Forecasting & Stress Testing — Best in Class Solutions

Prescient Models is an industry leader in loan and deposit forecasting in the financial services industry. Our recent paper with the Federal Reserve Bank of Philadelphia used some of our recent innovations to review otherwise hidden origins of the US mortgage crisis.

Services

Prescient Models has a flexible business model where we adapt our solutions to the needs of our clients. We provide

Model development, either independently or collaboratively with your in-house team.
Modeling libraries for licensing, to kick-start your in-house development.
Onsite training courses, to team your team from basic to best practices.
Model validation, to provide an expert second-opinion on models built in-house or by others.

Applications

Single-purpose models are over. Our technology can be applied to a range of bank products and creates models that can be used for many purposes.

We routinely create models for

Retail loans: card, auto, mortgage, home equity, personal loans, personal lines, and student loans
Small business loans
Commercial real estate (CRE)
Time deposits
Demand deposits

We excel at models that provide monthly or quarterly forecasts for short-term to long-term pricing, account management, loss reserves, and valuation. Our clients typically use our technology to predict

Delinquent accounts and balances
Account attrition and balance pay-down
Default (Charge-off) accounts and balances
Recoveries
Revenue
Net Margin
Net present value

Technology

The Prescient Models team is renowned for its expertise in retail lending, vintage models, and stress testing. Vintage-aggregate models are exceptional in the accuracy-complexity trade-off, but they don't support account-level actions.

Prescient Models is leading the industry again with the introduction of account-level Age-Period-Cohort models + behavior scoring. Using the technology highlighted in the Philly Fed paper above and detailed in our recent academic paper, we have bridged the gap being vintage-aggregate models and scoring models. Rather than scores that just rank-order, our models incorporate macroeconomic scenarios to provide a direct monthly, account-level forecast of any of the items listed above.