A0422
Title: High-frequency market manipulation detection with a Markov-modulated Hawkes process
Authors: Ioane Muni Toke - CentraleSupelec (France) [presenting]
Timothee Fabre - SUN ZU Lab (France)
Abstract: The focus is on a self-exciting point process defined by a Hawkes-like intensity and a switching mechanism based on a hidden Markov chain. Previous works in such a setting assume constant intensities between consecutive events. The model is extended to general Hawkes excitation kernels that are piecewise constant between events. An expectation-maximization algorithm is developed for the statistical inference of the Hawkes intensities parameters as well as the state transition probabilities. The numerical convergence of the estimators is extensively tested on simulated data. The model is applied to high-frequency cryptocurrency market data on a top centralized exchange. The focus is on two high-frequency market manipulation strategies which are wash trading using many orders and pinging. The goodness of fit of the model is benchmarked with the Markov-modulated Poisson process, and the effectiveness of the model is demonstrated in detecting past suspicious activities over a large out-of-sample dataset.
