Ask HN: Where did Zillow data science go wrong?

7 points·hbarka··6 comments

News coming out now about the layoffs at Zillow due to heavy losses in their home-flipping program. They have a treasure trove of data, perhaps one of the largest, related to the housing market.

View on HN

I’m guessing if they aren’t able to recover after the market gains from the last two years, their problems are in the labor involved in flipping. Probably holding houses longer than anticipated because they didn’t realize how hard it is to get tradespeople to show up for literally anything.

I wonder if they contributed to the housing shortages by buying up all the fixer uppers in the remotely affordable price range.

Similar to tradespeople not showing up at ports, exacerbating the supply chain. Zillow may have figured out the digital programming problems but the analog work of literally improving the homes did not transfer to the same principles.

I’m pretty experienced financially, and in my opinion they bought high and sold low. This is bad.

They may have felt they could use some sort of “momentum trading” or another commodity trading strategy which just never worked out in reality. If I was to guys why it didn’t work out I’d bet on it being due to the normal friction involved in the buying/selling/conveyancing/etc processes which vary considerably based on local and state level laws.

Perhaps ML and data analysis methods aren’t the magical panaceas believed? Or possibly they needed to read Taleb?

They didn't have insight into the labor markets in local areas and didn't realize they wouldn't be able to hire contractors fast enough or cheap enough to flip the homes they purchased.

Reply on news.ycombinator.com