I used Prophet and personally I do not have any problems, but I agree with the criticism that the tool it’s extremely focused in ergonomics that abstracts important aspects of the tool that can be used to built better models [1].
I thought the biggest issue wasn't with the models themselves, but how Zillow decided to apply and act on them, which is why it didn't work in practice.
So on average their predictions may have been pretty good, but since each transaction also depends on the other party to accept their offer, and whether they get outbid, most of their predictions where the offer actually goes through would be on the tail end of where they slightly overestimated the price.
I was lucky to make and learn from that mistake pretty quickly with some algorithmic trading on much smaller amounts. With housing transactions being much larger and slower, you wouldn't learn this lesson until it was too late. Models never perform as well in practice as they do in theory, and you need to remember to account for both known unknowns and unknown unknowns.
Great comments! I've learned a lot from them. I'm just getting started with algorithmic trading and time series modeling, so I appreciate your insights.
[1] - https://ryxcommar.com/2021/11/06/zillow-prophet-time-series-...