Mid-price estimation for European corporate bonds: a particle filtering approach

TL;DR

This paper introduces a Bayesian particle filtering approach to estimate mid-prices of European corporate bonds in illiquid markets using real-time partial information available to dealers.

Contribution

It develops a novel particle filtering method tailored for real-time mid-price estimation in illiquid bond markets, leveraging partial market data.

Findings

Effective real-time mid-price estimation demonstrated

Method outperforms traditional static models

Applicable to other illiquid asset markets

Abstract

In most illiquid markets, there is no obvious proxy for the market price of an asset. The European corporate bond market is an archetypal example of such an illiquid market where mid-prices can only be estimated with a statistical model. In this OTC market, dealers / market makers only have access, indeed, to partial information about the market. In real time, they know the price associated with their trades on the dealer-to-dealer (D2D) and dealer-to-client (D2C) markets, they know the result of the requests for quotes (RFQ) they answered, and they have access to composite prices (e.g., Bloomberg CBBT). This paper presents a Bayesian method for estimating the mid-price of corporate bonds by using the real-time information available to a dealer. This method relies on recent ideas coming from the particle filtering / sequential Monte-Carlo literature.