Abstract
Bayesian Non-parametric modelling of durations in high-frequency finance
by
Anuj Mishra
In this paper, we propose an infinite regime Autoregressive Conditional Duration (ACD) model. The stationarity, moments and tail behaviour of the proposed model are studied. The conditional density of duration is modelled in a Bayesian approach by putting prior on the space of transition densities using Dirichlet process mixtures(DPM). We establish the asymptotic property of the Bayesian procedure by proving the consistency of the sequence of posterior distributions.
Committee
Workshop
Key Dates
Communication
StatFin Main Webpage