I don’t trust testimony. I’m dubious of any claim I can’t test first-hand, but my dubiety increases with the number of hands involved. Moreover: It is a simple fact of human nature that people initiate conversations for a reason, and that reason is very often guile. Even where it is not, if I don’t know you know what you’re talking about, I’m happy to assume you don’t.
Oh? You have skin in the game? Well, that’s very different. Now I really don’t trust you! 😉
And thus we introduce a big bucket of freshly-extruded iBuyer testimony, brought to us by MarketWatch: “Why buying and selling a house could soon be as simple as trading stocks.”
The thesis is not defended, but these things never are. I don’t want to seem to be too critical of the reporter here. Reporters are generalists, and an actual software engineer could not discern fact from fiction in the specious spewings of iBuyer spokespersons.
Here’s what I know first-hand: Zillow does its buy-side pricing in Phoenix by human CMA. Second-hand: I am told by people who did them that Phoenix buy-side pricing by OpenDoor and OfferPad are also done by human CMA – and I am hugely popular among VCs just now for demonstrating OpenDoor’s recruitment of property valuation analysts.
I don’t know of any iBuyers buying truly blind, by AVM only, although all of them should. The race is to get the paperwork first, and to the victor go all the ensuing spoils – assuming there are any, net-after-everything.
In other words: Every breathless claim you have read about “pricing algorithms” is almost certainly bullshit. iBuying is done by human CMA for now – by Realtors, as Brian Brady points out – and that seems unlikely to change. AI is a cargo cult, after all – a machine built to cheat people who insist machines can’t cheat. “Ya want blockchain with that, chump?”
The article is simply the repetition of the same breathless bullshit, the transcription of undefended claims, so read it for yourself. This, from Zillow’s Krishna Rao, is choice:
What does that mean in real life? Zillow sees the listing price as a “machine learning” exercise, he said.
“That machine can look at what the relative demand is for homes like this, relative supply, how that’s trended, and take these gobs of data and crunch it down into a particular listing price. Over time, as that home is listed, we then get more and more granular information — how well is the home showing? Are we seeing lots of tours, lots of offers? And use that to refine our strategy.”
As I have demonstrated, all three Phoenix iBuyers seem to be pricing for resale by a mark-up on the purchase price consisting of a multiple of the rehab costs. There might be software doing all this heavy lifting, but a spreadsheet seems much more likely – and certainly more than is needed for what could easily be a penciled calculation.
Another quote, this one from Knock’s Stephen Freudenberg:
He offers an example: A family might spend $100,000 remodeling a kitchen but add only $50,000 to their house’s listing price because properties in the surrounding area, which are comparable listings, might not have such upmarket kitchens. “So they’re stuck with what the neighborhood sold for, but, if we’re actually looking at the data, then everyone could theoretically get a better deal.”
I related that example to my wife, just so I could watch her eyes roll. What do the sellers of over-improved homes get? Market, just like everyone else. Why? Because you are selling to people, not to an idealized perfect-schmoo, and wise people resist eating other people’s mistakes.
Meanwhile, here’s better news: The machines that could never, ever cheat anyone are also incapable of achieving racist outcomes in real estate:
To Dahlia Brown, the Knock customer in Marietta, having an algorithm at the heart of the real-estate market may help counter human bias by limiting “some of the historical practices that maybe have kept certain people from home ownership,” she said. “This process actually seems as fair and equitable as it could be.”
That map is Zillow in the core of Metropolitan Phoenix, everything sold to date. The Madonna-with-child outline comprises most of the Phoenix and Glendale neighborhoods that are typically redlined. It’s price and year built, not racism, resulting in the effect. But it’s hard to miss a hole that big…
I do wish reporters would take well-informed experts along on these interviews to ask the follow-up questions. The claims reported here and everywhere are absurd, and none are backed by evidence that I have seen – where pricing-by-algorithm, at least, would seem to be a big fat lie.
One lie, all lies? Tell me why not?