Not All Automated Valuation Models (AVMs) Are the Same
At the heart of every home sale or mortgage deal is the need for an accurate estimate of the property’s value.
And the real estate and mortgage industries have increasingly relied on automated valuation models or AVMs to estimate property values.
The best AVMs will estimate a home’s value with a high degree of accuracy, performing the calculations instantly and removing personal bias and human error.
But not all AVMs are the same.
An automated valuation model is only as good as the data and methodology behind it. While traditional AVMs rely heavily on recent comparable sales, today’s AI-powered valuation models can learn from decades of property transactions, historical pricing patterns, and hyperlocal market trends to generate more accurate estimates. Even with advanced AI, however, one of the biggest challenges remains capturing the subtle differences between properties located in the same general area.
Take, for example, a subdivision with similar homes built around the same time. Square footage, number of bedrooms and bathrooms, lot size, and other property characteristics may be the primary factors influencing value differences within that neighborhood. AI-powered AVMs can analyze these relationships at scale, identifying patterns and value drivers across millions of properties.
In other cases, location plays an even larger role. A home’s value may change significantly if it sits just a block away in a different ZIP code, school district, or housing market. Advanced AI models can combine property characteristics with hyperlocal neighborhood trends, historical market behavior, and geographic context to better understand how values evolve over time. A valuation model that captures these nuances can produce more reliable and accurate estimates, even in markets where traditional comparable-sales approaches struggle.
How ATTOM’s Automated Valuation Model Works
ATTOM has created AVMs with this challenge in mind. Estimates are rooted in precise property and neighborhood boundaries where the AVM can input hyperlocal data to make accurate calculations and statistical analysis.
ATTOMIZED AVMs are powered by its national warehouse of property data, which provides granular property data on millions of properties in every state. The data is updated daily.
This coverage is robust and covers 99% of the population. (ATTOM notably doesn’t provide estimates for mobile homes or homes located on farmland, as these properties are generally not a good fit for AVM estimates.
To obtain accurate results, ATTOM’s AVM uses a three-step approach.
- Step one focuses on data quality. Raw property and transaction records are cleaned and validated to remove errors, implausible values, and non-arm’s-length transactions such as foreclosures, short sales, and inter-family transfers. Additional filters catch below-market transactions that carry no distress flags by comparing observed prices to local market trends. This step is critical and high-quality inputs are essential for producing trustworthy valuations.
- Step two makes historical prices current. ATTOM constructs hyperlocal price indices at the census block level to track how prices have moved over time within each micro-neighborhood. Every historical transaction is adjusted to reflect today’s market conditions, allowing the model to learn from decades of sales history rather than just the most recent transactions. This dramatically expands the amount of useful data available for valuing each property.
- Step three estimates each property’s value using AI. Multiple machine learning models work together, each capturing a different dimension of value. These models are blended dynamically based on the data available for each home, producing a single estimated value along with a confidence score. ATTOM’s models provide a low and high estimate, and median range. About 80% of valuations are within 10% of the sale price and 92% within 20%.
ATTOM also has a Rental AVM that can provide estimates on the monthly rental value for roughly 80 million single-family homes, townhomes and condos.
Why Automated Valuation Models are Important
Automated valuation models are part of a broader technological revolution in real estate. AVMs lower transaction costs and make it easier to complete real estate deals, whether that be in a home sale, refinancing or home equity lines of credit (HELOCs).
AVMs are useful for several purposes and professions. Mortgage bankers and real estate brokers use computer estimates all the time in home sales and refinancing activity. Real estate investors can estimate the value of individual properties or an entire portfolio. Insurance agencies can use them to help develop a risk profile.
But technology is also bringing big changes in real estate. All aspects of the real estate market are becoming more transparent.
Consumers have more tools to do research on their own and more choices in how to buy and sell homes. Traditional real estate professionals need to demonstrate their value to clients.
AVMs are another technological tool that industry professionals can use to sharpen their game, separate themselves from competitors, and help clients as their trusted advisors.
With ATTOM’s easy-to-use platform, the broker or agent can easily generate market reports for clients, do analysis of the local market, or use accurate data to enhance their marketing materials.
ATTOM’s AVMs were designed to produce the most reliable and accurate value estimates in the market. They can help build confidence with clients, close deals more cost effectively and give your business a leg up.
