Trustpilot
Loading...
Loading...Loading...An automated valuation model (AVM) is a computer program that uses algorithms, public records, and sales data from similar properties to give an estimate of a property's value without having to see it in person.
You might have used an AVM without knowing what it was called. There is an automated valuation model that runs behind the scenes and gives you an instant price estimate when you type an address into a real estate website. An AVM is a computer program that uses information from tax records, public records, MLS listings, and past sales to figure out how much a property is worth right now.
The idea isn't brand new. Freddie Mac made its Home Value Explorer AVM more than 20 years ago to help with internal risk management and portfolio analysis. The scope has changed. AVMs process millions of data points across the country every day and give you a value estimate in seconds. For example, a traditional in-person appraisal can take one to three weeks from the time it is scheduled until it is delivered.
AVMs are basically trying to answer the same question as a human appraiser: What would a buyer who wants to buy this property pay a seller who wants to sell it in its current condition? The way they get there is different. An AVM doesn't walk through a house, measure rooms, and make decisions about how good or bad it is. Instead, it crunches numbers. That speed is a big plus in a market that moves quickly, but it also has pros and cons that every home buyer should know about before making a big financial decision.
An AVM is based on what real estate professionals call a "comparable sales approach," but it's done by a computer instead of a person. The software finds your target property, gets its details from public databases, and then looks for similar properties that have sold in the same area recently. It will use statistical models to account for the differences between the properties and come up with a value.
Most AVM providers use some combination of three main methods. The hedonic pricing model looks at a home and breaks it down into its individual parts, such as square footage, number of bedrooms, lot size, and age. It then gives each part a dollar value based on how much the market pays for those parts. The repeat-sales model looks at how the value of a single property changes over time by comparing the prices of several deals on the same home. The hybrid model combines parts of both by adding machine learning algorithms that can find patterns in thousands of deals.
The amount of data that modern AVMs can handle is what sets them apart from older ones. A good AVM today can get information from county recorder offices, local tax assessor databases, MLS feeds with both active and closed listings, and even geographic information system data that shows flood zones, school boundaries, and how close you are to amenities. Some more advanced models use satellite images and permit records to find improvements or new buildings that haven't been recorded in tax records yet.
The output is more than just one number. Most professional-grade AVMs also give the user a confidence score that shows how accurate the estimate is for that property. A high confidence score means that the model had good data to compare it to and that the property fits in with what is already known. A low confidence score means that the model is less sure, which is usually the case when the property is unusual, the comparable data is sparse, or the market has been unstable lately.
The accuracy of any AVM depends entirely on the data feeding it. This is a basic rule of any statistical model, but it matters especially here because property data in the United States is fragmented across thousands of local jurisdictions, each with its own recording standards and update schedules.
Public records form the backbone of most AVM databases. County recorder offices maintain deed transfers, mortgage recordings, and lien filings. Tax assessor offices track property characteristics like square footage, lot dimensions, year built, and assessed value. These records are usually reliable for basic physical attributes, but they can lag behind reality. If you finished a basement, added a bathroom, or replaced your roof, that work might not show up in public records for months or even years.
MLS data gives AVMs access to current and historical listing information, including asking prices, days on market, and final sale prices. This is some of the freshest data available, but MLS coverage isn't universal. Some markets have better data sharing agreements than others, and properties sold off-market (like many investor deals) won't appear in MLS feeds at all. You'd be surprised how many borrowers are confused when the online estimate for their home doesn't reflect a renovation they completed last year, and the reason is almost always a gap in the data pipeline.
Tax assessment records provide another layer. Assessors in most jurisdictions revalue properties on a regular cycle, sometimes annually, sometimes every three to five years depending on state law. The assessed value is calculated for property tax purposes and often sits below market value. AVMs treat assessed values as one input among many rather than as a definitive price.
AVMs show up at several stages of a mortgage transaction, though their role varies depending on the situation and the risk profile of the loan.
When you apply for mortgage, the lender needs to verify that the property securing the loan is worth enough to justify the loan amount. For many conventional loans, Fannie Mae's Desktop Underwriter or Freddie Mac's Loan Product Advisor will run an AVM as part of the automated underwriting process. If the AVM returns a high-confidence estimate that supports the loan amount, the system may offer an appraisal waiver (sometimes called value acceptance), meaning the lender can skip the traditional in-person appraisal altogether. This can save borrowers money and speed up the closing timeline. AmeriSave evaluates each deal individually to get the best approach for a specific property and borrower.
Lenders and investors who hold mortgage portfolios use AVMs to track the collateral value of their loans over time. If property values in a particular area drop, the portfolio manager will get early warning that certain loans are approaching a higher-risk loan-to-value ratio before problems develop. This kind of monitoring helps keep the mortgage market more stable overall.
When borrowers apply for a home equity loan or a home equity line of credit, the lender needs to know how much equity is available. An AVM can provide a quick estimate for this purpose, especially for lower-risk transactions. In some cases, the lender may supplement the AVM with a drive-by appraisal or a desktop review rather than ordering a full interior inspection. I've worked with home buyers in the Dallas-Fort Worth area who were able to move faster on a home equity application because the AVM data for their neighborhood was strong enough to support a streamlined process.
AVMs play a growing role in refinance transactions. If you already have a mortgage with a lender and you're looking to refinance, the lender may already have a property valuation on file. An AVM can confirm whether the property has held or gained value since the original purchase. Fannie Mae and Freddie Mac have both expanded the circumstances under which they'll accept AVM-based valuations for refinances, especially for borrowers with strong payment histories and lower loan-to-value ratios. When AmeriSave processes a refinance application, we look at the full picture to determine what valuation method makes the most sense for your loan.
How close does an AVM get to the real value? That depends on the property, the market, and the specific model being used. The standard metric for AVM accuracy is the median absolute percentage error, which measures the midpoint of how far off the estimates are from actual sale prices. Well-performing AVMs in active markets with good data coverage can achieve median absolute errors in the 3% to 7% range for properties with high confidence scores. For lower-confidence properties, the error margin can widen past 15% or even 20%.
Let's put that in real numbers so you can see what it means for your wallet. Say you're looking at a home that eventually sells for $375,000. If the AVM estimate came in at $363,750, that's a 3% error. Not bad. But if the AVM was off by 10%, the estimate would have been either $337,500 or $412,500. On a property worth $375,000, a 10% miss in either direction could mean the difference between getting your loan approved, needing a bigger down payment, or overpaying because you thought the home was worth more than it actually was.
Confidence scores give you a way to gauge how much weight to put on a given estimate. Most AVM providers scale their confidence scores from 0 to 100. According to industry research, properties with confidence scores above 90 tend to have median absolute errors around 5%, while properties scoring between 70 and 80 may have errors closer to 10% to 12%. Properties scoring below 60 are considered unreliable, and most lenders won't make a lending decision based on an AVM alone in those cases.
AmeriSave weighs these confidence thresholds when deciding whether an AVM-based valuation will work for a particular loan. When the data supports it, using an AVM can save you money on appraisal fees and shave days off your closing timeline. When the data doesn't support it, a traditional appraisal is usually the safer choice.
AVMs work best on standard properties in active markets where there is a lot of recent and comparable sales data. They have a hard time when the data is sparse, old, or just can't show what makes a property different from its neighbors.
The hardest things to deal with are properties that are one-of-a-kind or not standard. The AVM doesn't have a lot of information to work with if you're looking at a custom-built home with features that are different from anything else in the area. Mixed-use properties, homes on lots that are too big or too small, and properties in rural areas where sales are rare all have the same problem. I have seen AVM estimates that were off by 20% or more in rural Texas because the closest sale that was similar was eight miles away and two years old.
Condition is another thing that people don't see. An AVM can tell you the square footage, the year it was built, and the tax assessment, but it can't see that the homeowner tore down the kitchen last summer or that the foundation has a crack in it. Two houses on the same street built in the same year can be worth very different amounts depending on how well they've been cared for. Public records alone can't tell you that.
Markets that change quickly are also a problem. If prices in a neighborhood are changing quickly in either direction, the AVM might be using sales that are no longer relevant by the time you get the estimate. This was especially clear when prices went up quickly or when the market corrected itself and values could change faster than the data could be updated.
The federal government has taken a more active role in regulating AVMs in recent years, driven by concerns about accuracy, fairness, and the expanding role these models play in lending decisions. The Dodd-Frank Wall Street Reform and Consumer Protection Act directed six federal agencies to develop quality control standards for AVMs used in mortgage transactions. That mandate sat unfulfilled for over a decade before regulators finally acted.
The Consumer Financial Protection Bureau, along with the Office of the Comptroller of the Currency, the Federal Reserve, the FDIC, the NCUA, and the Federal Housing Finance Agency, issued a final rule requiring mortgage originators and secondary market issuers to adopt policies, practices, procedures, and control systems to make sure AVMs meet five quality control standards. Those standards call for ensuring a high level of confidence in estimates, protecting against data manipulation, avoiding conflicts of interest, requiring random sample testing and reviews, and complying with nondiscrimination laws.
That last standard is worth paying attention to. There's been growing concern that AVMs could inadvertently perpetuate bias in property valuations, particularly in communities of color. If the historical sales data that feeds an AVM reflects past discrimination in lending or appraisal practices, the model could produce estimates that undervalue homes in those neighborhoods. The nondiscrimination requirement pushes AVM providers to test their models for bias and take corrective action where it's found. AmeriSave is committed to fair lending practices across every step of the mortgage process, and that includes how property valuations are handled.
For borrowers, the practical impact of these regulations is that the AVM estimates your lender relies on are subject to oversight and quality controls that didn't formally exist before. This gives consumers real protection, especially as AVMs take on a larger role in lending decisions.
It's tempting to think of AVMs and traditional appraisals as competing methods, but they really serve different purposes. A better way to think about it is that they're complementary tools in a property valuation toolkit, and the right choice depends on the situation.
A traditional appraisal involves a licensed, state-certified professional physically inspecting the property, measuring its dimensions, evaluating its condition, photographing the interior and exterior, and researching comparable sales. The appraiser then writes a detailed report that documents their findings and their opinion of value. This process usually costs between $350 and $550 for a standard single-family home and takes one to three weeks to complete. The result is a professional opinion grounded in both data and firsthand observation.
An AVM will get you an estimate in seconds at a fraction of the cost, sometimes for free when it's bundled into a lender's technology platform. The estimate is based entirely on data and algorithms, though, with no human judgment about condition, curb appeal, or the kind of intangible things that can make one home worth more or less than its neighbors. There's no site visit, no photographs, and no one verifying that the property data in public records is accurate and current.
For most purchase transactions, lenders still require a traditional appraisal. That's because the stakes are high, and the lender wants a human professional confirming the collateral value before committing hundreds of thousands of dollars. But for refinances, home equity products, and portfolio monitoring, AVMs have become an accepted and even preferred tool because they can deliver useful information at scale. The mortgage industry is moving toward a hybrid approach that uses AVMs for screening and triage, with traditional appraisals reserved for transactions where the data alone isn't enough. AmeriSave uses this approach to help borrowers get the most efficient and accurate valuation process for their specific situation.
Knowing how AVMs work can help you make better choices at different points in the home-buying process.
First, the quick estimates you see on real estate websites are outputs from AVM. You can use them to get a general idea of what homes in a neighborhood are selling for, but you shouldn't use them to set your offer price. Don't take them as the final answer; use them as a starting point for your research. Your real estate agent's comparative market analysis and the final appraisal are better ways to figure out how much a house is worth if you're making an offer.
Second, if your lender offers an appraisal waiver based on AVM data, that can be a real plus. You'll save money on the appraisal fee and probably close faster. Before you decide to skip the independent professional assessment, think about whether you feel comfortable doing so, especially if you've never owned a property before. If you've lived in your home for a long time and want to refinance it, the risk is lower. If you're buying a new home, especially if you think it might have problems, having a licensed appraiser walk through it gives you extra protection.
Third, don't worry if an AVM estimate comes in lower than you thought it would. It doesn't mean that the house costs too much. It could just mean that the model didn't have all the information. An AVM might not give an accurate value if your home has had recent renovations, a finished basement that isn't shown in public records, or a layout that isn't typical for homes in your area. When AmeriSave looks at your loan application, our team doesn't just look at the numbers. They also look at the whole picture of the property you're buying.
Automated valuation models have changed how the mortgage industry figures out how much a property is worth. Before, it was all up to individual appraisers and took weeks to get an answer. Now, it's much faster and easier. They are a helpful way to quickly figure out how much a home might be worth, and they will continue to play a bigger role in the underwriting process for refinances, home equity products, and even some purchase deals.
An AVM estimate is not the same as an appraisal. It can't see what a licensed professional can see, and the accuracy of the estimate depends on how good the data is that it gets. Smart home buyers don't take AVM outputs as the last word on value; they see them as one piece of information among many. AmeriSave can help you understand how property values work and which mortgage option is best for you if you want to get the most for your money when you buy a home.
The accuracy of AVM changes based on the property and the quality of the data it is based on. In active markets with good comparable sales data, AVMs that work well can give estimates with median absolute errors of 3% to 7% for properties that are very confident. Traditional appraisals are usually more reliable for individual transactions because they involve a licensed professional coming to the house and making decisions about its condition and features. Most lenders still want a traditional appraisal for most purchase loans. If you want to know how the appraisal process will affect your loan, AmeriSave's appraisal guide has all the information you need.
Yes, in some cases. Fannie Mae and Freddie Mac let lenders use AVM-based valuations for some refinance transactions and even some purchase loans through their appraisal waiver programs. Your eligibility depends on things like your loan-to-value ratio, payment history, and how much data is available for properties like yours. Most first-time home buyers and many other people who buy a house still need a traditional appraisal. At AmeriSave, you can learn about desktop appraisals and other options for appraisals to see what might work best for you.
AVMs use a lot of different data sources, such as public property records from the county recorder and tax assessor offices, sales and listing data from the Multiple Listing Service, deed transfer histories, and geographic information system data. Some models also use satellite images, permit records, and indicators of market trends. The estimate's accuracy depends directly on how fresh and good this data is. AmeriSave's guide to fair market value explains the different methods and how they compare to these valuations if you're interested.
Yes. The CFPB, the Federal Reserve, and the FHFA, along with three other federal agencies, set quality control standards for AVMs used in mortgage transactions under Section 1473 of the Dodd-Frank Act. The rule says that mortgage originators and secondary market issuers must have policies and controls in place to make sure that AVM estimates are accurate, not subject to data manipulation, not biased, tested at random, and in line with nondiscrimination laws. To make sure your property valuation is done fairly and accurately, AmeriSave follows all federal lending rules.
A confidence score shows how accurate an AVM estimate is for a certain property. Most of the time, scores are between 0 and 100. The model had strong comparable data and the property fits well within established pricing patterns, so the estimate is more likely to be correct if the scores are higher. Lower scores mean that there wasn't much comparable data, the property is unusual, or the market is unstable. These scores help lenders decide if they can trust the AVM or if they need a traditional appraisal. At AmeriSave, you can find out more about how to value property.
AVMs use data that is available to the public, which may not always be up to date. The AVM won't know about any work you've done on your home that hasn't been recorded in tax assessor or public record databases. This includes things like remodeling your kitchen, finishing your basement, or making other improvements. AVMs also can't directly check the condition of a home, so a home that has been well cared for and one that hasn't might get the same estimate even though they have the same recorded features. Location within a neighborhood can also matter. For example, corner lots, cul-de-sacs, and being close to busy roads can all change value in ways that data can't always show. You can learn about what makes your home worth more by reading AmeriSave's guide to getting ready for an appraisal.
AVMs have a hard time accurately valuing new construction and recently renovated homes. There may be few or no comparable sales from the same development for new builds, especially in the early stages of construction. The AVM has to use data from a larger area that may not show how much new homes cost. The problem with renovated properties is that public records don't always show how much work was done. It could take months for the data to show that a full gut renovation has turned an old house into a modern home. In either case, a traditional appraisal is usually the best option. Find out what affects the value of a home at AmeriSave.
You have choices if a low AVM estimate is making it hard to get a loan or changing the terms of your loan. First, ask your lender if you can get a traditional appraisal to get a more accurate value based on a physical inspection. You can also show proof of recent improvements, sales that are similar to the model's but that it may have missed, or other proof that the value should be higher. If your loan is going through automated underwriting and the system offers an appraisal waiver that you don't like, you can usually ask for an appraisal anyway. The people at AmeriSave can help you understand your choices and find the best way to move forward.
Fact Checked
