June 06, 2023 | Stamford, CT — Artificial intelligence (AI) and machine learning (ML) are transforming the way investors trade highly liquid fixed-income products like U.S. Treasuries. Over a longer horizon, AI/ML could have an even bigger impact on markets for less liquid and transparent products like mortgage-backed securities, municipal bonds and interest-rate derivatives.
In liquid “flow” products like U.S. Treasuries and fixed-income ETFs, AI/ML are moving markets in the direction of trade automation in much the same manner as in global equity markets. For complex products such as municipal bonds, MBS and CLOs, AI/ML is used to enhance predictive pricing analytics and the management of market, credit and liquidity risk.
“AI/ML can be particularly useful for highly nuanced securities like municipal bonds,” says Audrey Blater, Senior Analyst for Coalition Greenwich Market Structure & Technology and author of Fixed-Income Markets Want In on the AI/ML Craze. “The technology can determine the relationships between factors without prior beliefs about the data—circumventing some of the pitfalls of highly deterministic models that wind up being spectacularly wrong.”
In the muni market, there are many price determinants including tax, municipality, state, and other considerations, making the risk profile of each security unique. Changes in municipal bond pricing are related to the overall municipal market and tax codes, local nuances, state nuances, a host of idiosyncratic risks, and more. AI/ML has the ability to constantly evaluate and learn from new data faster and more accurately than humans and, perhaps, other types of historical pricing models and proxies.
In the case of the U.S. mortgage market, AI/ML can be used to more accurately explain and predict MBS pricing and risk. Advanced technology is capable of analyzing millions of loan documents—many of which are submitted on paper. Now, intelligent document processing allows market participants to aggregate and understand the underlying fundamentals of MBS through document digitization.
Biggest Impact of AI/ML Yet to Come
Approximately 60% of mortgage traders and 41% municipal bond traders believe they have already fully adopted AI/ML. However, Coalition Greenwich is skeptical about the accuracy of those assertions, as there are still widespread misunderstandings of what AI, and especially ML, are really all about.
“In many cases, rules-based programming and other types of models are responsible for analytical and predictive outcomes—not true AI/ML,” says Audrey Blater. “Nevertheless, there is an entire universe of fixed-income products and markets that will eventually be transformed with AI and ML—including the massive market for interest-rate derivatives."