FICO® Score 10 T Decisively Beats VantageScore 4.0 on Predictability

An analysis by FICO data scientists has found that FICO Score 10 T significantly outperforms VantageScore 4.0 in mortgage origination predictive power.  An independent third-party study found that VantageScore 4.0 generates a minimal predictive improvement over Classic FICO – so minor that, when properly adjusting the comparison due to a truncation anomaly described below, it should be questioned if VantageScore 4.0 even beats the 20-year-old Classic FICO at all.  At the same time, FICO Score 10 T far exceeds Classic FICO in detecting loan losses.  In fact, FICO Score 10 T’s improvement over Classic FICO was shown to be five times better than VantageScore 4.0’s improvement, with FICO Score 10 T detecting 18% more defaulters in the critical score decile commonly used for mortgage originations versus just 3.4% for VantageScore 4.0. 

Truncated Data: VantageScore’s Illusion of Performance

Importantly, both Classic FICO and VantageScore 4.0 have been compared on conforming mortgage data, which truncates the performance calculation of Classic FICO below the 620 cut off. While the performance of Classic FICO should be measured through its entire score range from 300 to 850, the historical database only allows for measuring the strength of Classic FICO from 620 and 850.  This truncation effect negatively impacts the calculation of Classic FICO’s strength by 15% or more, artificially boosting VantageScore 4.0’s performance in comparison – a boost big enough that it calls into serious question whether VantageScore now actually beats Classic FICO at all.

VantageScore 4.0 Punishes Non-Homeowners—FICO Score 10 T Does Not

FICO Score 10 T handily wins despite VantageScore’s attempt to improve the performance of VantageScore 4.0 by including mortgage-specific variables in its model.  But, including mortgage-specific variables penalizes people who have never owned a home. Under VantageScore millions of Americans—including young people, members of the military, and people from disadvantaged groups—will have lower scores than they otherwise would, merely because they have never owned a home. That’s unfair. Forcing lenders to tell people from disadvantaged groups through adverse action notices that they have been rejected for a mortgage because they don’t currently own a home is irresponsible.  FICO Score 10 T includes rental data while not penalizing people for not owning a home. 

Impact of VantageScore’s Changes to VantageScore 4.0 After Submission to GSEs

Nearly two years after both FICO and VantageScore submitted their models for evaluation by the GSEs and FHFA for accuracy and reliability, VantageScore decided to modify VantageScore 4.0 to exclude a large category of predictive credit file information – Medical Debt Collection Records.  VantageScore’s post validation change raises serious questions of a potential breach of the integrity of the FHFA and GSE Validation and Approval of Credit Scores process that would impact the VantageScore/Classic FICO comparison cited in this white paper.  Not only does it undermine determining whether VantageScore 4.0 now even outperforms Classic FICO, it raises serious doubt of whether the VantageScore 4.0 model evaluated by the GSEs and FHFA under the Joint Enterprise Credit Score Solicitation process is even the same VantageScore model used to generate scores in the historical dataset made available by the GSEs for industry modeling and transition.  Most troubling, it creates concerns about whether the current new model was even approved through the required process.  If model developers are allowed to discretionarily change their models after GSE validation, no constraints would remain for any model developers to refrain from altering their models to curry favor with stakeholders, including lenders who transfer the risk to the GSEs and taxpayers. In fact, they would be highly incented to do so.  The race to the bottom is no longer theoretical and, if allowed to continue, could have significant and irreversible impacts on the mortgage ecosystem. 

More Loans, Better Pricing with FICO Score 10 T

Based on enhanced prediction, use of FICO^® Score 10 T rather than VantageScore 4.0 by market participants throughout the mortgage industry will drive significantly more loan approvals for prospective borrowers and, due to better model performance for mortgage insurers, investors and others, improve mortgage pricing and lower costs for borrowers—benefiting millions of Americans.

Given this superior performance, it’s no surprise that lenders representing over $300 billion in annual mortgage originations and $1.5 trillion in mortgage servicing portfolios have already adopted and started trading loans with FICO Score 10 T outside of the GSE conforming market. 

The Bottom Line

In head-to-head competition, VantageScore is the loser and it’s not even close. The mortgage industry deserves better than a runner-up score. FICO Score 10 T isn’t just better – it’s the trusted standard. It wins on predictive accuracy. It wins on fairness. For lenders, investors, and consumers, and any stakeholder concerned about safety and soundness, FICO Score 10 T is the clear choice. 

Click here to download the full white paper: FICO® Score 10 T Materially Outperforms VantageScore 4.0