As featured on Urban Turf
Most prospective homebuyers or people interested in real estate are familiar with Zestimate, Zillow’s tool that approximates home values. However, the tool has been getting some negative attention in the wake of a class-action lawsuit in Illinois that decries the tool’s inaccuracies as undermining home sales.
Today, Zillow announced the Zillow Prize, a contest that will award $1 million to whomever can improve the tool’s precision.
Zestimate’s accuracy has already improved since its unveiling 11 years ago, from monthly updates garnering an estimate with a 14 percent error rate to instantaneous updates that put half of all Zestimates within 5 percent of the home’s eventual sales price.
Analysis factors include data such as comparable sales, price per square foot, tax assessments and last sales price. This data is then plugged into a larger algorithm that constantly assesses the weight of these factors in each census tract in order to recommend an estimated home value.
“Zillow is by far the most widely-used real estate site by a mile, and so Zestimates are prominent and not everybody will always agree with them,” Zillow Group CEO Spencer Rascoff commented during a CBS interview earlier this morning. “Ultimately, the Zestimate is a great starting point, really good information, but every home is worth what a buyer will pay for it and somebody applies human judgment on top of that.”
Now, Zillow is aiming to leverage that “human judgment” by essentially crowdsourcing a solution to improve the computational accuracy of the Zestimate home valuation algorithm.
A public qualifying round, requesting models that will lessen the tool’s residual error, will close next January and narrow the contestants down to 100 teams. The final round will require teams to create algorithms that actually predict home sales prices. The winner will be selected January 15, 2019.
As for the lawsuit? Rascoff dismissed it as “frivolous” and “without merit”.