top of page
  • Writer's pictureRick Haskell

Scorecards are a Game-Changer

Updated: Jun 21, 2023

Are you a small to mid-sized lender, perhaps a regional bank or credit union, or maybe an up-and-coming specialty finance company with big visions and big goals. If so, this article is for you. It is intended to be educational, and to help you take your company to the next level.


Arguably the most transformative technology in all of lending over the past 2 decades has been the adoption of scoring technologies. In the past, consumers applied for credit and were decisioned by a loan officer who reviewed the credit bureau, application and deal structure, and made a judgment call on who should be approved or declined. On the collections side, past due accounts were by and large treated the same in terms of contact strategy and timing, and critical decisions whether or not to repossess a vehicle for example were gut-feeling decisions, often made on emotion and with bias and inconsistency.


As scoring technologies came on the scene, the larger enterprise lenders began building custom models, called scorecards, to automate most of their mission-critical decision-making. The result of all this automation helped them to grow better-performing loan portfolios, faster, more accurately, and more profitably since staffing counts were dramatically reduced made possible by all this automation.


And so the risk management industry, previously having success in healthcare and medical circles, began infiltrating the financial services sector—bringing with it a false-narrative that statistical models are difficult to build, and require mathematicians with advanced degrees.


Truth-be-told, building these models is easy as pie with the right tools, and my experience is the most powerful scorecard models of all are those built by business experts with years of hard-earned experience in their trade. A mathematician cannot learn your business in days, but indeed—you can learn to build scorecards in a day or two.


Allow me to introduce Lendisoft MODEL BUILDER! The scorecard builder for the rest of us.


The industry primarily recognizes 2 types of scorecards:

  1. Expert Scorecards – Are structured exactly like Machine Learning Models, but you use your business acumen and hard-earned experience to choose the variables and set the points (weights). These are also known in the industry as judgmental scoring models. You can build unlimited Expert Scoring models in Lendisoft, and use them to calculate our risk-based tools like Odds-of-Roll, Likelihood of Default, Likelihood of Collectability, and more. Expert Scores also drive our Risk Grades (A-E), which can be used to schedule email and SMS campaigns, and even control the timing around agents calling precise risk-segments within your past due accounts. Expert Scores also unlock the ability to do skill-based assignments, empowering you to assign your toughest accounts to your top collectors, etc.

  2. ML Scorecards – Are logistic regression derived models (or other machine-trained method)—ours are powered by IBM SPSS, and use a “supervised machine learning” algorithm that automatically picks the most predictive variables and optimally sets the points (weights) in your model. ML models are known to be more powerful than expert scores since they are empirically derived models and machine-trained.

In either case, the use of scoring models in your day-to-day operations will help you optimize your decision making and drive efficiency.

  • If you intend to build Expert Scorecards and you are a Lendisoft client, you can do so right out-of-the-box in all 3 of our product variations. Lendisoft LMS is a modern LMS built around risk management and collections optimization. If you are looking for a modern LMS that’s 100% cloud-based and a no-code solution with advanced customization options, look no further than Lendisoft LMS!

  • If you desire the enhanced power of ML models, we offer 3 ways to bring ML scoring technologies to your organization:

    1. Subscribe to Lendisoft Model Builder and get 1 full year of unlimited scorecard development for $10,000 (annual subscription). Lendisoft Model Builder facilitates the development of each ML scorecard as a project, and gives you a dedicated workspace with predefined templates to carry out the development steps in sequential fashion. Logistic regression is performed using IBM’s market-leading SPSS statistics tool for maximum accuracy and predictive power.

    2. Let Lendisoft build your scorecard - While other scoring consultants charge upwards of $30,000 per scorecard model, our development cost is only $10,000 per scorecard. We use industry best-practice modeling techniques, and perform logistic regression using IBM’s market-leading SPSS statistics tool for maximum accuracy and predictive power.

    3. One-on-One Training – For the ultimate learning experience, we’ll train you by building your ML scoring model together, and interactively. This includes 2 all-day sessions, followed by multiple homework and review sessions as you build your first scoring model under our guidance and tutelage. This service includes a 1-year subscription to Lendisoft Model Builder. We will train you on IBM’s market-leading SPSS statistics tool for maximum accuracy and predictive power. The cost for this exclusive engagement is $25,000.

For more information about our scorecard development, risk management tools, and training services, please contact us at (949) 888-5556.

Comments


Commenting has been turned off.
bottom of page