PROBABILISTIC MODELING OF SELF EXCITING COUNTING PROCESSES WITH LATENT RISK RELAXATION

Main Article Content

HUYNH HUU DINH
TRINH PHUOC TOAN

Abstract

We study a class of self exciting counting processes with latent risk relaxation. Unlike a classical Poisson process, our model allows each event to temporarily increase the chance of future events. We couple the event count with a hidden risk variable that increases at event times and decreases between events. We derive the evolution equation for the joint law, obtain closed equations for the first order moments, and derive formulas for event times and inter event times. The model also provides a link between Poisson type dynamics, accumulation based self excitation, and exponential memory effects.

Article Details

Section
Basic Science, Culture, Art