Zillow’s problem was not with their home value forecast model. The problem was with applying this model to the real world. Any model is going to have some properties that it values too low and some that it values too high. There assumption was that this would balance out in the end. That was their mistake. In the real world you only acquire the ones that you value too high. Because no one is going to sell it to them for less than it’s actually worth, they only acquired the properties where they were willing to overpay for, which is what they did. I am not the slightest bit surprised that their business model didn’t work because they consistently over paid for properties. I would’ve been quite shocked had it turned out otherwise.