Stochastic Modelling with Randomised Markov Bridges

TL;DR

This paper develops explicit filtering formulas for estimating hidden variables using noisy observations from randomised Markov bridges, with applications in finance and commodity pricing.

Contribution

It introduces a general filtering approach for randomised Markov bridges and constructs a skew-normal bridge example for financial modeling.

Findings

Explicit filtering formula derived for RMB-based observations

Skew-normal RMB constructed and analyzed

Applications demonstrated in commodity and emission price models

Abstract

We consider the filtering problem of estimating a hidden random variable $X$ by noisy observations. The noisy observation process is constructed by a randomised Markov bridge (RMB) $(Z_t)_{t\in [0,T]}$ of which terminal value is set to $Z_T=X$. That is, at the terminal time $T$, the noise of the bridge process vanishes and the hidden random variable $X$ is revealed. We derive the explicit filtering formula, governing the dynamics of the conditional probability process, for a general RMB. It turns out that the conditional probability is given by a function of current time $t$, the current observation $Z_t$, the initial observation $Z_0$, and the a priori distribution $\nu$ of $X$ at $t=0$. As an example for an RMB we explicitly construct the skew-normal randomised diffusion bridge and show how it can be utilised to extend well-known commodity pricing models and how one may propose novel stochastic price models for financial instruments linked to greenhouse gas emissions.