Links and Resources

Cutting-edge Research and Development.

Portfolio Defense has developed a number of innovative solutions and new approaches to predictive risk analytics over 20 years in the industry. We incorporate these new learnings into a variety of client projects.

Our work has included:
- the development of "hybrid" predictive variables which identify previously hidden interactions, improve predictive power yet simplify implementation;

- innovative new technique of Disparate Impact Optimization© which measures and reduces negative lending impacts on protected classes and vastly improves fair lending compliance/opportunities for community lending;

- testing and incorporating new data sources including debit bureau data, sub-prime bureaus, fraud data, and increased demographic data in models to increase model performance, reduce data cost, and remain readily implementable;

- Financial Distress Scores© permit portfolio managers to anticipate problem borrowers and create a strategic response to a recessionary economy.

- The development of Economic Risk Adjustors© to help adjust account management strategies based upon local economic conditions. For example, an account which has high Revolving Utilization might behave in a riskier fashion in an MSA with high unemployment, low income growth, HPI home prices falling and a high density of home foreclosures.

Risk Management Conference Materials.
New Data Sources for Sub-Prime Decisioning
Designing a CB-based Surrogate Model Development Database
Understanding Variable Interactions for Better Scorecards
eFunds DebitBureau Overview
Multiple Model Generations in Subprime Lending
Mark Zandi economy.com Presentation
Improving Response Model Results with Economic and Geographic Data
Mortgage Portfolio Models - the Competing Risks of Prepayment and Default
Home Equity Benchmark Results