Automated Valuation Models (AVMs) Explained Quiz

Answers and explanations appear as you take the quiz.

1. Why is the Zestimate's error much lower for homes currently listed for sale than for off-market homes?
   • Listed homes are always in dense suburbs with many similar houses
   • The list price, fresh photos and description give the model strong new information
   • Zillow sends an appraiser to every listed home
   • Off-market homes are excluded from Zillow's error statistics
2. Which AVM technique takes the price change of homes that sold more than once and applies that growth to a property's last sale price?
   • Hedonic regression
   • Computer vision condition scoring
   • Repeat-sales indexing
   • Comparable-sales reconciliation
3. When did the federal AVM quality control rule for mortgage originators and secondary market issuers take effect?
   • October 1, 2025
   • January 1, 2021
   • June 30, 2023
   • The rule was proposed but never finalized
4. Which of the following is NOT one of the quality control factors required by the 2024 AVM rule?
   • Protecting against manipulation of data
   • Requiring random sample testing and reviews
   • Complying with applicable nondiscrimination laws
   • Requiring AVM vendors to publish their model source code
5. Why do AVMs tend to be less accurate in Texas than in many other states?
   • Texas bans the use of machine learning in real estate
   • Texas homes are all custom-built
   • Texas is a non-disclosure state where sale prices are not recorded publicly
   • Texas requires AVMs to ignore square footage
6. A vendor reports a forecast standard deviation of 10 percent for an estimate of $400,000. Assuming roughly normal errors, what does that suggest?
   • The estimate is guaranteed to be within $10,000
   • About two-thirds of true values should fall between roughly $360,000 and $440,000
   • The model is wrong 10 percent of the time
   • The home will appreciate 10 percent next year
7. If an AVM's median error for off-market homes is 7 percent, what can you conclude?
   • Every home is within 7 percent of its true value
   • No home is off by more than 14 percent
   • Half of off-market homes are estimated with more than 7 percent error
   • The AVM is less accurate than any appraisal
8. Which testing practice would make an AVM look more accurate than it really is?
   • Scoring estimates produced after the model already saw the list price or pending sale
   • Freezing estimates before sale information is available
   • Measuring hit rate separately from accuracy
   • Comparing performance by county and price tier