Zillow's Home Price Zestimate Accuracy Still Sucks

People pay an awful lot of attention to Zillow’s Zestimates of home prices and, to their credit, Zillow is constantly working to improve their valuation algorithm. In fact I think it was in June of last year that they launched their $1 MM Zillow Prize to be awarded to the team that could most improve the accuracy of the Zestimate. That contest is still going on but in the meantime I think it’s good to check in on the current accuracy since I haven’t analyzed this since December 2015 when I compared several automated home value estimators.
First, let’s look at Zillow’s official reporting of their own accuracy.

  • For the Chicago metro area they only give themselves 3 out of a possible 4 stars.
  • 43.3% of the homes have more than a 5% error
  • 23.2% of the homes have more than a 10% error
  • 11.5% of the homes have more than a 20% error

And, yes, I’ve flipped their accuracy reporting around to focus on the errors because when you pull up a home value you need to understand how likely it is that it could be very inaccurate. But here is the big problem. When you get a Zestimate you have no way of knowing whether or not that particular home is going to be part of that 23.2% that has more than a 10% error – i.e. there is a 23.2% chance that the error is going to be huge. Even 10% of a $1 MM home is a lot of money and when you consider that the error could be positive or negative that translates into a $200K range in valuation – possibly much more.
We also decided to check the accuracy for a few randomly selected homes that have sold in some of the higher income areas of Chicago in the last month. We went through the list and compared the sale price to the Zestimate just prior to listing. Here are the results but it is also interesting to note that a significant number of homes didn’t even have a Zestimate until they were listed. That certainly helps Zillow’s reported accuracy since they then have the list price as an important data point. So we did not include those in our analysis.

Note that there are some huge errors in these anecdotal results, which reinforces my point above. You just never know when you are staring at a BS number. Like…check out the forecast for unit 520 at 550 N Kingsbury
As I mentioned above, we used the Zestimate that was in effect just prior to the listing. The reason we did that is because Zillow appears to have done exactly what I suggested back in that December 2015 blog post. They now immediately adjust their Zestimate based upon the listing price when a home hits the market. And you can bet that their accuracy is based upon these adjusted numbers. Just to give you an idea of how big of an effect this can be consider how the Zestimate changed for that condo at 550 N Kingsbury when it hit the market. Their Zestimate took a dive from $1.6 MM to $582 K overnight. Of course, most of the time they won’t have the benefit of a listing to factor into their Zestimate so why count those scenarios in their accuracy measure?

Zestimate history for 550 N Kingsbury, Unit 520
Once Zillow saw unit 520 at 550 N Kingsbury hit the market way below their Zestimate they adjusted it down.

#Zillow #Zestimate #HomeValueEstimates #AutomatedValuations
Gary Lucido is the President of Lucid Realty, the Chicago area’s full service discount real estate brokerage. If you want to keep up to date on the Chicago real estate market, get an insider’s view of the seamy underbelly of the real estate industry, or you just think he’s the next Kurt Vonnegut you can Subscribe to Getting Real by Email using the form below. Please be sure to verify your email address when you receive the verification notice.

Enter your email address:Delivered by FeedBurner


Leave a Reply