ML and Bond Pricing
Bond origination and OTC trading continues to be manual, risky, and problematic. The problem comes from figuring out how to price the bond. It is not easy to measure credit risk. There are multiple liquidity buckets and different issuers. Credit risk assessment is a major problem and it impacts price discovery.
Measuring credit pricing risk and valuation requires using many different datasets. Dynamic monitoring of credit risk implies using dynamic data that includes factors such fundamentals, alternative data, transactional data, and behavioral data to price risk.
Many models are being developed and deployed. Some are based upon older machine learning techniques including support vector machines, regression, and k-nearest neighbor. Deep learning and neural networks are now being used to improve the outcomes.
AIAI offers a comprehensive report on these developments.