The lack of a standardized metric for Automated Valuation Model (AVM) confidence scores has historically made it difficult to compare AVM models across vendors, because each vendor uses its own proprietary scale; a high score from one provider doesn’t necessarily align with a high score from another.
The MISMO® Common Confidence Score (CCS) addresses this by establishing a uniform, vendor-agnostic metric. By providing a consistent method for quantifying the reliability of AVM values, the CCS enables clearer risk communication and better decision-making across the industry. The following frequently asked questions explain CCS and how it works alongside other scoring metrics, such as the VeroVALUE® Confidence Score (VCS).
What is the MISMO® Common Confidence Score (CCS)?
The MISMO® Common Confidence Score (CCS) is a new, standardized, vendor-agnostic metric designed to quantify the reliability of automated property valuations or collateral assessments. Instead of relying on vendor-specific confidence indicators on an Automated Valuation Model (AVM), it provides a common, interpretable score that lenders, servicers, investors, regulators, and other housing finance professionals can consistently understand and compare across models, vendors, and use cases. The result is clearer risk communication, easier governance, and better decision-making, without dictating how any model is built or who supplies it.
Purpose of a CCS
An AVM estimate is a property’s market value at a point in time. A confidence score indicates how reliably an AVM value is expected to represent that market value.
The purpose of the MISMO® CCS is to provide a consistent, standardized way to communicate confidence across all AVM providers, helping stakeholders feel part of a unified effort to improve risk interpretation and decision-making.
Why is the CCS necessary?
Before the CCS, proprietary AVM confidence scores were not comparable across providers. Different providers use different scales and meanings (e.g., “92,” “B,” or “High”) for the same property, making direct comparison difficult. Likewise, a “92” may mean two completely different things for two different AVMs.
Many providers already rely on Forecast Standard Deviation (FSD), a statistically sound measure of uncertainty. For example, an FSD of 0.08 implies a 68% (one-standard-deviation) probability that the market value falls within ±8%. While consistent and meaningful, FSD is often considered too complex and difficult to explain for broad adoption. Moreover, assumptions required to calculate FSD, such as the normality of errors, are often not correct.
The CCS addresses this gap by preserving the rigor of FSD while providing a simpler, more interpretable score that does not rely on untenable assumptions, making confidence measurement more accessible for stakeholders unfamiliar with complex statistics. It also supports compliance with the Dodd-Frank Act’s AVM Quality Control Standards by enabling lenders to properly evaluate the risk of using automated valuations.
What does the CCS enable?
The CCS allows lenders, servicers, investors, regulators, and other housing finance professionals to:
- Compare confidence across AVM providers.
- Communicating risk using a shared language.
- Improve risk-based decisions, including those made by non-specialists.
Providers may continue to offer additional confidence indicators, such as FSD or proprietary scores, alongside the CCS, to ensure they feel supported in their current practices as they adopt new standards.
How does the CCS differ from the VeroVALUE Confidence Score (VCS)?
While both metrics measure the reliability of a valuation, they use different statistical “rulers.” The CCS is designed for industry-wide uniformity, while the VeroVALUE® Confidence Score (VCS) is optimized for the specific nuances of the VeroVALUE® engine.
Customers have long used the VCS as a quantifiable measure of valuation accuracy. VCS is interpreted using the median absolute error (MAE) metric. For example, a VCS of 95 correlates to an MAE of 5%, 85 to 10%, 75 to 15%, etc. In practice, a large group of valuations with a VCS of 95 will have an MAE of approximately 5%. Lenders commonly rely on VCS when using the VeroVALUE AVM.
The CCS is calculated similarly but correlates with P10 rather than MAE. P10 represents the percentage of valuations that fall within +/-10% of the subject property’s market value. As a result, while a VCS of 95 (MAE of 5%) is a typical score, achieving a CCS of 95 (P10 of 95%) is significantly more difficult and would be less likely to be seen in practice. Since P10 values in AVM testing commonly fall around 60% to 80%, CCS values of 60–80 generally indicate more typical performance.
The two scores use different statistical approaches, which naturally result in different score ranges.
At a Glance, Side-by-Side Comparison
Feature | MISMO® CCS | Veros® VCS |
Statistical Basis | Correlates to P10 (Percentage of valuations within plus or minus 10% of market value) | Correlates to MAE (Median Absolute Error) plus Proprietary Indicators |
Scores | A score of 60+ is considered strong. | A score of 90+ is considered strong. |
Best Use Case | Benchmarking different AVMs against a single industry standard. | Setting internal thresholds and cascades for VeroVALUE® AVMs. |
Does MISMO® CCS change how the VCS is calculated?
No, the VCS remains unchanged and continues to use the same proprietary methodology that customers rely on for model-specific confidence. The CCS is provided in addition to, not in place of, the VCS.
Is there any guidance to follow before using the MISMO® CCS?
Yes, the user should test that the CCS values from that AVM vendor correlate with P10 per the MISMO® standard. If they do, direct comparisons with CCS values from other similarly tested AVM vendors are possible. If they do not, the AVM vendor’s CCS values should not be used until they can reliably produce CCS values in accordance with the MISMO standard.
Will Veros continue to support the VCS in the long term?
Yes, Veros will continue to maintain and support the VCS as a core component of the VeroVALUE® AVM. The addition of CCS is about expanding customer choice, not replacing existing tools. VeroVALUE users leveraging VCS to establish AVM thresholds or as a part of AVM cascades will be able to continue doing so.
CCS Validation and Testing Guidance
Before relying on an AVM’s CCS, users should verify that the score accurately reflects the model’s performance in accordance with the MISMO standard. For example, if an AVM reports a CCS of 90 across many properties, the observed P10 should be near 90%. If observed performance is materially lower (e.g., 70%), the CCS may not be reliable for comparison across AVMs or for risk-based decisioning. It should not be used to compare with other AVMs’ CCS values.
In Summary
The adoption of the MISMO® CCS provides a more consistent framework for evaluating AVM reliability. Veros remains committed to supporting the VCS for internal cascades and model-specific performance, while offering the CCS to facilitate broader industry benchmarking.
For further clarification, contact support@veros.com or your Veros sales representative. Technical details and the full MISMO® AVM CCS Standards Reference Guide are available at: https://www.mismo.org/standards-resources/mismo-product/automated-valuation-model-common-confidence-score-standard-and-guidance
