Technical advances mean Zestimates are computed in nearly real time. And now data scientists, engineers and visionaries from around the world are focusing on them, too.
For many Americans, their home is one of their largest financial assets. The typical homeowner has a little over half of their wealth tied up in their home, according to a Zillow analysis of the Federal Reserve Board, Survey of Consumer Finances.
That’s why the Zestimate® home value – and improving its accuracy – is so essential. While it’s not an appraisal, the Zestimate, based on a complex mathematical equation, gives a homeowner a starting point for tracking their home’s worth.
A home’s value is ultimately determined by what someone else is willing to pay for. An algorithm can’t really take into account the perfect paint color, the way light filters through a window or that feeling of “home” when a buyer walks in the door.
Instead, the Zestimate algorithm pulls a host of publicly available records from county assessors: past sales, square footage and past valuations. On top of that, public records like road networks and neighborhood surroundings (views, parks and other amenities) round out the millions of data points that go into Zestimates.
Zillow is constantly working to improve the Zestimate. When Zillow launched 11 years ago, the Zestimate had a median absolute error rate of 14 percent. Today, the algorithm’s accuracy is within 4.3 percent nationwide, meaning half of Zestimates nationwide were within 4.3 percent of the final selling price, and half are off by more than 4.3 percent.
Some of the improvements are purely technical. For example, Zillow recently transitioned all data to the cloud so the team can compute the Zestimate in nearly real time. Zillow can process three times as much data, which means homeowner updates are processed faster, and the team of data scientists can iterate improvements more quickly.
And now Zillow has offered the $1 million Zillow Prize, inspiring the brightest scientific minds to improve the Zestimate. The competition, launched in May, has attracted more than 15,500 individuals who have downloaded the competition dataset. More than 2,500 data scientists, engineers and visionaries from 76 countries have submitted an average of 350 entries a day.
That’s a lot of great minds working to improve Zillow’s automated home valuations of 100 million homes across the country.
Learn more about Zillow Prize.