HSBC Developed Treasury Yield Prediction Tool

The bank launched a machine-learning model to forecast the direction of 10-year Treasury yields.

Updated on Sept. 22, 2026 in Stock Markets

Isometric editorial illustration of interlocking geometric metal plates, representing a structured framework for predictive financial data analysis. — HSBC has launched an automated tool named DUSTIN designed to forecast future movements in 10-year U.S. Treasury yields for market participants. AI Illustration. Upload story photo >

Live Poll

Would you trust a machine-learning model's projection to guide your personal investment decisions?

HSBC has designed a machine-learning model named the Directional U.S. Treasury Indicator, or DUSTIN, to predict future movements in 10-year Treasury yields. The tool currently assigns a 28% probability to the likelihood of yields rising.

Why it matters

The model aims to provide market participants with a predictive framework for interest rate shifts by analyzing complex market dynamics. It specifically highlights the relationship between yield curve changes and long-term instrument performance.

The DUSTIN model achieves 65% out-of-sample accuracy while utilizing 17 distinct variables, including momentum and volatility metrics. It currently forecasts a decline in 10-year Treasury yields over the coming weeks.

The players

HSBC

HSBC is a multinational banking and financial services organization that provides a wide range of corporate and investment banking products.

The details

By processing inputs like credit performance and market momentum, the model converts raw data into ranked variables to compare disparate market scales. HSBC reports that rising 2-year yields currently serve as a counter-indicator for the 10-year trajectory.

The model provides a forecast window covering the next 21 trading days.

Market Dynamics

This tool follows the pattern set by other algorithmic trading platforms in applying machine learning to macro-interest rate forecasting. It reflects a broader shift toward automated, data-ranked analysis within the global fixed-income markets.

Retail and institutional investors can use these indicators to adjust their bond portfolio allocations based on projected yield trends. The forecast for declining 10-year yields over the next 21 trading days may influence timing decisions for fixed-income purchases.

The takeaway

Advanced modeling allows banks to better quantify the relationships between different points on the Treasury yield curve. Investors should consider these automated forecasts as one of several signals when navigating interest rate volatility.