KBRA Altman Launches High Yield Return Forecasting Model: A Differentiated Approach Using Machine Learning Allows the Model to Keep Pace with Changing Market Conditions

KBRA Altman, a division of KBRA Analytics, is pleased to announce the launch of its new High Yield Return Forecasting Model. Unlike traditional approaches that examine bivariate relationships between the level of credit spreads and other economic variables, the model uses machine learning techniques applied to a large data set of economic and market variables. The model identifies historical forecasting relationships between those variables and future returns on a diversified index of high-yield corporate bonds.

The advantages of the model are severalfold. First, in evaluating its efficacy, we do not allow model forecasts to use forward-looking data. This simulates investor behavior since investors don't have access to future information when making portfolio decisions. Second, while the model starts with many potential forecasting variables, it selects a small subset of these which proves to be the statistically "best" one. Third, the model changes over time. The world changes over time, and a model that seeks to predict future financial outcomes cannot be static. A consequence of this is that today's forecasting variables may not be tomorrow's forecasting variables. Finally, the model is easy to interpret and apply. Given the current values of the best set of forecasting variables, the model produces a year-ahead forecast for the return of a diversified index of high-yield corporate bonds.

The year-ahead return forecast now stands at 1.6%, suggesting that the high yield asset class is offering paltry returns, reflective of the low rate environment, stretched valuations, and still evident stimulus-fueled environment. The current best set of forecasting variables includes HY spreads, the level of US T-bill rates, the lagged default rate on HY bonds (as tracked by KBRA Altman), HY realized return volatility, and a measure of aggregate corporate leverage in the US.

KBRA Altman will publish quarterly reports providing updated model HY return forecasts, as well as detailed analysis of the model components. Subscribers of KBRA Altman can access the report here. We'll discuss the model and the current model forecast at our upcoming webinar "KBRA Altman: Picking Up the Pieces: Perspectives on Credit Markets Post-COVID."

Through its exclusive collaboration with Professor Edward Altman, PhD, KBRA Altman provides in-depth analysis of the defaulted U.S. corporate bond and loan market via a collection of price indices and data that track current and historical defaults as well as recovery experiences of corporate issuers. Click here to sign up for free access to the KBRA Altman portal where you can view our insights and topical research.

About KBRA Analytics
KBRA Analytics, LLC (KBRA Analytics) is our premier product platform for high quality data and advanced analytics. Our seasoned teams of industry specialists across each product provide unparalleled insight creating a foundation of deeper analysis and rapid discovery for users. KBRA Analytics is an affiliate of Kroll Bond Rating Agency, LLC (KBRA). KBRA is a full-service credit rating agency registered in the U.S., designated to provide structured finance ratings in Canada, and with credit rating affiliates registered in the EU and UK.