Grand Unified Tech

Understand the borrower.Anticipate the cash flow.

INTELLIGENCE FOR AGENCY MBS

Complex markets.
Clearer foresight.

Prepayments shape the life of a mortgage-backed security. We introduce a unique Deep-learning prepayment model designed for MBS pricing and OAS calculation.

Explore our solutions ↗
Mortgage expertise.Machine intelligence.Unified.
PREPAYMENT INTELLIGENCEILLUSTRATIVE MODEL

Ahead of the curve.

Different rate paths. A clearer view of possibilities.

Conditional prepayment rate (CPR)%
Illustrative prepayment forecastThree hypothetical prepayment paths diverge over twelve months. These curves are for design illustration only, not actual model output.24181260JANAPRJULOCTDECFORECAST
HistoricalBase caseRate scenarios
FROM BORROWER SIGNALS TO PORTFOLIO PERSPECTIVE01 / 03

OUR FOCUS

01

Model Performance

Model how borrower behavior and changing market conditions may influence agency MBS prepayments.

BEHAVIOR → FORECAST
03

Model Future Forecasting

Provide next-month prepayment forecasts, enabling users to monitor model performance in real time

SIGNALS → UNDERSTANDING

THE GRAND UNIFIED APPROACH

Domain knowledge.
Deep-learning.
One perspective.

Mortgage markets are driven by people, incentives, and change. Our focus is bringing those dimensions together through Deep-learning.

01

Agency Mortgage Data

Fannie Mae Single-Family Loan Performance Data since 2009 and Freddie Mac MBS loans since 2021 are used for our Deep-learning Prepayment Model Training

02

Deep-learning Model Training

Two model versions (202412 and 202609), sharing the same deep learning architecture, are trained on data through December 2024 and September 2026, respectively, to demonstrate model robustness.

03

Prepayment Projection

The model forecasts Agency MBS prepayment behavior throughout the life of the bond, enabling reliable cash flow projections and option-adjusted spread (OAS) valuation.

Forecast visualization is illustrative and does not represent live data or validated model performance.