by Mark Dreux, DMS
“An approximate answer to the right problem is worth a good deal more than an exact answer to an approximate problem.” – John Tukey
Do you wonder if your credit model could be better if it was refit? Knowing when and if to commit to the time and expense of a full model rebuild challenges many lenders. But according to defi SOLUTIONS partner Digital Matrix Systems (DMS), there is an easy solution.
Creating Investigative Scoring Models
DMS FastTrack Modeling™ can help you determine the expected lift of a new credit model at a fraction of the time and cost it would take to build and test it. FastTrack Modeling uses archived credit data and DMS attributes to create an investigative scoring model. In as little as one week, this unique process can show you whether your existing credit models can be improved. The FastTrack Modeling process provides an efficient method of performing an initial analysis on the lift expected from a new model and lets you see the value of the model before committing to the full cost of building it. If the lift isn’t sufficient, you can direct your time and money elsewhere.
Client Case Study
A FastTrack model recently beat one client’s existing model by more than 20%. On average clients have experienced a 15-20% gain in volume and 10-15% reduction in overall losses while maintaining current marginal risk.
Of course the client decided to move forward with the full model build and, when all was said and done, they found that the actual lift of the final model exceeded the expected FastTrack results. Furthering the benefits, DMS employed the FastTrack process for the client to complete a cost-effective evaluation of an alternative data source to help determine the value adding the data source to the client’s portfolio services.
How Does It Work?
Using historical credit data, a core set of DMS attributes, and performance data, FastTrack Modeling uses a proprietary automated modeling tool to create a “quick and dirty” challenger model to your existing scoring model. The two models are compared using measures such as C values, KS, and gains tables to estimate the lift of a full modeling effort. C value and KS are two commonly used statistics that summarize how effective a model is at rank ordering a population of applicants from low to high risk. If the lift is sufficient, DMS can complete the full modeling process for you that includes these steps:
- An extensive data cleaning
- A robust compliance review
- Inclusion of all attributes
- A detailed validation
- Implementation of the new model
The best part? DMS leverages the work done in the FastTrack Modeling process to help reduce your overall project cost.
DMS, the BIG KAHUNA sponsor at this year’s defi FEST 2017, demonstrated FastTrack Modeling during its partner roundtable session and also presented “DMS in the TestLab” at a special session during the event. TestLab enables clients to take control of their test environments by simulating credit bureaus with both customized and reusable test data.
For more information about FastTrack Modeling, go to: www.dms.net/fast-track-modeling
Check out defi Partner Spotlight on DMS.
Get a copy of the DMS FastTrack Modeling presentation here.
