The advantage of leveraging multiple AVMs for property valuation

By ICE Mortgage Technology

August 11, 2025

As automated valuation models (AVMs) continue to become more advanced, real estate and mortgage industry professionals must transition from solely relying on the more traditional manual appraisal methods to adopting modern valuation solutions. These can help reduce costs, streamline workflows, and better inform business decisions earlier in the property valuation process. Further, the right AVM solutions can help provide more efficient and credible options when it comes to property valuation.

AVMs can be used independently or as part of a solution that leverages multiple models. Utilizing the latter option can help provide the user with optimal results by covering a wider range of valuation metrics each AVM specializes in, such as accuracy, coverage and hit rate.

In this article, we discuss the benefits of using a solution with multiple AVMs. For more on AVM basics and the role they play in assessing property value, read our blog, What is an AVM? Everything you need to know.

Using multiple AVMs to determine a property’s value

A single AVM can be a capable method of obtaining property value, but using a solution that pulls from multiple AVMs can be a more reliable, cost-effective valuation strategy. Valuation solutions that leverage multiple AVMs often use a “waterfall” approach, which is a structured sequence of AVMs that pulls from multiple or differing data sources, mathematical models and geographical criteria to help provide the right property valuation model output based on user-defined criteria.

Using a waterfall approach can help support certain valuation metric requirements needed to maintain compliance, meet satisfactory confidence thresholds and secure more reliable valuation outputs. Additionally, an individual AVM may perform differently based on the geography, property type and living area. By leveraging a solution with multiple AVMs or an AVM waterfall, these three factors are already taken into consideration to provide the most appropriate valuation output for the specificities of the subject property.

The waterfall method employs a four-step approach to obtain the property valuation needed:

Define valuation criteria: The process begins with pre-determined criteria such as data availability, property characteristics (e.g., geographic location or type), confidence scores or error margins. These parameters help determine which AVM model within the waterfall will provide the most reliable results.
Rank AVMs by performance: AVMs are then evaluated based on various performance indicators. These benchmarks include model accuracy in predicting property values, as well as ability to meet user requirements for stability, reliability and compliance.
Execute the desired valuation result: The waterfall then evaluates all AVMs in its stack based on the user-defined criteria, calculating and considering all valuation models until an acceptable output is achieved, based on desired criteria.
Validate the output: The output then undergoes further validation to confirm if it meets the quality and accuracy standards of the stakeholder. This process may involve methods such as verifying that the confidence score surpasses a predefined threshold. In some cases, stakeholders may choose to conduct additional third-party testing.

Advanced solutions to drive property valuation accuracy

Using a solution with multiple AVMs or an AVM waterfall may offer a more reliable approach to obtaining a property value than using a single AVM. However, its success depends on a rigorous testing and validation process, as well as consistent monitoring. This helps to provide valuation results that are transparent, consistent and accurate, fostering stakeholder confidence while supporting regulatory requirements.

Types of AVM testing

User acceptance testing (UAT): User acceptance testing validates the valuation output through their individualized testing methods. Testing should also be subject to any form of randomization or any type of random specification that either a stakeholder or independent testing agency can bring to the table. This process helps to provide transparent and objective results.
Performance benchmarking: Performance benchmarking analyzes the accuracy of each AVM against historical property data and regularly compares the AVMs used against other models available on the market to determine accuracy and hit rates.
Outlier analysis: Outlier AVM outputs are analyzed to detect and resolve inconsistencies in property valuations. This helps to establish optimal AVM performance across various market conditions, property types and regions.

Continuous monitoring

AVMs are consistently monitored to produce highly accurate results and lower the number of outlier valuation outputs. Maintenance involves continuous tracking of model performance and market conditions. Any economic shift may influence data accuracy. Regular updates and refinements to the models’ algorithms based on this tracking can help optimize valuation outputs.

Transparency

Lenders and real estate professionals need to engage with an AVM provider that regularly discloses their data sources and processes.

How ICE’s Property Valuation Model can provide accurate and objective property values

ICE’s Property Valuation Model leverages multiple ICE AVMs and selects the most appropriate output based on a variety of factors, including the geography of the property and model’s preference data, resulting in optimal performance. Each of the individual AVMs rely on the same data but use different algorithms and techniques when arriving at valuation estimates. Property Valuation Model delivers automated, configurable preference data based on user-defined rules to determine which model to leverage for the subject property.

Increased efficiency

By deploying our Property Valuation Model, users can seamlessly automate property valuations across different scenarios in an efficient and data-driven manner. This drastically reduces the turnaround times for property valuations and minimizes inconsistencies that may exist in other valuation methods.

Enhanced accuracy and regulatory compliance

Using multiple AVMs is an improved way to obtain high-quality results and support regulatory compliance. By leveraging various data sources and methodologies, the risk of inaccurate or inconsistent valuations can decrease significantly. This is due to covering a wide range of various valuation performance metrics with multiple AVMs that have been evaluated for accuracy, data integrity, objectivity and subject to random testing.

Broader coverage

Different AVMs specialize in unique property segments or market conditions. By leveraging multiple models, Property Valuation Model helps provide greater geographic and property-type coverage, which can eliminate blind spots caused by the limitations of utilizing only one model.

Expedited turn times

Property Valuation Model also helps streamline the property valuation process, reducing the dependency on manual processes. The solution can also be used to assist, inform, expedite and in some cases be used in place of the traditional appraisal process. A quicker process can lead to faster decision-making and reduced costs for lenders and other users.

Risk mitigation

Property Valuation Model reduces reliance on a single model, thereby mitigating risks linked to limited data or inherent model biases. Leveraging multiple AVMs helps provide robust valuations, even in volatile or atypical market conditions. Proper testing and ongoing monitoring of the ICE AVMs that make up Property Valuation Model is also key to reducing compliance risk and securing the high hit rates and accuracy users desire.

Configurability

Property Valuation Model provides flexible configuration of preference data to meet specific business needs based on loan types, regions or property categories.

Achieve the right valuation results

By adopting ICE’s Property Valuation Model, users can unlock the potential for more accurate and efficient results. To learn more about how our Property Valuation Model supports compliance and provides objective and consistent modeling through high-quality data, optimal hit rates, proven methodologies, rigorous testing and complete transparency, click here.