AVM API - Automated Valuation Model for Property Values

Get a lending-grade estimated market value for any US residential property, with the confidence interval and the comparable sales behind it.

What this actor does

• Full valuation. Estimated market value with its low and high bounds, the forecast standard deviation, a confidence score, and the probability the value lands within 5, 10 and 20 percent.
• Comparable sales and listings. Every comparable the model reasoned from as its own row: address, coordinates, distance from the subject, size, beds, baths, year built, list price and date, sold price and date.
• Value only. Just the number, for pricing a long list cheaply.
• Interactive what-if. Re-value a property against the living area, lot size, bedrooms, bathrooms, year built and condition rating you supply, and see what the change is worth.
• Cascade across models. Optionally try several lending-grade models in turn to raise the hit rate on properties a single model cannot price. The model that answered is recorded on every row.
• Backdated values. Ask for the value as of a past date to audit a decision that was made then.
• Pay per result. You are charged for rows that carry a real record. Addresses the provider cannot value are free.

Bring your own key

This actor calls the provider with your own API key, sent in its x-api-key header. Paste it into the API key field, or set it as the DATA_API_KEY secret. Nothing runs on anyone else's credentials, and a run without a key ends cleanly with an explanation rather than failing.

Example output

{
  "recordType": "valuation",
  "requested": "1600 Pennsylvania Ave NW, Washington, DC 20500",
  "found": true,
  "avmId": "1234567",
  "marketValue": 762000,
  "lowValue": 701000,
  "highValue": 823000,
  "forecastStdDev": 0.08,
  "forecastStdDevPercent": 8,
  "confidenceScore": 92,
  "providerName": "Clear Capital",
  "providerModel": "ClearAVM",
  "effectiveDate": "2026-09-01",
  "standardizedAddress": "1600 Pennsylvania Ave NW, Washington, DC 20500",
  "apn": "0143-0001",
  "propertyType": "Single Family Residence",
  "bedrooms": 4,
  "totalBathrooms": 3,
  "livingAreaSqFt": 2450,
  "yearBuilt": 1992,
  "soldComparableCount": 6
}

FAQ

What is an AVM?

An automated valuation model estimates what a property is worth from public records, MLS data and recent comparable sales, without anyone visiting it. Lenders, capital markets teams and investors use lending-grade AVMs for prequalification, portfolio monitoring and underwriting support. This actor returns the estimate along with the model's own statement of how confident it is.

What is the forecast standard deviation?

The forecast standard deviation, or FSD, is how far the model expects to be off, expressed as a fraction of the value. An FSD of 0.08 means the model expects to land within about eight percent. It is the single most useful number for deciding whether to trust a valuation. This actor stores it both as the raw fraction and as a percentage, and the "maximum forecast standard deviation" input is a fraction between 0 and 1, not a percentage.

How accurate is it?

Every valuation row carries the model's own confidence measures: the FSD, a confidence score, an estimated error, and the probability the true value falls within 5, 10 and 20 percent. Accuracy varies by market and by how much recent comparable activity exists, which is exactly what those measures capture. Set a maximum FSD to reject valuations the model is not confident about.

Can I see the comparable sales?

Yes. Turn on "Include comparable sales and listings" in full valuation mode and each comparable becomes its own row, with the address, coordinates, straight line distance from the subject property, gross living area, lot size, beds, baths, year built, zoning code, list price and date, and sold price and date.

What does the what-if mode do?

Interactive mode re-values a property as if its characteristics were different. Supply a larger living area, an extra bedroom, a later year built or a better condition rating and the model returns the adjusted value alongside the characteristics it started from. It answers questions like what a renovation is worth before you commit to it. Because one set of characteristics is applied to every address in the list, run one property at a time when they differ.

Can I get a value as of a past date?

Yes. Set "Value as of date" to a date in YYYY-MM-DD form and the model returns what the property was worth then, which is what you need to audit a lending or purchase decision after the fact. Availability depends on which model answers.

What does it cost?

Pricing is pay per result: 0.02 USD per valuation and 0.004 USD per comparable property, with Apify platform usage included. Addresses the provider cannot value are not charged. Retrieving a valuation you already paid for by its id costs nothing extra on the provider's side. You also need your own agreement with the valuation provider, which bills separately.

Does the actor fail if my key is wrong?

No. A missing key, a rejected key, an exhausted allowance or an empty input all end the run cleanly with a message saying what to change. Only a real outage or a bug fails the run.

Keyword map

AVM API, automated valuation model API, property valuation API, home value estimate API, house price estimate, real estate appraisal API, lending grade AVM, forecast standard deviation, FSD confidence, comparable sales API, real estate comps API, property value lookup, mortgage collateral valuation, portfolio valuation, retro AVM, backdated property value, what-if property valuation, renovation value estimate, US residential property values, ClearAVM alternative