Almost two years and more than 3,800 teams later, Zillow has a winner in the $1 million competition to improve the accuracy of the Zestimate, the much-debated tool used by the Seattle-based real estate tech company to measure home valuation.
The company announced on Thursday that the winner of the so-called Zillow Prize will be revealed on Jan. 30. Last March, Zillow announced the top three teams to advance out of the first round and compete for the grand prize. Some competitors and their theories on the Zestimate, are featured in the video above.
The winning team had to beat the accuracy of a Zillow benchmark model when evaluated against real-time home sales between August and October 2018, the company said in a news release. The winner’s algorithm also needed to score the largest improvement to the Zestimate home valuation’s accuracy among all the final competitors.
While the Zestimate’s margin of error is currently 4.5 percent nationwide, the valuation is always a hot-button topic across the real estate landscape as home sellers, buyers and agents grapple with whether the valuation benchmark is too high or too low or just right.
Zillow co-founder and executive chairman Rich Barton called the Zestimate “very provocative and personal and a little voyeuristic” in a 2016 GeekWire interview discussing how the company came up with the tool.
Zillow has always said that the Zestimate, introduced in 2006, was designed to be just a starting point to help people estimate the value of a home for free. There are valuations on more than 110 million homes across the U.S. today, and Zillow says that “constantly improving the Zestimate’s accuracy is a top priority.”
“Awarding the Zillow Prize is nearly two years in the making, and we’re thrilled to finally be naming the champion,” Stan Humphries, creator of the Zestimate and Zillow Group’s chief analytics office, said in a statement. “We’ve been impressed by the imagination and ingenuity all our competitors showed in improving the Zestimate. The winner’s improvements to the Zestimate will not only give people more accurate home valuations and knowledge about one of their biggest life investments, but also show the innovation that’s possible when we crowdsource ideas and put these types of challenges into the hands of data scientists around the world.”
Zillow Prize attracted contestants on teams representing 91 countries and became one of the most popular machine learning competitions ever on Kaggle, the platform administering the contest, according to the company.