Navigating an ever changing tide

Navigating an Ever-Changing Tide

Date: June 22, 2020

Dear Colleague:

When VantageScore came on to the scene just about 14 years ago, a bright light was shown on the fact that consumers didn’t have just one credit score and that lenders have access to many different models that produce different scores for consumers.

The recent introduction of a new model from our competitor FICO and the ensuing media coverage points to the fact that there remains many questions and more education is clearly necessary. I thought I would take this opportunity to address some of the questions here.


First of all, why do we develop new models? While I cannot speak to our competitor’s motives, I can share that most credit scoring models use similar types of credit file information but we use that data differently by implementing modern mathematical architectures in an effort to build the most predictive model possible. How much weight is put on a particular characteristic (such as utilization of available credit, payment history, age and mix of credit activity, etc.) and how these are measured, will vary from one model to another, leading to differences in scores. Extensive testing is employed to verify the most predictive outcomes.

A useful analogy is to look at the different rankings for sports teams. Different rankings use different inputs and formulas to rank teams from best to worst, often in an effort to predict who will win the championship. At the end of the year, some rankings were more accurate than others.

Credit score models are similar. Multiple models are available to lenders, introduced at different time periods, and some will be more predictive than others. Newer models, in the case of VantageScore, use the latest innovations to be more predictive and accurate.


Changing economic environments also might compel a credit score model developer to introduce a new model. For example, if volatility is introduced into the system (such as the additional risk that was driven into the environment by The Great Recession), the need may arise to introduce new models to take that volatility into account. Indeed, we introduced VantageScore 3.0 as the economy emerged from The Great Recession and then launched VantageScore 4.0 after that economic impact was mostly washed out of the system.


In order to more accurately reflect the changing dynamics and to provide the highest predictiveness and accuracy, a model developer will update a model and, at the same time, take advantage of any new data types and technological advances that might add to the model’s ability to score the broadest possible consumer population in a safe and sound manner.

Let’s take our latest model as a case study. When we developed VantageScore 4.0, we recognized the predictive power of trended credit data. Trended credit data examines credit activity over a period of time (sometimes as much as 24 months) and more accurately reflects a consumer’s true creditworthiness by tracking the trajectory of credit behaviors instead of a snapshot of credit activity from the previous month.

VantageScore’s use of trended credit data recognizes and appropriately rewards those who are taking steps to improve their credit health over time. For example, the way VantageScore uses trended credit data rewards those with high utilization rates who consistently pay down outstanding balances as well as those who regularly pay more than the minimum amount due on outstanding balances. Conversely, trended credit data will alert a lender to early indicators of a deterioration in credit quality


With any model version update, credit scores are expected to shift because more predictive model would apply new – more accurate – scores to some consumers. With VantageScore 4.0, we saw that more consumer scores increased than decreased when compared with our earlier model, VantageScore 3.0, but these shifts aren’t as informative as they appear. Here’s why:

A new version of a credit score model only matters when a lender chooses to adopt it. And that takes a long time from when a model is initially introduced, and between such times scores continue to shift based on the updates to the credit files. The point is: these estimates of how scores will shift, will also shift over time!

From a consumer’s standpoint, I get it. Bart Simpson’s famous “Ay, caramba!” line comes to mind.

But here’s what consumers need to know:

  1. For the most part, credit score models like the VantageScore models are all based on information reported in the consumer credit files.
  2. All models will reward positive behaviors, such as on-time payments and low utilization rates on credit cards and reduce scores for consumers who have mishandled their credit accounts (e.g., missed payments and high balances). This basic foundation of credit scoring remains true whether the model uses trended credit data or not.
  3. Understand that the score you might check on your own isn’t likely to be the EXACT score a lender will use.
  4. You can use the free score to understand directionally whether you are likely to be approved for a loan and at what terms.
  5. And be sure to shop around for the best terms for your loan; make those lenders compete for your business!


Barrett Burns

CEO and President, VantageScore Solutions, LLC