Predictive Modeling

Predicting customer behavior is a critical step in managing the risk/reward balance. Accurate prediction of customer behavior is paramount to proper risk/reward decisions as well as critical input to proper financial accounting.

A key decision is needed each time a business creates a new product offering, approves a loan, works-out a default or prepares a loss reserving estimate.  In each case, the cost and revenue quantities should be completely understood a priori. 

Predictive model development has many facets and can span the range from simplistic to extremely complex.  We specialize in predictive tools that support prepayment, delinquency, default and loss-given-default (LGD) outcomes.  Solutions are developed using a wide range of modeling techniques primarily based on statistical methods including logistic regression, survival models and discriminate analysis.

We build predictive models for a wide range of applications essential in mortgage banking, consumer lending and enterprise risk management areas. The Predictive Modeling solution includes the following components that produce a rigorous and stable framework:

• Predictive Modeling
  • Predictive model design, development, implementation and testing
    
    
• Scorecard / Model Discriminant Power
  • Performance measures that determine the “fit” of the model
    
    
• Scorecard / Model Stability
  • Monitoring measurements that determine the stability of your model
    
    
• OCC and external audit consideration