14 ms·

Crazy. I recently sold a condo to Opendoor for significantly more than I would have even thought to list it for. When I negotiated with Opendoor after their ini

Crazy. I recently sold a condo to Opendoor for significantly more than I would have even thought to list it for. When I negotiated with Opendoor after their initial offer, I pointed out a recently sold condo (days before, in similar condition, layout and finishes) in the same complex that sold for much higher than Opendoor offered. Within hours Opendoor came back matching that same selling price. The kicker? It was a unit that Zillow bought. The algorithms are fooling themselves... Opendoor matches Zillow who matches Opendoor and that's how you get ever increasing offers. Edit: oh, and now Zillow has that unit on the market, priced 14% lower than they paid for it, after 3 price cuts so far. Edit 2: Phoenix/Scottsdale AZ market

AareyBaba 5y ago
I suspect the 'AI software' running these companies is using linear regression to predict housing prices and one of the inputs is the price of similar houses nearby.
- c141charlie 5y ago
  What could possibly go wrong with that approach? :-)
- danrocks 5y ago
  Maybe some reinforcement learning algorithm running as well, where the maximum reward is winning the bid. Hence the algorithm just goes and lays waste on other bidders. Gotta get that infinite reward!
  - mrfox321 5y ago
    I doubt RL is involved. That would be soooo risky. Bidding algorithms can model the price of winning the auction, you can maximize profit without fancy RL
- nitwit005 5y ago
  I recall in one of the previous crashes it turned out a firm had a model that couldn't handle homes going down in value. A linear model is superior to that.
- jitl 5y ago
  This is basically how human-lead appraisals work in most markets.
- streetcat1 5y ago
  should move to xgboost.
- SantalBlush 5y ago
  In fairness, I would think humans include nearby home prices as one of their parameters as well.
  - LanceH 5y ago
    A Broker Price Opinion (BPO) is typically set by appraisal, comparable sales and comparable listings. Depending on how cookie cutter the neighborhood is, there are many different levels of "appraisal" from drive by photos to a deep inspection.
  - KingMachiavelli 5y ago
    Human/realtor appraisals are often very poor. They will take 3-4 nearby & similar homes and basically add/subtract the differences from the home they are comparing. Then they basically just average the adjusted sale prices. They might add a bit onto that price since prices go up over time. So this has some obvious issues: * Areas without a ton of very similar houses that have also sold recently will basically have no 'comps' to use. * It's really easy to keep identifying differences (pros/cons) until the adjusted prices equal each other but it's hard to know if all of the meaningful differences have actually been identified. * It gives average homes and average buyers a huge advantage. This model assumes that the housing market is hot but not hot enough that someone will pay 10-15% more for some specific feature. Anything unique to a home/property is only going to be worth a fraction of the time & money it would cost to add. Anything super common (kitchen remodel, finished basement, etc.) can actually add 100% or more of it's cost to the houses value because the buyer is paying with 5x or more leverage so paying a bit extra to have it included. This is also why it's often better to fix a few things as the seller than give a discount on the sale price - the buyer will often pay a premium to get things move-in ready since it doesn't impact their monthly costs significantly.
- devnull3 5y ago
  lol ... this infact is the sample problem which Andrew Ng explains in his famous Machine Learning course on Coursera!
  - hbarka 5y ago
    Wait, can you elaborate? I’m sincerely interested in going over that material.
- davnn 5y ago
  There was a Kaggle Competition in case you missed it. At least here you can see what kind of features they are using and what models people employed. See https://www.kaggle.com/c/zillow-prize-1 https://www.kaggle.com/c/zillow-prize-1
  - xapata 5y ago
    Zillow Offers team was isolated from the Zestimate team. They didn't share algorithms.
    - laminarflow 5y ago
      AFAIK this is false; Offers began transacting at the Zestimate price in several markets earlier this year
      - xapata 5y ago
        Could be stale information. I talked to them a while back. Still, transacting at the Zestimate price doesn't mean they're privy to the Zestimate algorithms.
bombcar 5y ago
I've been noticing lots of "new lower price" emails from Zillow; I think the market has cooled off - but how much of it being hot was this investor competition?
- crackercrews 5y ago
  It's also the last rush of sales before things slow down for the holidays. There's also a chance that interest rates will be higher next year. That would bring home prices down.
- datavirtue 5y ago
  A lot of people were jumping in with their rundown old houses asking rediculous (lose your ass) prices and I'm seeing them all revise as of late. This was seen mostly in rural areas--Im looking for a farm. I saw quite a few houses that sold a few years prior for much lower prices. The disparity in estimated values and previous sales versus the list price was astronomical in most cases. The banks will happily give you a loan for whatever price so there is no check on the runaway prices except for your own personal knowledge of the historical market demand in an area.
  - coffeecat 5y ago
    I'm not sure I follow your logic; if you're concerned with buying a house now, then prices from years ago are irrelevant. You can speculate that prices have risen too quickly and that they'll revert back to their earlier level, but that's ultimately just speculation. Comparable sale prices from the past few months are the thing to look at, to determine an offer that's likely to be accepted. One thing to keep in mind about list prices: if the sellers do a good job of estimating the current market price, then the house will sell quickly. If you're checking listings infrequently, there's going to be some sampling bias in the listings you see; overpriced houses that aren't selling will be disproportionately represented in active listings.
- iav 5y ago
  price reductions are seasonal in real estate. Peak new listings are around Super Bowl Sunday and Labor Day. There are very few price reductions in that period. The best quality homes at attractive prices sell first. By late October, the total number of listings on the market starts to decline, and price reductions become common.
onlyrealcuzzo 5y ago
TBF - wasn't the Phoenix metro up almost 40% in one year? After a 14% cut, that would still be an absurd YoY increase (I think still the largest single year increase in a major metro in the US on record?). I get that Zillow is losing 14% - but the market is not yet even back to normal appreciation - let alone "crashing".
- roland35 5y ago
  Here is the case shiller index for Phoenix - it is absolutely wild since 2020! https://fred.stlouisfed.org/series/PHXRNSA/ https://fred.stlouisfed.org/series/PHXRNSA/
  - lsh123 5y ago
    One thing I don’t understand is that many people who buy houses in Phoenix believe in Global Warming getting worse in the short term (source: lived in Phoenix for a year, considered buying a house there, and spoke with many folks who just moved or were about to move).
    - KingMachiavelli 5y ago
      Hot and dry is not that bad from a global warming standpoint. Evaporative cooling is very effective and efficient so it's actually less energy intensive heating in colder locations. The plentiful amount of space and days of sun makes solar energy very practical. The nights can get pretty cold so it means that certain evening hours are actually very comfortable.
      - xapata 5y ago
        Except for drought problems.
    - railton 5y ago
      Have you read "The Water Knife" by Paolo Bacigalupi? Mostly set in Phoenix when the water has run out. https://www.goodreads.com/book/show/23209924-the-water-knife https://www.goodreads.com/book/show/23209924-the-water-knife
roland35 5y ago
You think Zillow would have one person take 15 minutes to do a quick look at the offer prices first!
- aaronwall 5y ago
  They could have charged a review fee to have an appraiser visit & make accurate price estimates & then let the homeowner keep the Zillow certified valuation label if they don't sell it to Zillow.
monksy 5y ago
Oh the new REITs playing the algro game. This is cute.
analyte123 5y ago
If local real estate agents and developers can manage to manipulate the "Zestimate", and I'm pretty sure they are at least trying, you can imagine the payoff.
- Hjfrf 5y ago
  Developers could certainly do it. A single padded first sale on a new development would be factored into offers. Classic wash trade.
avidphantasm 5y ago
Gee, what could go wrong with over fitting incompressible models?
monkeybutton 5y ago
Madness.. Its not like such a feedback loop hasn't been encountered before in systems. I like this example: https://www.michaeleisen.org/blog/?p=358 https://www.michaeleisen.org/blog/?p=358
stefan_ 5y ago
It's hard to tell if this is "tHe MaRkEt Is CoLlApSiNg" or just a Knight Capital 2.0. Leaning towards the latter.
qqqwerty 5y ago
This is an interesting lesson in algo trading run wild. Roughly, house prices have a fundamental ceiling, and it is determined by the size of the monthly payment that banks will allow the typical homebuyer to assume when approving a loan. If Zillow was the only iBuyer in a particular market, then that ceiling would likely have held, as all other non-zillow sales would still be operating under the loan approval constraint, and that would get reflected in the comps. But in a market with multiple ibuyers, none of which are capital constrained, it would make sense that they run up the prices against each other, past what the local homebuyers can afford. Not sure how much that actually played a role here. But could be fun to do some back of the napkin math to see if that was the case in some of these markets.
- derefr 5y ago
  > house prices have a fundamental ceiling, and it is determined by the size of the monthly payment that banks will allow the typical homebuyer to assume when approving a loan. In the more irrationally-hot markets, why assume that housing is even being purchased on a mortgage by your average home-buyer, rather than being purchased cash-in-hand by a private or corporate investor looking to park wealth they've already generated? (Even if that isn't the majority of houses in those markets, the sales like that that do happen would still have an impact on property values in the affected neighbourhoods.)
  - trident5000 5y ago
    Because the real estate sector is like 30-40 trillion. 2007 was the result of a massive infusion of debt by unsophisticated retail "investors" who bid prices up far higher than even their rental value. Lending standards dont allow this today. If big institutional money is going to cause a frenzy it would need to be irrational money managers that somehow clear their purchases through a panel and none of them say anything. Some point out that blackrock is on the other side of this zillow unloading but they have trillions in assets and only 60 billion in real estate. This is the largest money manager in the world and thats all the exposure they have.
    - 5y ago
wpietri 5y ago
I don't know how many people remember the Lean Startup movement and Steve Blank's book "Four Steps to the Epiphany" which is about how to take one's product hypotheses and rigorously validate them before scaling up. But somewhere there's a side note that basically said, "If we end up in another bubble, ignore all of this, because in a bubble being careful and capital-efficient isn't a great strategy." I think about that a lot these days given the absolute oceans of capital sloshing around. I'm glad you got a piece of it! But this is a really weird way to redistribute excess wealth.
rflrob 5y ago
My favorite story of algorithmic pricing gone mad was discovered by a postdoc in my grad school lab, leading to an out of print book being listed on Amazon for over $23 million: https://www.michaeleisen.org/blog/?p=358 https://www.michaeleisen.org/blog/?p=358
- tomcam 5y ago
  I feel pretty smart because I immediately found that same book on eBay for only $8 million
  - iab 5y ago
    Textbook arbitrage - congrats on the fortune
    - TheCoelacanth 5y ago
      Unfortunately, I have a feeling that the bid-ask spread on that book is approximately $23 million.
      - tomcam 5y ago
        So that’s why I couldn’t flip it
    - CRConrad 5y ago
      How do you know it was a textbook?
- bduerst 5y ago
  Whenever I see items like that, all I can think is embezzlement or tax evasion - but then again you'd think they would do that on less unique items (like $1200 for playstations).
- withinboredom 5y ago
  I worked on software that did algorithmic pricing. It was a race to the bottom because some bozo wanted to be on the first page of sellers. So they’d price it one penny under the lowest price. The next seller would match it. Before you knew it, the product was worth dollars or cents. We also had people do it on purpose (we called it price bombing), but that was easy to negate since we filtered out competitors with low ratings. But there went that bozo, price matching and beating the price by a penny… It was madness trying to code around people not thinking about their effects on the market and still turn a profit. Spoiler: we never did turn a profit on bozo powered listings, because at some point, it’s better to free the inventory space instead of waiting for the bozo to sell their items and let the price climb back up over the course of months or years.
ChuckMcM 5y ago
Very much like the $5000 books on Amazon where the algorithms are anticipating flipping an existing listing. Congrats on winning this round! The next step might be interesting which would be to go back and re-purchase the Condo once the price corrects to the 'real' price and bank the difference. I have read this article and the one before it and it sounds very much like some folks who fell in love with their own idea and got early feedback that it was "working", went all in and never checked to see if it was still working. The lesson should have been, "When you perturb an emergent system, never assume that the current state is the new steady state." The most interesting thing for me back when I was in college was a discussion on of the professors on feedback systems gave on LA's freeway systems. There are three major interconnected freeways, 405, 110, and 10 which at the time formed a triangle. Now there is the 105 which cuts off the tip so perhaps a smaller triangle. The professor shared a paper that tracked "brake waves" which were aggressive braking maneuvers that would "propagate" backwards on crowded freeways. The paper showed that at the right time of day, an aggressive braking on any of these three segments could result in your own braking wave to "lap around" and hit you again. Sort of a ringing of the system. This sort of effect can be present in any system that isn't centrally organized but is instead a result of the interactions within the system. Buying and selling real estate is such a system, especially when you are a 'market maker' in that system. There are a lot of papers on how HFT trading algorithms interfere (both constructively and destructively) with each other on wall street exchanges. Just interesting stuff in my opinion.
- iab 5y ago
  Traffic fascinated me - please post a link to that paper
  - ChuckMcM 5y ago
    I will see if I can find it, not all papers published in the 80's are online. (and I just looked the professor is emeritus at USC now but still around so I'll send him the question too)
    - ahmadss 5y ago
Alex3917 5y ago
> Opendoor matches Zillow who matches Opendoor and that's how you get ever increasing offers. Or Opendoor buys a house from themselves at way over market to trick Zillow into bankrupting itelf.
gibsonf1 5y ago
That doesn't sound like an algorithm mistake, just like a policy to match sales nearby at the same price because of the assumption that prices will just keep going up coupled with the stupidity of not tracking what you've bought so far - yikes! Oh, on second thought, I guess it could be algo as that previously bought nearby condo could have been priced using some crazy algorithm.
- not2b 5y ago
  "Prices will keep going up" was a key belief that triggered the 2008 real estate crash. They keep going up until they don't, and then the house-flippers who weren't careful enough and the lenders that approved "liar loans" get burned badly.
  - chii 5y ago
    > the lenders that approved "liar loans" get burned badly. the lenders "knew" of the bad loans, but because these lenders are going to on-sell the loan to some other investor, they didn't care as they profit off the sale, rather than the loan repayments. It's a moral hazard.
    - erik_seaberg 5y ago
      The loans were also bundled into opaque securities and the “low-risk” tranches were rated and priced for a typical rate of uncorrelated defaults, not a widespread crash.
      - aaronwall 5y ago
        The rating agencies would rate just about everything https://www.cnbc.com/id/27321998 https://www.cnbc.com/id/27321998 Official #1: Btw (by the way) that deal is ridiculous. Official #2: I know right...model def (definitely) does not capture half the risk. Official #1: We should not be rating it. Official #2: We rate every deal. It could be structured by cows and we would rate it. And the parts that were so junky they couldn't be rated got remixed into another deal to get rated.
lindseymysse 5y ago
I wonder how much of our economy, that alternately humming along and collapsing around us, is simply algorithms talking to eachother? How much economic movement is being generated by arbitrage ( https://www.ecomcrew.com/amazon-arbitrage/ https://www.ecomcrew.com/amazon-arbitrage/ ) ? How much online traffic is generated by advertising bots interacting with each-other? How much are housing prices being juiced by algorithms? I think the environmental crisis is already effecting the supply chain, I don't think we humans can see it yet. Are we Soviet Russia circa 1931, pretending like everything is going swimmingly while an enormous amount of us are about to starve?
- JasonFruit 5y ago
  Can you elaborate on what makes you think environmental factors are affecting the supply chain?
  - lindseymysse 5y ago
    Well, first of all, an unequal food supply chain has generated survival diets close to civilization, spreading new diseases to the entire world (hi Covid!, next door to bird flu). Second, there are droughts popping up all around the world, our fisheries are collapsing but our supply chain might be absorbing the problems of it until suddenly it doesn't anymore. That is my deep worry -- that we are seeing the start of a generalized system collapse.
mysterydip 5y ago
reminds me of some bots on amazon someone pointed out a while back. Basically the only two sellers of some obscure book are two bots with competing algorithms that keep slowly increasing the price.
cyanydeez 5y ago
this is yahoo 2: electric marketing dot com bust. it was purely about marketing driving up its own value via poorly considered metrics that invariably included positive reenforcement loops. eg, yahoo sells an ad for a company which sells ads on yahoo.
- aaronwall 5y ago
  Some of those were even formally recognized as revenue trading agreements to show growth.
bogomipz 5y ago
Is Opendoor an MLS/appraiser that also got into the flipping market? It looks like they are just another MLS service but the article states "It will be interesting to see how Opendoor reacts."
ashtonkem 5y ago
I think it’s a bit like when Amazon bots would occasionally drive up the cost of some random book into the millions, except they managed to fool each other into actually buying the damned thing.
lordnacho 5y ago
I spent a lot of time doing quant models for financial products. Two lessons: - It's easy to get wrong. If you see a lot of opportunity all the time, it's your model that's wrong, not the market. Very trivial things can mess up your model, like not understanding what the input data means. - Good trades are hard to get, common rule of thumb that your boss will tell you. Someone selling a house too cheaply? Either there's lots of other buyers or there's something about the house that you didn't think of. I remember seeing an FX volatility model that almost never traded. It always said "the market is right, plus minus costs". Now and again it would ding and we'd carefully try to get it done in the market, you wouldn't do like Zillow and surprise all the sellers with a massive payday. I also wrote an arbitrage bot from Bitstamp to MtGox once. I looked at it, didn't turn it on. Percentage wise it was a massive arb, but you couldn't see this in the raw numbers: credit risk on one of the legs. It just shows you that you still need to understand how things actually work. The model is only a calculator for quantifying opportunities that you understand. This is perhaps the oddest thing about this story. Zillow must have run into several situations where they were paying more than what's sensible, and their staff must have reported this? Surely at the start of an algo buying program, you are vigilant to evidence that the program is wrong?
- sennight 5y ago
  > Percentage wise it was a massive arb, but you couldn't see this in the raw numbers... Counterparty risk, that is where the profit came from. I did the same thing, but went into it knowing what the exposure was and how to mitigate it: don't leave crypto in an exchange's hot wallet any longer than it takes to execute a trade and take profit only on exchanges that you could legally pursue in the event that they fail to execute a USD transfer. I also anticipated the debanking that followed... as far as I know I'm still blacklisted by one bank and two money transfer services. What I didn't anticipate was how many exchanges would get hacked and what that would look like to anybody who aggregated the transactions: I get cold called by actual financial institutions a few times a year, always looking to bulk up their dark pools - they somehow have it in their heads that I'm sitting on billions of dollars in BTC. They likely don't have enough of the puzzle to put together the fact that arbitrage doesn't take much when you only need three confirmation blocks - USD wire transfers were the bottleneck.
bserge 5y ago
So when will algorithms start outbidding people on houses they can afford? Finally, found a home you can buy? ClosingDoors, Chuggle and WetDreamVerse make a better offer near instantly. Oh and it doesn't stop there. Now you can't afford a car, either. Or a computer. Or a power generator. But it's fine, just rent them from Memezon!
thom 5y ago
In addition to back testing models like this, it’s always worth running forward simulations of what would happen if some/most/all agents in the simulation shared your new strategy.

AareyBaba 5y ago

I suspect the 'AI software' running these companies is using linear regression to predict housing prices and one of the inputs is the price of similar houses nearby.

c141charlie 5y ago
What could possibly go wrong with that approach? :-)
danrocks 5y ago
Maybe some reinforcement learning algorithm running as well, where the maximum reward is winning the bid. Hence the algorithm just goes and lays waste on other bidders. Gotta get that infinite reward!
- mrfox321 5y ago
  I doubt RL is involved. That would be soooo risky. Bidding algorithms can model the price of winning the auction, you can maximize profit without fancy RL
nitwit005 5y ago
I recall in one of the previous crashes it turned out a firm had a model that couldn't handle homes going down in value. A linear model is superior to that.
jitl 5y ago
This is basically how human-lead appraisals work in most markets.
streetcat1 5y ago
should move to xgboost.
SantalBlush 5y ago
In fairness, I would think humans include nearby home prices as one of their parameters as well.
- LanceH 5y ago
  A Broker Price Opinion (BPO) is typically set by appraisal, comparable sales and comparable listings. Depending on how cookie cutter the neighborhood is, there are many different levels of "appraisal" from drive by photos to a deep inspection.
- KingMachiavelli 5y ago
  Human/realtor appraisals are often very poor. They will take 3-4 nearby & similar homes and basically add/subtract the differences from the home they are comparing. Then they basically just average the adjusted sale prices. They might add a bit onto that price since prices go up over time. So this has some obvious issues: * Areas without a ton of very similar houses that have also sold recently will basically have no 'comps' to use. * It's really easy to keep identifying differences (pros/cons) until the adjusted prices equal each other but it's hard to know if all of the meaningful differences have actually been identified. * It gives average homes and average buyers a huge advantage. This model assumes that the housing market is hot but not hot enough that someone will pay 10-15% more for some specific feature. Anything unique to a home/property is only going to be worth a fraction of the time & money it would cost to add. Anything super common (kitchen remodel, finished basement, etc.) can actually add 100% or more of it's cost to the houses value because the buyer is paying with 5x or more leverage so paying a bit extra to have it included. This is also why it's often better to fix a few things as the seller than give a discount on the sale price - the buyer will often pay a premium to get things move-in ready since it doesn't impact their monthly costs significantly.
devnull3 5y ago
lol ... this infact is the sample problem which Andrew Ng explains in his famous Machine Learning course on Coursera!
- hbarka 5y ago
  Wait, can you elaborate? I’m sincerely interested in going over that material.
davnn 5y ago
There was a Kaggle Competition in case you missed it. At least here you can see what kind of features they are using and what models people employed. See https://www.kaggle.com/c/zillow-prize-1 https://www.kaggle.com/c/zillow-prize-1
- xapata 5y ago
  Zillow Offers team was isolated from the Zestimate team. They didn't share algorithms.
  - laminarflow 5y ago
    AFAIK this is false; Offers began transacting at the Zestimate price in several markets earlier this year
    - xapata 5y ago
      Could be stale information. I talked to them a while back. Still, transacting at the Zestimate price doesn't mean they're privy to the Zestimate algorithms.

bombcar 5y ago

I've been noticing lots of "new lower price" emails from Zillow; I think the market has cooled off - but how much of it being hot was this investor competition?

crackercrews 5y ago
It's also the last rush of sales before things slow down for the holidays. There's also a chance that interest rates will be higher next year. That would bring home prices down.
datavirtue 5y ago
A lot of people were jumping in with their rundown old houses asking rediculous (lose your ass) prices and I'm seeing them all revise as of late. This was seen mostly in rural areas--Im looking for a farm. I saw quite a few houses that sold a few years prior for much lower prices. The disparity in estimated values and previous sales versus the list price was astronomical in most cases. The banks will happily give you a loan for whatever price so there is no check on the runaway prices except for your own personal knowledge of the historical market demand in an area.
- coffeecat 5y ago
  I'm not sure I follow your logic; if you're concerned with buying a house now, then prices from years ago are irrelevant. You can speculate that prices have risen too quickly and that they'll revert back to their earlier level, but that's ultimately just speculation. Comparable sale prices from the past few months are the thing to look at, to determine an offer that's likely to be accepted. One thing to keep in mind about list prices: if the sellers do a good job of estimating the current market price, then the house will sell quickly. If you're checking listings infrequently, there's going to be some sampling bias in the listings you see; overpriced houses that aren't selling will be disproportionately represented in active listings.
iav 5y ago
price reductions are seasonal in real estate. Peak new listings are around Super Bowl Sunday and Labor Day. There are very few price reductions in that period. The best quality homes at attractive prices sell first. By late October, the total number of listings on the market starts to decline, and price reductions become common.

onlyrealcuzzo 5y ago

TBF - wasn't the Phoenix metro up almost 40% in one year? After a 14% cut, that would still be an absurd YoY increase (I think still the largest single year increase in a major metro in the US on record?). I get that Zillow is losing 14% - but the market is not yet even back to normal appreciation - let alone "crashing".

roland35 5y ago
Here is the case shiller index for Phoenix - it is absolutely wild since 2020! https://fred.stlouisfed.org/series/PHXRNSA/ https://fred.stlouisfed.org/series/PHXRNSA/
- lsh123 5y ago
  One thing I don’t understand is that many people who buy houses in Phoenix believe in Global Warming getting worse in the short term (source: lived in Phoenix for a year, considered buying a house there, and spoke with many folks who just moved or were about to move).
  - KingMachiavelli 5y ago
    Hot and dry is not that bad from a global warming standpoint. Evaporative cooling is very effective and efficient so it's actually less energy intensive heating in colder locations. The plentiful amount of space and days of sun makes solar energy very practical. The nights can get pretty cold so it means that certain evening hours are actually very comfortable.
    - xapata 5y ago
      Except for drought problems.
  - railton 5y ago
    Have you read "The Water Knife" by Paolo Bacigalupi? Mostly set in Phoenix when the water has run out. https://www.goodreads.com/book/show/23209924-the-water-knife https://www.goodreads.com/book/show/23209924-the-water-knife

roland35 5y ago

You think Zillow would have one person take 15 minutes to do a quick look at the offer prices first!

aaronwall 5y ago
They could have charged a review fee to have an appraiser visit & make accurate price estimates & then let the homeowner keep the Zillow certified valuation label if they don't sell it to Zillow.

monksy 5y ago

Oh the new REITs playing the algro game. This is cute.

analyte123 5y ago

If local real estate agents and developers can manage to manipulate the "Zestimate", and I'm pretty sure they are at least trying, you can imagine the payoff.

Hjfrf 5y ago
Developers could certainly do it. A single padded first sale on a new development would be factored into offers. Classic wash trade.

avidphantasm 5y ago

Gee, what could go wrong with over fitting incompressible models?

monkeybutton 5y ago

Madness.. Its not like such a feedback loop hasn't been encountered before in systems. I like this example: https://www.michaeleisen.org/blog/?p=358 https://www.michaeleisen.org/blog/?p=358

stefan_ 5y ago

It's hard to tell if this is "tHe MaRkEt Is CoLlApSiNg" or just a Knight Capital 2.0. Leaning towards the latter.

qqqwerty 5y ago

This is an interesting lesson in algo trading run wild. Roughly, house prices have a fundamental ceiling, and it is determined by the size of the monthly payment that banks will allow the typical homebuyer to assume when approving a loan. If Zillow was the only iBuyer in a particular market, then that ceiling would likely have held, as all other non-zillow sales would still be operating under the loan approval constraint, and that would get reflected in the comps. But in a market with multiple ibuyers, none of which are capital constrained, it would make sense that they run up the prices against each other, past what the local homebuyers can afford. Not sure how much that actually played a role here. But could be fun to do some back of the napkin math to see if that was the case in some of these markets.

derefr 5y ago
> house prices have a fundamental ceiling, and it is determined by the size of the monthly payment that banks will allow the typical homebuyer to assume when approving a loan. In the more irrationally-hot markets, why assume that housing is even being purchased on a mortgage by your average home-buyer, rather than being purchased cash-in-hand by a private or corporate investor looking to park wealth they've already generated? (Even if that isn't the majority of houses in those markets, the sales like that that do happen would still have an impact on property values in the affected neighbourhoods.)
- trident5000 5y ago
  Because the real estate sector is like 30-40 trillion. 2007 was the result of a massive infusion of debt by unsophisticated retail "investors" who bid prices up far higher than even their rental value. Lending standards dont allow this today. If big institutional money is going to cause a frenzy it would need to be irrational money managers that somehow clear their purchases through a panel and none of them say anything. Some point out that blackrock is on the other side of this zillow unloading but they have trillions in assets and only 60 billion in real estate. This is the largest money manager in the world and thats all the exposure they have.
  - 5y ago

wpietri 5y ago

I don't know how many people remember the Lean Startup movement and Steve Blank's book "Four Steps to the Epiphany" which is about how to take one's product hypotheses and rigorously validate them before scaling up. But somewhere there's a side note that basically said, "If we end up in another bubble, ignore all of this, because in a bubble being careful and capital-efficient isn't a great strategy." I think about that a lot these days given the absolute oceans of capital sloshing around. I'm glad you got a piece of it! But this is a really weird way to redistribute excess wealth.

rflrob 5y ago

My favorite story of algorithmic pricing gone mad was discovered by a postdoc in my grad school lab, leading to an out of print book being listed on Amazon for over $23 million: https://www.michaeleisen.org/blog/?p=358 https://www.michaeleisen.org/blog/?p=358

tomcam 5y ago
I feel pretty smart because I immediately found that same book on eBay for only $8 million
- iab 5y ago
  Textbook arbitrage - congrats on the fortune
  - TheCoelacanth 5y ago
    Unfortunately, I have a feeling that the bid-ask spread on that book is approximately $23 million.
    - tomcam 5y ago
      So that’s why I couldn’t flip it
  - CRConrad 5y ago
    How do you know it was a textbook?
bduerst 5y ago
Whenever I see items like that, all I can think is embezzlement or tax evasion - but then again you'd think they would do that on less unique items (like $1200 for playstations).
withinboredom 5y ago
I worked on software that did algorithmic pricing. It was a race to the bottom because some bozo wanted to be on the first page of sellers. So they’d price it one penny under the lowest price. The next seller would match it. Before you knew it, the product was worth dollars or cents. We also had people do it on purpose (we called it price bombing), but that was easy to negate since we filtered out competitors with low ratings. But there went that bozo, price matching and beating the price by a penny… It was madness trying to code around people not thinking about their effects on the market and still turn a profit. Spoiler: we never did turn a profit on bozo powered listings, because at some point, it’s better to free the inventory space instead of waiting for the bozo to sell their items and let the price climb back up over the course of months or years.

ChuckMcM 5y ago

Very much like the $5000 books on Amazon where the algorithms are anticipating flipping an existing listing. Congrats on winning this round! The next step might be interesting which would be to go back and re-purchase the Condo once the price corrects to the 'real' price and bank the difference. I have read this article and the one before it and it sounds very much like some folks who fell in love with their own idea and got early feedback that it was "working", went all in and never checked to see if it was still working. The lesson should have been, "When you perturb an emergent system, never assume that the current state is the new steady state." The most interesting thing for me back when I was in college was a discussion on of the professors on feedback systems gave on LA's freeway systems. There are three major interconnected freeways, 405, 110, and 10 which at the time formed a triangle. Now there is the 105 which cuts off the tip so perhaps a smaller triangle. The professor shared a paper that tracked "brake waves" which were aggressive braking maneuvers that would "propagate" backwards on crowded freeways. The paper showed that at the right time of day, an aggressive braking on any of these three segments could result in your own braking wave to "lap around" and hit you again. Sort of a ringing of the system. This sort of effect can be present in any system that isn't centrally organized but is instead a result of the interactions within the system. Buying and selling real estate is such a system, especially when you are a 'market maker' in that system. There are a lot of papers on how HFT trading algorithms interfere (both constructively and destructively) with each other on wall street exchanges. Just interesting stuff in my opinion.

iab 5y ago
Traffic fascinated me - please post a link to that paper
- ChuckMcM 5y ago
  I will see if I can find it, not all papers published in the 80's are online. (and I just looked the professor is emeritus at USC now but still around so I'll send him the question too)
  - ahmadss 5y ago

Alex3917 5y ago

> Opendoor matches Zillow who matches Opendoor and that's how you get ever increasing offers. Or Opendoor buys a house from themselves at way over market to trick Zillow into bankrupting itelf.

gibsonf1 5y ago

That doesn't sound like an algorithm mistake, just like a policy to match sales nearby at the same price because of the assumption that prices will just keep going up coupled with the stupidity of not tracking what you've bought so far - yikes! Oh, on second thought, I guess it could be algo as that previously bought nearby condo could have been priced using some crazy algorithm.

not2b 5y ago
"Prices will keep going up" was a key belief that triggered the 2008 real estate crash. They keep going up until they don't, and then the house-flippers who weren't careful enough and the lenders that approved "liar loans" get burned badly.
- chii 5y ago
  > the lenders that approved "liar loans" get burned badly. the lenders "knew" of the bad loans, but because these lenders are going to on-sell the loan to some other investor, they didn't care as they profit off the sale, rather than the loan repayments. It's a moral hazard.
  - erik_seaberg 5y ago
    The loans were also bundled into opaque securities and the “low-risk” tranches were rated and priced for a typical rate of uncorrelated defaults, not a widespread crash.
    - aaronwall 5y ago
      The rating agencies would rate just about everything https://www.cnbc.com/id/27321998 https://www.cnbc.com/id/27321998 Official #1: Btw (by the way) that deal is ridiculous. Official #2: I know right...model def (definitely) does not capture half the risk. Official #1: We should not be rating it. Official #2: We rate every deal. It could be structured by cows and we would rate it. And the parts that were so junky they couldn't be rated got remixed into another deal to get rated.

lindseymysse 5y ago

I wonder how much of our economy, that alternately humming along and collapsing around us, is simply algorithms talking to eachother? How much economic movement is being generated by arbitrage ( https://www.ecomcrew.com/amazon-arbitrage/ https://www.ecomcrew.com/amazon-arbitrage/ ) ? How much online traffic is generated by advertising bots interacting with each-other? How much are housing prices being juiced by algorithms? I think the environmental crisis is already effecting the supply chain, I don't think we humans can see it yet. Are we Soviet Russia circa 1931, pretending like everything is going swimmingly while an enormous amount of us are about to starve?

JasonFruit 5y ago
Can you elaborate on what makes you think environmental factors are affecting the supply chain?
- lindseymysse 5y ago
  Well, first of all, an unequal food supply chain has generated survival diets close to civilization, spreading new diseases to the entire world (hi Covid!, next door to bird flu). Second, there are droughts popping up all around the world, our fisheries are collapsing but our supply chain might be absorbing the problems of it until suddenly it doesn't anymore. That is my deep worry -- that we are seeing the start of a generalized system collapse.

mysterydip 5y ago

reminds me of some bots on amazon someone pointed out a while back. Basically the only two sellers of some obscure book are two bots with competing algorithms that keep slowly increasing the price.

cyanydeez 5y ago

this is yahoo 2: electric marketing dot com bust. it was purely about marketing driving up its own value via poorly considered metrics that invariably included positive reenforcement loops. eg, yahoo sells an ad for a company which sells ads on yahoo.

aaronwall 5y ago
Some of those were even formally recognized as revenue trading agreements to show growth.

bogomipz 5y ago

Is Opendoor an MLS/appraiser that also got into the flipping market? It looks like they are just another MLS service but the article states "It will be interesting to see how Opendoor reacts."

ashtonkem 5y ago

I think it’s a bit like when Amazon bots would occasionally drive up the cost of some random book into the millions, except they managed to fool each other into actually buying the damned thing.

lordnacho 5y ago

I spent a lot of time doing quant models for financial products. Two lessons: - It's easy to get wrong. If you see a lot of opportunity all the time, it's your model that's wrong, not the market. Very trivial things can mess up your model, like not understanding what the input data means. - Good trades are hard to get, common rule of thumb that your boss will tell you. Someone selling a house too cheaply? Either there's lots of other buyers or there's something about the house that you didn't think of. I remember seeing an FX volatility model that almost never traded. It always said "the market is right, plus minus costs". Now and again it would ding and we'd carefully try to get it done in the market, you wouldn't do like Zillow and surprise all the sellers with a massive payday. I also wrote an arbitrage bot from Bitstamp to MtGox once. I looked at it, didn't turn it on. Percentage wise it was a massive arb, but you couldn't see this in the raw numbers: credit risk on one of the legs. It just shows you that you still need to understand how things actually work. The model is only a calculator for quantifying opportunities that you understand. This is perhaps the oddest thing about this story. Zillow must have run into several situations where they were paying more than what's sensible, and their staff must have reported this? Surely at the start of an algo buying program, you are vigilant to evidence that the program is wrong?

sennight 5y ago
> Percentage wise it was a massive arb, but you couldn't see this in the raw numbers... Counterparty risk, that is where the profit came from. I did the same thing, but went into it knowing what the exposure was and how to mitigate it: don't leave crypto in an exchange's hot wallet any longer than it takes to execute a trade and take profit only on exchanges that you could legally pursue in the event that they fail to execute a USD transfer. I also anticipated the debanking that followed... as far as I know I'm still blacklisted by one bank and two money transfer services. What I didn't anticipate was how many exchanges would get hacked and what that would look like to anybody who aggregated the transactions: I get cold called by actual financial institutions a few times a year, always looking to bulk up their dark pools - they somehow have it in their heads that I'm sitting on billions of dollars in BTC. They likely don't have enough of the puzzle to put together the fact that arbitrage doesn't take much when you only need three confirmation blocks - USD wire transfers were the bottleneck.

bserge 5y ago

So when will algorithms start outbidding people on houses they can afford? Finally, found a home you can buy? ClosingDoors, Chuggle and WetDreamVerse make a better offer near instantly. Oh and it doesn't stop there. Now you can't afford a car, either. Or a computer. Or a power generator. But it's fine, just rent them from Memezon!

thom 5y ago

In addition to back testing models like this, it’s always worth running forward simulations of what would happen if some/most/all agents in the simulation shared your new strategy.