Ten years after the Zestimate’s debut, you might expect online home valuations to have become about as accurate as they can be.
But real estate analytics company HouseCanary shows that industry observers think there’s still room for improvement.
HouseCanary recently bagged $33 million in a Series A funding round that included participation from the family investment office of Eric Schmidt, the executive chairman of Google parent company Alphabet, Alpha Edison, ECA Ventures and Raven Ventures.
The startup claims to have the “most accurate valuations available online.” It can value and forecast over 100 million properties in 18,000 U.S. residential markets, using a massive data set, predictive analytics and intuitive visualization tools, the company said in a press release.
Customers include Sotheby’s International Realty, national brokerage Movoto and Blackstone’s Invitation Homes, which operates about 50,000 single-family rental properties.
“We want to do for residential real estate what Bloomberg did for financial services — build a platform that will be a beacon of accuracy and transparency, enabling greater speed and confidence in residential real estate transactions,” said HouseCanary CEO Jeremy Sicklick in a statement.
HouseCanary says its valuations have a nationwide median absolute prediction error of 5 percent, meaning that half of estimates for homes sold in October were within 5 percent of their sales price while the other half were off by more than 5 percent.
The company suggested that this accuracy level should improve as HouseCanary continues to fine-tune its predictive models. That’s happened over the last year: the October 2016 error rate of 5 percent was down from 5.6 percent in July.
Zillow and Redfin publish “median error rates” to convey the accuracy of their valuations. It wasn’t clear if “median error rate” meant the same as “median absolute prediction error” (the accuracy measure used by HouseCanary).
Zillow says the Zestimate has a median error rate of 4.5 percent — which, notably, is down by nearly half from just eight months ago.
Redfin, meanwhile, claims a median error rate of 1.82 percent for for-sale homes, and 6.37 percent for off-market homes.