Say a seller wants a home value estimate for an active listing she sees online, but she’s not ready to use a real estate agent.
What would be more useful to her? An automated valuation that factors in the property’s actual list price, or one that excludes that price — instead relying only on comparable sales and other data?
That’s a sometimes-overlooked question raised by an independent study commissioned by Redfin that — rather conveniently — found that Redfin’s automated valuations for property listings are considerably more accurate than online estimates from Zillow and homes.com.
Looking at properties across the U.S., the study found that half of Redfin’s estimates for recently sold listings were within 2.06 percent of their sales price, compared to 5.95 percent for Zillow estimates (Zestimates) and 10.26 percent for Homes.com estimates.
Redfin commissioned market research firm SSRS to conduct an independent analysis for the study.
SSRS reported evaluating a random and “independently assembled data” set of 5,000 recently sold properties spread across more than 7,000 postal codes. It compared the sales prices of these homes to the estimates that Redfin, Zillow and Homes.com had pegged to the properties shortly after they went under contract but before they sold.
“Our findings are clear: Redfin performed significantly better than Zillow and homes.com at predicting home-sale prices,” said Dr. Aniruddha Banerjee, SSRS’s senior vice president of advanced analytics, in a statement. “We hope that others attempt to replicate our results. Our methodology for coming to this conclusion was consistent with the rigorous approaches taken at academic institutions.”
But these results come with at least two important caveats:
- First, the study only analyzed estimates that were assigned to properties when they were listed for sale. It did not compare estimates of properties before they were listed to how much they sold for later on.
- Second, Redfin bakes the list price of a for-sale home into its valuation of that home.
Zillow, on the other hand, purposely does not.
Why not include list prices?
Zillow spokeswoman Emily Heffter offered an analogy to rebut the study from Redfin.
Imagine someone wanted to guess how much you weighed, and started out by asking what you thought you weighed, she said. Now, imagine you told the person you thought you weighed 150 pounds.
“OK, I guess you weigh 150 pounds,” the person responds. Then you step onto a scale and the needle points to 150.
“Yeah, I’m a great guesser,” the person exclaims.
By baking list prices into its for-sale property estimates and then publicizing a better accuracy rate than estimates that ignore list prices, Redfin is acting like that smug, imaginary prognosticator, she argues.
‘More data points’
But the analogy oversimplifies Redfin’s approach.
Redfin’s estimates of listings do vary somewhat from their actual list prices, obviously indicating that the brokerage bases its estimates of for-sale properties on more than just their list prices.
Redfin spokeswoman Rachel Musiker said that, while Redfin’s estimates of for-sale properties do consider their list prices, such estimates are more accurate than estimates for off-market properties “because there are more data points and more-up-to-date information available about those homes than there are for homes that haven’t been on the market for a while.”
What’s more useful?
So what’s more useful?
Estimates for on-market properties that tend to be more accurate because they factor in those properties’ list prices or estimates that try to gauge their value without their list prices?
“People want an opinion of value that’s independent, that’s truly independent,” argues Zillow Group Chief Analytics Officer Stan Humphries. “With Redfin’s approach, you’re not really getting a second opinion. It will be accurate if that listing price is accurate. If it’s not, it won’t be accurate.”
He noted that the Zestimate’s published national median error rate for all properties — both on and off-market — of 4.5 percent is lower than Redfin’s published error rate for off-market properties of 6.43 percent. (Meanwhile, Redfin claims a median error rate of 1.83 percent for for-sale listings, slightly lower than SSRS’s finding.)
Humphries claimed to have seen some Redfin estimates for listings that differed dramatically from what those listings sold for, whereas the Zestimates of those properties came much closer to the mark, he said.
This illustrates how Zestimates are more likely to flag properties that are mispriced than Redfin’s estimates, he said.
But then again: What happens if a Zestimate misses the value of a for-sale property by a wide margin simply because it did not use the home’s list price?
That, of course, can create its own problems if buyers and sellers place more faith in Zestimates than Zillow advises — which some real estate agents gripe that their clients often do. Data, of course, does not always tell the full story of a home’s condition and layout — which is why a property owned by Zillow CEO Spencer Rascoff sold for so much less than its Zestimate.
So where does Homes.com stand on the whole issue?
“We can agree that this study proves that ALL home value estimates are inaccurate,” said Homes.com President David Mele. “And a consumer should use them as a starting point in collaboration with a qualified real estate professional.”
“With that said,” he added. “Homes.com is continually working to improve the accuracy of our estimates, as we know that they are important to consumers.”
Editor’s note: This story has been updated with comments from Homes.com.