are fully characterized by their first and second order statistics through a Fredholm integral equation of the second kind. Using recent advances in solving partial differential equations with physics-informed neural networks, a numerical procedure is provided to solve this integral equation in high dimensions. Together with an adapted training pipeline, a generic set of hyperparameters is given that produces robust results across a wide range of kernel shapes. An extensive numerical validation is conducted on simulated data. Two applications of the method are finally proposed for the analysis of the microstructure of cryptocurrency markets. In a first application, the influence of volume is extracted on the arrival rate of BTC-USD trades and in a second application, the causality relationships and their directions are analyzed amongst a universe of 15 cryptocurrency pairs in a centralized exchange.
