Zillow's model is fine in the aggregate, which is what the parent was suggesting it for. They ran into trouble profiting from it because adverse selection hits them hard -- they'll only be able to buy from the very homeowners whom the model most overpredicts the value of, because of unobservables it can't model.

None of that relates the question of whether Zillow, on overage, correctly predicts how much the market, in the aggregate, has moved.

I think you're confusing separate issues because they both merge into a narrative of "zillow bad".

> I'm not sure that it makes sense to trust their pricing models.

Zillow, Redfin, etc list at prices that may be inflated, but they always list the purchased pricing as far back as they can and raise estimate to last sale if higher.

I don't think his point is Zillow specifically, but the housing market as it exists today. Zillow is his example.