The Zillow estimate used to be spot on in our neighborhood, for several years even. Recently they all plunged to 10-20% below market value, even though all sales data puts it at the previous mark. The weird thing is that they also went back and recalculated old "Zestimate" data as well. On April 2nd the estimate for our house was $617,453, which is more or less what it would sell for. It had been above $600k since June, 2016.
This month it was $544k, and there's no sign it ever got above $580k in the historic graph. It claims it was $546k on March 31st.
One thing I did notice was a new button underneath the estimate: "I disagree with my Zestimate" which, surprisingly enough, is a lead gen for realtors, not a method to try and get them to change.
I know it seems weird to have the "old" data change. But, it's not observed data, it's estimates made at a point in time. The estimate comes from the property itself as well as the algorithm. If they changed the algorithm, do you think it would be better to recalculate the old values, or show a huge swing in your home value, implying some sort of market trend or issue with your house? If I were implementing it, I think I'd have to change the old data. I'd like to think I could sneak in an explanation / comparison, but that might get nuked for usability/UI reasons.
I used to run an online service predicting college admissions. As I would update the algorithms to incorporate new results, users' predictions would change. At first, I thought people would be pleased. Soon, I discovered that people really don't like the idea of their prediction changing when they have done nothing to cause it to change. Subsequently, I would lock in the values that had been "seen" by the user, and allow them to update their prediction, if they wanted to, when new models came online. This eased a lot of the users' expressed frustration.
They should be more transparent about their algo though. That would then allow users to actually know how they came to the conclusion that they did. Hopefully this lawsuit will reveal that algo.
Why would the lawsuit force them to reveal trade secrets? I don't see how you have any right to know how they arrived at the number they did. I can say publicly that I value your home at $5 and my neighbor bob can say its worth a million dollars.
These are both opinions and anyone seriously considering spending half a million or more can consider spending a modest amount on an appraisal.
There's a difference between an opinion proffered by a neighbour and an opinion from a large real-estate company.
The reality is people do trust the site, so law should take that into account. If a situation is making life harder for actual people, the fact that their behaviour isn't economically rational is not a good enough reason to ignore the problem.
Zillow is a information company providing a free ballpark estimate of a vast array of properties, not a real estate company. This is an important distinction. Neither party is paying them for an accurate estimate ergo they are not obligated to provide an estimate sufficient for either buyer or seller, which is why both sides are best advised to PAY for an accurate estimate before closing a transactions worth hundreds of thousands.
I'm honestly not seeing the problem here. Don't such people already hire experts to asses value before spending half a million dollars? I'm seeing the claim that inaccurate estimates lead to difficulties selling but I have yet to see the proof. Maybe those having trouble selling really are asking for more money than people are willing to pay.
...and Uber is just a software company, and Mastercard/Visa just facilitate payments.
The specific classification of a company is irrelevant. If they've induced their own gravitational field, then they're responsible for the consequences.
Assessors are obligated to be qualified because they are obligated to their clients who pay them money in exchange for their services.
How much did the seller in question pay zillow and what was promised in return?
The fact that person a's action foo can negatively effect you doesn't imply that the law or ethics precludes a from doing foo. A must be either violating your rights, the law, or have a duty to you that would preclude foo.
Its not clear how you could have a right to silence others speech in order to earn more money, its not clear what duty zillow has to sellers, and its not clear that the law actually precludes zillow from underestimating the value of your property.
In fact it looks an awful lot like none of those apply, and further that allowing anyone who feels like their property is undervalued to use threat of a lawsuit to silence discussion would preclude anyone from providing public analysis of property values whether accurate or not and thus harm the public at large. Real harm unlike the unproven harm previously discussed.
Zillow, should make it very clear then that their estimates are complete works of fiction and any resemblance to actual value is purely coincidental and unintended.
Well, it's really not a complete work of fiction... it's a correlated value estimate based on previous sale, and other property sales in a given market over time based on factors of the home (size, bedrooms, bathrooms, etc)... It may well not be accurate, as there may be other features that increase or decrease value.
That said, I would posit that if you're selling for more than a 20% variance from the Zestimate, that you are probably over/under-valued for the market. Just because you have gold-leaf throughout your home doesn't mean that it'll sell for a million when all your neighbors are half that.
When you seek the services of an agent and pay them money to help you buy or sell real estate they represent to you that they are qualified to assess the value of the real estate being bought and sold and are somewhat morally obligated to you by virtue of the money you paid them to provide you with accurate information.
Professional licensing requirements are supposed to ensure they are sort of kind of qualified.
Someone you didn't hire which is providing a ballpark guess on the value of your property to help people do preliminary research before they buy has minimal enough obligation to buyers and none at all to you.
We can see that the obligations of the agents above attach both to the fact that they are brokering 6-7 figure transactions and to the fact that you contracted them. Its hard to imagine what you believe the obligation is supposed to attach to in zillows case.
Its not even clear that regulating zillow according to each individual states laws would accomplish save placing unconstitutional limits on free speech. Plantiff doesn't even want that they want a club to compel zillow's silence or desirable speech. Surely cloaking it in bs doesn't make this any less unconstitutional.
The problem in Zillow's case may simply be the legal system... And that will vary greatly from state/county to state/county. The real estate market is very protective of itself.
Their algorithm is the very core of the Zestimate product. The value of the IP is incredibly important for the company, and they would be very stupid to reveal any details of its inner workings.
I live in a hot real estate market (L.A.), and I've spent about 5 minutes (no exaggeration) looking online at the prices of houses in my neighborhood, knowing the actual closing price of recent sales, and my first thought was that prices were about 10-20% below market value.
It's funny that this is obvious to someone like myself who has spent virtually no time dealing with real estate, let alone people heavily entrenched in the process.
This is the same in Seattle, where a house's asking price can be 550, its zestimate (or Redfin estimate) will be around 570, and the price it actually goes for invariably ends up around 700. This has been the case for months, and it can almost seem like some sort of conspiracy between realtors and these meta-agencies to game the market. There are plenty of comps that show prices are WAY above what the 'estimates' show, and yet these professionals are seemingly oblivious. Fear of another 2008? Too much success drumming up bidding wars by underpricing? Whatever the reason, it's pretty infuriating as a buyer because it creates a sort of information asymmetry which makes a major decision really painful.
The properties in GP's post are not likely to be sold at auction - very few homes are sold that way in the US.
The 'bidding war' referred to is a standard listing/sale that receives multiple offers. If the interest is high enough, the seller may counter all of the offers and reveal the highest price, forcing the bidding.
Even without revealing the highest offer, the offers themselves constitute a silent auction, which is where a lot of the information asymmetry comes from. If I could speak to each of the other people and tell them I had X amount to spend we could all agree to walk away or make offers without going through a bunch of hassle and (for 32 of 33 offerers) heartbreak. The process really highlights to me how broken certain types of 'market' can be when there's a monopoly of information.
toomuchtodo: I am not sure if i would call it "true" value. The true value is what you'd get if you were to sell it again. Maybe at that same day, which sounds to me like the runner-up bid.
> It's funny that this is obvious to someone like myself who has spent virtually no time dealing with real estate, let alone people heavily entrenched in the process.
I think you're completely missing the point of the tool and the last sentence in OP's post... They're getting it wrong on purpose to get signups.
I really like the Zestimate tool (if it's accurate), but it's pretty sleazy for them to use it this way. Most people won't research the market price for every house they see, and as long as the Zestimate is reasonable, they'll assume the sales price is too high.
My neighborhood is the opposite. Zillow will value a house at, say, $650k and it sells for $520k. It's not the condition of the house, either, since it's pretty consistent.
For my address, it's currently showing 468k, and says that's a drop of 137k from one month ago. I'm positive that 468k is around 100k too high to be realistic. I bought for 265 in 2010, and I've been seeing units around me sell for around 350k recently. Redfin agrees with me, showing sales between 303 and 380, depending on unit size and upgrades.
It's a rough estimate, but it is not always lower. A place near me was listed for $625k below the Zestimate, and is selling for 725k below the Zestimate (which was $2.4M). A below market rate sale in your area, such as a foreclosure auction, can drive automated price estimates down. A transfer of ownership to a trust, which can be at an old sale price or even zero dollar, is tough to discern in the metadata of some localities and will drive the $/sqft down.
Given the many wild overestimates, I don't suspect that they are trying to have their model dampen price estimates to drive referrals. However, the price estimates at Redfin seem much more accurate, so their model could use some work.
I just went and tried this. The Zestimate for my place is... ludicrous. Probably 30% below market value. What's more, similarly sized apartments on the same floor vary by 15%!
I'd always thought this number was dumb but wow. This is even more ludicrous than I could've believed.
How can it be this far off with apartments with recent sales history in a building with lots of unit (and thus lots of sales activity)?
Could this be a case of them A/B testing on the Zestimate model to maximize lead generation with the "disagree?" link?
similarly sized apartments on the same floor vary by 15%
Is that so unreasonable. In the building I used to live in for example the apartments on one side of the house had a beautiful view from the balcony and the ones on the other side didn't. The view side apartments consistently sold for more. I've also looked at apartment buildings where the floor size is basically them same for two apartments, but the layout was different, and people where willing to pay more for the apartments with the better layout.
Highly unlikely. You do not cause this level of pricing swings in an A/B test. Additionally, A/B testing this much of a price swing heavily discounts the value of the tool.
They have my house around 680K. If I were to guess what I could really sell it for I'd guess around 580K, based on a recent nearby sale. My house is a little odd and there aren't many good comps for it.
At one point a few years ago the Zestimate was about 540 and my house was appraised for 450.
And when we bought it 5 years ago the Zestimate was at 380 and we bought it for 420.
I once heard to expect the Zestimate to be off by ~20% in either direction. That seems to be holding here.
It has been at least $100k way too high in our neighbourhood for well over a year. I'm not sure quite how they go so far off, there has been a slow but steady rate of house sales around to set their data from. Consistently the Zestimate has been noticeably over the market value, and it has been annoying sellers.
I just looked at my house and I think the Zestimate is too high.
I bought the house 9 years ago and the Zestimate is about $10k more than I paid. It's right in line with what the previous owner was asking when he sold it but I wasn't willing to pay that and I likely couldn't sell it for that much.
If true and you can prove it, this may be valuable information in prosecuting Zillow for some pretty illegal and perverted stuff. I hope you will share this information with someone who can punish Zillow for this.
Punish zillow for what? Seems to be a confusion between something would benefit me/may harm me and the other side having any sort of obligation to consider your position.
An inaccurate estimate from someone who you paid nothing and whom owes you nothing may make your life more difficult but this doesn't imply they owe you anything.
Market manipulation is illegal. Whether or not there's a business relation between buyer and Zillow or seller and Zillow is immaterial. All the DA needs is an initial suspicion that someone related to Zillow profited from the manipulation.
What market manipulation? First we haven't seen proof that zillow's zestimate is effecting real prices. Presumably after initial research stages people have an actual assessor come in before closing.
Then you have to show that they profited from such manipulation rather than you know from ads.
It looks more like they changed their algorithm and rather than show imaginary swings in the value of homes revised their historical projections. This makes tons of sense if you understand the historical data to mean this was what the property was worth in 2010 as opposed to in 2010 this is what we thought it was worth.
In fact this interpretation is the logical one. Nobody cares what zillow thought in 2010 they care what the historical value was.
> The weird thing is that they also went back and recalculated old "Zestimate" data as well (...) there's no sign it ever got above $580k in the historic graph.
It makes it illegal to "develop a real estate appraisal" without a license. An appraisal is defined as simply "an opinion of value". So if you are driving down the street and say to your spouse, "Look at that house, it's gotta be worth a million bucks," you just committed a misdemeanor.
This law as it's written has got to be unconstitutional.
> nor does this Act apply to the procurement of an automated valuation model. "Automated valuation model" means an automated system that is used to derive a property value through the use of publicly available property records and various analytic methodologies such as comparable sales prices, home characteristics, and historical home price appreciations.
If that sentence applies, I don't think they really have a case...
Good point. But what does "procurement" mean in this context? The plain meaning is "obtain", but that doesn't really make sense here. The word isn't used elsewhere in this document but is presumably defined somewhere.
If it is interpreted to mean "use", then the case is a loser as you say. But I wouldn't want to be the lawyer tasked with arguing to the judge that "the legislature meant 'use' but decided to say 'procurement' instead".
Generally in law you use a law dictionary, e.g. http://thelawdictionary.org/procurement/, which defines it as "the entire process of purchasing goods". "Entire" there suggests a broad term, hence interpreted inclusively in the context it appears. So in context it's probably something like "commercial use". But there are other questions of applicability, e.g. whether the legislature intended that sentence to only apply to financial loan decisions. It seems like there's no indicator of legislative intent on the original bill https://openstates.org/il/bills/96th/HB1015/, hence it would probably be the plain meaning as an independent clause, but IANAL...
In my experience (7 years as a corporate lawyer), a law dictionary would only be consulted for legal terms of art, which "procurement" is not. For most terms, a regular dictionary definition is used.
At any rate, I don't see how the definition cited would clear up the matter. It's a big leap from "the entire process of purchasing goods" to "commercial use". Like "use", which I discuss above, "commercial use" is a common term of art, and it would be strange to say "procurement" if what you mean is "commercial use".
Well, I was trying to shift the focus to "automated valuation model". That's used in other places and seems to be a term of the art in finance, e.g. this Kansas consumer bill: http://kslegislature.com/li_2016/b2015_16/measures/documents... Hence Zillow probably doesn't qualify, since it doesn't do loans or banking.
But the courts are so random; I'm guessing this case will be determined primarily by how much the judge that gets assigned likes technology. It'll be interesting to follow up on.
I'm not sure I understand your point. By visting the website one is procuring the estimate through an automated model in return for the ads and tracking you agree to by visting. Legal language is straight forward in this case.
It does because in this sense the model is the result of the automated valuation. You are thinking too much like a programmer when you think the automated valuation model is the model being procured.
I agree. So where is the line between an "off-hand comment" that's not subject to the law, and an appraisal that constitutes a misdemeanor? Is it illegal if you post your comment on Facebook? Or on your blog where you talk about various homes for sale in your neighborhood? Or only when you start charging people money to hear your opinion?
A citizen cannot look at the law and figure out if they're committing a crime, which is why I said the law is unconstitutional.
This is pretty typical of professional regulation language. Go look at your state's definition of "counseling" or "practice of law." At the limits, it covers almost any advice to anyone.
It's usually more difficult to sue, though. The power is usually given to some semi-governmental group. The Courts go out of their way to find professional regulations constitutional, and the regulators try to avoid claims that might get the entire regulation struck down. That's why Illinois never goes after non-commercial speculation about house prices.
But sometimes they overreach. There was an article in the WSJ a few days ago about some state's Barber Commission levying $50 fines on cosmetologists with striped poles out front -- for improperly implying that a licensed barber was on premises.
This woman probably isn't suing to enforce the regulation, as such. She could sue without the professional regulation, but the law sets a professional standard of care for appraisers that Zillow doesn't claim to meet, so she hopes it will help her case.
I imagine if you offer your opinion about the price in a professional context where your opinion can affect the actual value of the property then you might have to follow some rules.
If you drive by a house and say "this house doesn't look like it's worth shit" you're probably not going to wind up in court after someone complains that said house is worth quite a bit more than shit and you're driving the price down.
Laws actually can be vague while remaining Constitutional. Sometimes it's unavoidable. Court cases can be used to refine the edges.
Actually this has come up before with a similar law in Arizona, where the state tried to ban Zestimates. Volokh's analysis basically agreed with me (http://volokh.com/posts/1177792494.shtml). Though I couldn't find any info about what ultimately happened with that case.
There was never an actual legal case. There were just cease and desist letters with threats of a case. Before any actual legal case could materialize the law in AZ was changed to allow automated appraisals without certification as long as no money changed hands for the appraisal.
The line is almost certainly when you start using the estimate in the context of conducting commerce, which is a pretty straightforward and understandable distinction.
> A citizen cannot look at the law and figure out if they're committing a crime, which is why I said the law is unconstitutional.
Zillow is not a citizen.
They're also in the business of providing these numbers, which is obviously far from the same thing as making a comment while driving down a street. I think you're trying to blur a line that's pretty clear to most reasonable people.
IMDb now hides the ages of many actors[1] with the idea that unless they supply their own birth date it could be facilitating age discrimination. (I'm not sure if the Wikipedia article has the latest legal status or decisions)
A prohibition against unlicensed appraisals seems about as valid as hiding what is public information if you choose to make the effort and know where to look. However, I can see both decisions being overruled if appealed to higher courts.
> So if you are driving down the street and say to your spouse, "Look at that house, it's gotta be worth a million bucks," you just committed a misdemeanor.
No, but if you try and sell your opinion to make money that's when you're very likely crossing the line. An appraisal is something you pay for or make money from; opinions are not.
Unfortunately, what the law is is generally determined by more than the text of the law, but the law plus previous precedence so it's never enough to just look at the text of the law.
Realtors have always hated the Zestimate with a passion [0]. When they publicly called out Zillow CEO for selling his own home for 40% less than the Zestimate on it, Zillow magically released a superior algorithm [2].
[1] reminds of an interview with the founder of Ashley Madison.
He was going on and on about how western cultures were so regressive and prude about adultery and sex in general. The interviewer then asked him if he had ever used his own service. This elicited an uncomfortable non-answer that basically boiled down to "No".
"non-answers" are never good, of course, but i can easily see a way he could give a workable answer to that question. for example, he could say that some people are trapped in loveless marriages they can't get out of, but he never has been, so he's not the target customer.
of course i'm just playing devil's advocate, because all evidence points to ashley madison being a scam, with essentially no women on the site.
You would think real estate would prove to be one of the few real life fields where machine learning would be able to predict the prices pretty accurately. You have:
1. List price
2. "Favorites" on the MLS.
3. Square footage
4. Bedrooms
5. Bathrooms
6. Sold price for nearby properties
7. Sold price for nearby properties relative to their list price
And the list goes on... Is it that difficult to predict? Literally there are hundreds of other features that have a clear correlation to the sell price (amount of people who attend the open house, crime, proximity to busy street, etc.) that I'm not including. I guess the problem is, there's not any money in having an accurate prediction.
Disclaimer: I'm a former employee of Zillow Group, and still hold stock.
These online real estate listing sites are using machine learning to try and predict sale price. The main difficulties are:
1. 'Machine Learning' is a broad term, with lots of potential approaches, some of which could give more or less accuracy.
2. If you put too much effort into getting the prediction accurate, you're prone to overfitting.
2. The data is messy. Really messy. Really really messy. The major players in the field often have a vested interest in not rectifying the situation. Even when they do, the software that runs most MLS data exchanges is, well, old. To a certain extent, it doesn't matter if one particular house has accurate data, if all the surrounding data is inaccurate.
3. On the scale of things, home sales are actually pretty rare. Any particular home might only change hands half a dozen times, while the market demand for that home might swing wildly. Any given home sale has a pretty big influence on the valuations of nearby homes, and if that sale is 'inaccurate', i.e., the seller over- or under-paid what others would consider the fair price, it can have an outsized effect on algorithmic estimates of value.
4. It's much easier to tell when a machine-learned valuation is wrong, and by how much, compared with other common applications. If Netflix is telling you a movie is 4 stars, when you think it's more of a 3.5, it doesn't feel that "off" to you. If text-to-speech mistakes a word for another similar-sounding word, it feels understandable. When Zillow says you have $30k less than you think you do, it's more quantifiable, and has more emotional impact.
Tangentially, fun fact: while there's hundreds and thousands of potential features, if you do a PCA on it, basically 80% of a home's price is just "price per square foot in the nearby area". one feature. Of course, if you end up 20% over or 20% under based on that logic, you get sued.
Was the model aware of it's own error rate? For instance, could you say that this home is worth $500-650 thousand dollars?
Obviously, this is harder to convey to a homebuyer and for various reasons "hard" numbers are prefered, not least because they cause the illusion that you know what you are talking about.
If you had a really great model, you might be able to make some money with it--find underpriced (per the model) houses, buy them, and then try to sell them at the model's price.
Real estate markets are rarely liquid. Such models would have to take that into account, of course, and that isn't something I'd think is straightforward at all.
The reason I think it isn't straightforward to account for is based on several things: it isn't fungible, the purchase date can have little or no relationship to the price, and the means of purchase typically carry other time-based constraints that are also not necessarily related to price (although they can be). Furthermore, the liquidity in a given market may change fairly quickly and without any (apparent) other reason.
I haven't tried to develop a model for this, so this is just sort of a gut feel for how difficult feature engineering to account for how a market behaves in this respect might be. And that "gut feel" is mainly informed by my experience purchasing and selling property (for personal and rental use).
I'm not convinced that risk as used in this context contributes to price significantly in the residential real estate market. People in this market don't (typically) view transactions as investments and they don't view the home as assets. Lenders do, so the availability of mortgage funds may be linked to price and liquidity but I suspect that relationship is tenuous at best.
I imagine it's tricky because value per square foot varies quite a bit in adjoining neighborhoods. And sometimes within a neighborhood. And neighborhoods aren't necessarily well defined geographically. And, some neighborhoods go long periods of time without any houses being listed or sold.
> value per square foot varies quite a bit in adjoining neighborhoods. And sometimes within a neighborhood
The quality or rather "reputation" of the assigned public school is probably the best predictor of this variance. The problem is that the school's reputation can diverge from the official grades, so this is difficult to quantify.
Sure, but there are other drivers that would cause different value per square foot. A house with 12 foot ceilings, granite countertops, stone floors, wainscoting, built-ins, high-end appliances, tile roof, and so on...could be 100 yards from house with the exact same square footage, and none of that. It trips up the Zestimates around here. Especially in areas with larger lots, where bulldozing the old plain ranch house and replacing it with a higher end house is common. The per-square foot values are very different.
Edit: also, near power lines, near water tower, on cul de sac, next door to elementary school, etc.
That might be a good project actually. Try to predict previous houses prices from historical data. A retrospective analysis would be fun since you know what the "right answer" is and can adjust your model accordingly.
It seems this and predicting the stock market have some similarities, but I think real estate is less volatile.
>... buy them, and then try to sell them at the model's price.
At the liquidity and costs of marketing and buying-selling a property, this strategy would require you to find really high margin houses and those are snatched , marketed and sold within a week.
They buy houses from owners at a fixed price and then need to turn a profit on it. Therefore they need to have a good idea of what the eventual selling price will be.
There is a kaggle competition right now to predict house prices in russia. If you can do it better than anyone else, you can win tens of thousands of dollars.
There are also non-tangibles like knowing your buyer and how much you can get from them.
My mother passed away last year and I handled the home sale as I was the executor of the estate. The buyer was a neighbor who heard about my mother passing away and saw it as an opportunity to buy the house for so his mother could live less than 100 meters from him and help him and his wife with raising their kids.
He knew we'd be eager to sell and thought he'd get a deal, but I realized that the reason for him wanting my mother's house was because it was ideal for his particular circumstance. This buyer was a real estate agent in my mom's office (she was also a real estate agent). At the end of the day, I got about $40k premium over the appraised value (almost ~10% premium) just because I knew I had a motivated buyer.
Situations with motivated buyers and sellers happen all the time and adds a lot of volatility in sale prices, and is a confounding factor that I would expect to really complicate machine learning approaches to appraising the value of a home.
Usually this is mitigated by the appraisal. Lenders generally aren't willing lending more than a home is appraised for because of the increased risk. I'm sure this varies from state to state but I think this holds mostly true in general. That being said, the appraisals themselves are quite subjective and only based off a few similar properties sold recently, so I still agree it's a difficult problem.
Appraisals magically come in at the exact offer price unless the offer is just completely off the charts insane. Appraisals are often times BS or, at best, just whatever you need them to be within some rather large margin (+/- 10%? 20%?).
Sadly, my appraisal came in "magically" at list price, so I had to put down a lot more at close. It was 3 years ago so we've done well regardless. I think you are right, I'm pretty sure the appraisal is just a guess. Most people I know had the appraisal come in way high, so maybe I got the one honest person.
When I sold my last house I had no idea what it was worth (the market was going crazy at the time) so I called an appraiser to see if he would give me a good starting point.
The first question he asked was what I was listing it for. He also made it clear that he would not physically inspect the house or do anything more than a cursory comp search.
Real estate sales are just loaded down with parasites looking to skim off of every transaction. IMHO it is an industry badly in need of disruption, but sadly a lot of the middlemen have their positions enshrined in laws, regulations, and corporate policy. Another good example: why do we pay Realtors a percentage of the sale price instead of a fixed rate? They're basically glorified babysitters that are just there to make sure you aren't stealing everything from the homes.
There's plenty of money in it. Imagine if you could price the house correctly out of the gate, reducing the amount of time the home stays on the market. It is indeed a hard problem to solve using ML.
You'd need a few other things:
8. Floor count
9. Size and location of the bedrooms, bathrooms, kitchen
10. In-unit laundry (mostly for condos)
11. Proximity to neighbors
12. Existing tree coverage and other landscaping
13. Proximity to trails, transit, and other mostly-fixed landmarks
Real estate is not liquid enough that a 'correctly' priced house will sell immediately. You have to wait for a buyer that wants the exact featureset that house is offering.
Real estate has sort of a weird duality where it's viewed as a commodity in aggregate - but it is definitely not a commodity good to the individual buyers looking at houses.
In the Bay Area at least, you really cannot discount the Feng Shui attributes of a home [1] and I don't think there is any MLS type of database that takes this into account.
"It is difficult to get a man to understand something, when his salary depends upon his not understanding it!" -Upton Sinclair
Zillow has an accurate estimate somewhere in their database but it's not in their financial interest to make that estimate public. By low balling, they both gain more buyers and gain more sellers.
>By low balling, they both gain more buyers and gain more sellers.
Just looked up my house, and got a chuckle out of my wife when I told her the zestimate for our house was 61% higher than the price we paid two years ago. Needless to say, we're in not in a "hot" real estate market (i.e. city population ~60,000). The other shady thing it that Zillow says the sold price for the house we sold back then is $20,000 higher than actual sale price. I wonder what percentage of people think Zillow is underestimating vs. overestimating home prices.
When there's a string of bad reviews on Yelp, it's basically Yelp's way of telling the shop owner: "Gee, doesn't it suck that these spammers are bad mouthing your business without actually patronizing it? Well, if you work with us I'm sure we can weed out the spammers and only keep the honest (read: positive) reviews."
In this case, it's: "Gee, it's regrettable that our Zestimate for your house is so low. Only if we had a better source of data. Well, if you work with us I'm sure we can get a better (read: higher) estimate that's more in line with the true value of your house."
The buyers will flock to whichever market maker that advertises the lowest price, in this case Zillow. This in turn forces the seller to choose Zillow as well since that's where most of the buyers are.
Take a look at an actual appraisal document[0] for clues into what impacts a home's market value. The key is in selecting the 'Comps', homes that have sold recently and which are materially comparable to the home under appraisal.
Other random attributes that adjust the appraisal:
- Does the home have a view? Of what?
- What is the quality of construction?
- Garage or Carport?
- Pool?
- basement?
There are many variables, least of which is how active the market is today.
>And the list goes on... Is it that difficult to predict?
They do likely have a ML algorithm in the background. They even offer an Excel sheet where, for any county, you can see what % of sold homes are within, say, 20% or 10% of the Zestimate (it can be quite bad). And 20% is a huge error. Even 10% off is huge for most people.
What is hard to capture:
1. Condition of the house
2. Design of the house
3. Color of the house
4. Neighbors being annoying
5. Alignment of incentives for both party's real estate agents. Both profit more from a high price.
>You think estimating the value of every home in the country within 10% is bad?
It may be good from an "interesting problem" standpoint. But if your house's Zestimate at $300K, 10% means it's real value is somewhere between $273K and $334K.
Does anyone have a source of data for the variance of the home prices in the U.S.? Getting within 10% sounds spectacular, considering that I think you could get close to 10% price differences between different combinations of buyers and sellers on the exact same property. If someone were to come up with an algorithm for predicting the price of a steak dinner at a restaurant, using spotty 5 year old data, I think everyone would be impressed at getting it correct within $3 of $30.
>If someone were to come up with an algorithm for predicting the price of a steak dinner at a restaurant, using spotty 5 year old data, I think everyone would be impressed at getting it correct within $3 of $30.
As I said, as an academic exercise it's great. But we're not discussing whether Zillow deserves goodie points. The question is: Is this Zestimate useful for anyone?
From a buyer's or seller's POV, 10% is too high an error. Sure, if your house is only $100K, $10K off is not much. If your house is $300K, you're not likely going to get far by underbidding with $270K, and the seller will be deemed too optimistic for asking $330K (although with bidding wars, it occasionally does happen).
For either the buyer or the seller, being told the price could be between $270-330K is virtually useless in deciding what to buy/sell for. It's just too much variance.
With high numbers, absolute amounts matter more than relative percentage.
Note you have two problems:
1 - features that you can't observe (sentiment / how much someone likes the house or area);
2 - features that are too expensive to observe
Have a view of the ocean? In CA, that can be worth another $X00k or maybe even millions. Have ocean view rights, ie nobody can build between you and ocean? Even more. The neighboring property with a blocked view can be worth far less. That's going to be really hard to observe at scale.
Does the backyard, if any, get full sun? What is the soil like?
Have reasonable suspicions there's asbestos in the building, even if not formally acknowledged, and a $50k+ remediate would be required?
How much sunlight does it get, and from what side?
How old is the interior woodworking and how well has it been cared for?
How good is the internal soundproofing?
How much renovation is required? Do you still have knob & tube?
Even silly things like how good is the cell service. Which can vary just by moving 30 feet.
How good is the water pressure and how much $$ would it cost to upgrade it.
You could run a computer vision algorithm on the included photos to get a sense of the view. If a house has a good view they're going to include a picture of it.
I tried once. The challenges came down to too many variables relative to sample size.
You gave just a small subset of variables which can impact price by over 20%.
The most important when I looked seemed to be:
- Neighbourhood (even street in some cases)
- Number of Bedrooms
- Year of construction
- Year of last renovation
- Floor (apartment)
- View (especially for apartments)
- Number of Bathrooms
- Year of sale
That last one, year of sale, is important because it actually impacts your training data. Do you use five years of data even though local real estate prices have been increasing at 10-20%/year? Or can that be normalized even though some neighbourhoods are changing at different rates than others?
I restricted my research to 1BR apartments in the city of Vancouver. There are some unique challenges for real estate in Canada, but basically the best sample size I could get was about 2000 sales (I think this was all the 1BR sales in a given year). Very few had information on all of the variables I listed above. Suffice it to say, you can't train a model of any quality on that data.
I'm sure this varies by state and/or locale but one major thing missing from public data in my area is basement size, which can account for 25% or more of a house's size. Our city/county/state (not sure which one does it, really) only records above ground sqft, even if you have a walk out basement.
There are multiple startups working on real estate appraisals
The thing with RE is that its as imperfect as you can imagine. Its possible that the listing price varies from the real sales prices because its a part of the marketing campaign, which that alone puts a lot of noise into the market. (for example, putting a 100k house at 80k can get people in a bidding war ultimately irrationally or emotionally paying above 100k. Alternatively, a higher listing price might make the seller more adept and following a realtor.)
> I guess the problem is, there's not any money in having an accurate prediction.
Market fit is definitely one of the hardest parts of the problem.
Some companies have decided to actually buy the houses and sell them themselves!
The big thing Zillow is missing is buyer information. I will personally value a home differently than you will because we have varying desires.
For example, I put a low value on bedrooms past the second and multiple stories, but a high value on a large garage and big kitchen. You might value things completely differently.
What Zillow should do (or maybe I should) is gather information about the buyers and calculate a personalized Zestimate that reflects what the seller thinks the actual value of the home is. As a side-effect, this approach would probably be shielded from lawsuits because it's not an appraisal, but a personalized estimate of premium over market value.
Aside from the long list of unobservable attributes your ML algorithm can't learn, the biggest problem would be the small size of the dataset available, especially in the hottest markets (few sellers).
Many neighborhoods (not cities, or zip codes - too much variance in god that large) have on the order of single digits a dozen sales per year, meaning that in some cases the most recent 'comps' can be a month or two old. In a fast rising or dropping market, there's nowhere near enough data to train an algorithm before the prices have changed again.
Except appraisals include much more than what you listed there which typically needs to be seen individually on a case by case basis and isn't public data. For example looking at if a sink is leaking.
In my case I'm pretty sure a horrible yellow-brown shag carpet was the reason I paid less than list for my house (I removed it as soon as I got the key and was delighted to find hardwood floors underneath).
In most places, for most types of appraisals, an appraiser is unlikely to do more than glance at the sink, and primarily to determine the finish /trim level of the house, not determine whether or not it is leaking. (of course, they may catch an obvious leak and document it, but they aren't going to look too hard.)
In some areas where houses were developed around similar times to similar plans this works.
But, for example, here in Berkeley the price per square foot can vary wildly based on the condition of the homes. We have 100 year old homes that are barely habitable next to completely remodeled ones that can sell for a difference of $200 per square foot.
This is why each property is valued on a sq. foot cost. It naturally includes the "depreciation" of the structure and the land cost.
1. empty buildable lot (with zoning restrictions on max # buildable sq. feet)
2. tear down (add demo fees and maybe requirement to retain existing foundation to avoid extra taxes).
3. rehab
4. new construction
Which is normalized by location... rehab may cost more then new construction depending on the location and if it was a flip.
A $1m tear down can become a $2m+ house, or 4-5 $500k condos.
So many intangibles. People's emotions (on both sides) are one, but then you have the house with the nice kitchen, or that other house with a not-so-nice kitchen, but a flatter yard. And on and on.
Zillow uses plenty of ML, but what's their advertised accuracy? iirc it's something like within 20% 80% of the time?
Having been through a tax reassessment in our home town, I can say it is not that easy. We found there are many fine-grained location dependencies, and not that many sales, with the latter issue complicated by the after-effects of the 2008 debacle.
I think the difficulty is scaling the "art" of estimating prices. Any single home I could sit down and give a good estimate but to come up with a fully automated way to do that seems pretty tough still. I'm also not sure that the really hard parts are solved with ML as well.
Inspectors and appraisers do different jobs - appraisers don't look for rotting wood, sometimes they don't even enter the home (i.e. if they know the trim level used for a particular development)..
As long as there are already many anecdotes, I'll just add mine (I've always taken the zestimate as Zillow's working model that probably needs lots of tweaking; it's not like the MLS's are eager to give them data).
Anecdote: My folks live in a golf course neighborhood where their house is not on the course, and it's smaller than the rest. However, my folks house is valued based on these extravagant golf-course side properties that are actually quite a bit more $/sqft. Zillow needs to have a human go through manually and adjust for these subtle differences, IMO.
Despite dome shortcommings, Zillow is a godsend compared to the systems local MLS's use, just the listing dates/listing removed dates/ last sold for/ days on market dates are worth a lot. Realtors can give you this info, but it is not nearly as frictionless as it could be.
Zillow is still looking for that disruptive go-to-market strategy (they already have most of the tools). But, the Realtors aren't going to give up their ~$6K per transaction fees easily. They're much more connected than cab drivers.
This is an obsession of mine, but I'm not a professional. I dated/ lived with a Realtor for ~20mo, and I asked lots of questions about the workflow of the business.
I don't think Redfin has much of an edge in go-to-market strategy. Redfin's allows for more comprehensive data (MLS access because of membership), and Zillow's allows for scalability. Sales data sometimes show 3 months lag from time of transaction and entry into the county clerk's records which gives the MLS quite an edge itself.
I wonder how many other tech companies fit into this general category of "providing an equivalent to a regulated service without meeting the regulations". You've got the more obvious "sharing economy" type things which are in the news all over the place lately, but I bet there are a bunch of other companies like Zillow that have been flying under the radar for a while.
If a homeowner or homebuyer is paying someone to assess the value of a property, requiring expertise and training in inspection of a house such as materials, build quality, plumbing and whatnot, I could understand needing government approval.
This is just someone saying what they they think something is worth. Do I need a license before I can say "I think a certain car is only worth $25k instead of $30k"?
>If a homeowner or homebuyer is paying someone to assess the value of a property, requiring expertise and training in inspection of a house such as materials, build quality, plumbing and whatnot, I could understand needing government approval.
I've seen how appraisers appraise, and this comment brought out a laugh.
Their estimate is really no better than Zillow's.
They spend about 15-20 minutes looking at the property. They primarily measure dimensions and look at cosmetic stuff (no "inspection" whatsoever). Then they look at comparables and make an appraisal.
I asked one of them how they handle various items in their appraisal (solar panels, etc). The answer? "I have no idea what to do with those things, so I just ignore them."
The reality is many buyers will value stuff that the appraiser just plain ignores. This causes problems because the seller cannot increase their price due to those items - when it comes time for the buyer to get a loan, the appraisal will come out lower than the sale price and the bank won't approve.
I recall when I bought my house, the real estate agent had access to the appraiser.
Lots and lots of problems with the appraisal business. I don't know how much they make, but it's a good racket to get into that requires minimal work and pays well ($400-800 per appraisal - they spend more time driving to houses than examining them).
I think you're assuming way more thoroughness in the typical assessment of a property than is warranted by the practices. At least with respect to real estate taxes, in my experience assessments are pretty mechanical with respect to square footage and standard house features (number of bathrooms, etc.)
Assessments are different from inspections.
I don't really disagree with you though. There would seem to be a difference between someone being directly paid to assess something, perhaps largely based on standard formulas and practices and a web company providing housing estimates with no contractual relationship.
The actual Illinois law covers real estate appraisal, not other kinds (ie, cars). However, it defines an appraisal as the act of forming an opinion about the value of a piece of real estate. It then forbids doing it without a license. Not doing it for money, not communicating your opinion, but forming it. There are explicit exceptions for realtors, banks involved in real estate lending, and for private use (ie, appaising your own house).
So I would say Zestimates are clearly illegal according to the plain text of the law. However, I suspect this is one of those laws where a judge would find a convoluted reading to support his or her gut opinion that finding Zestimates illegal is ridiculous. I think you could also make a pretty reasonable claim that the freedom to form an opinion about the value of something is an unenumerated, but nonetheless inalienable right, and therefore the law is unconstitutional.
There are some rules around someone saying what a home is 'worth' when we were recently buying our house the realtor had a couple tools for estimating sale price but every email with them was extremely adamant that it was not an estimate of worth because that would require a full appraisal. Maybe she was being overly cautious but it was so constant I have to believe there were some very real restrictions she was working with.
So when you go to a company and ask for a bank loan secured against a car, or to insure your car against loss, you might want to have some common standards recognized by insurers or loan providers as to what constitutes a 'fair value'. You would be a bit upset if you filed an insurance claim on your car, and the underwriter came back and said 'I asked @lotsofpulp and they said your car is worth $25K instead of $30K'. You might question what credentials @lotsofpulp has to value cars.
Now, you might argue that there are market based or self-regulated solutions to that, and that's reasonable (fact is that a lot of the industry goes off 'de facto' standards for valuing cars, like KBB), but the fact is that sometimes a government regulation is necessary to prevent anticonsumer or anticompetitive practices, especially in the finance industry.
This isn't the same thing. This would be you putting up a web site that tracks cars and the ones for sale you'd say how much you think it's worth. Do it once, you're just a guy (or gal) saying something. Do it all the time and you become somewhat of an authority figure. That's the difference.
A home appraisal is not the same thing as a home inspection, the latter being much more thorough,looking for condition of home mechanics, and structure.
"...the fact that they offer market-value estimates and 'are promoted as a tool for potential buyers to use in assessing [the] market value of a given property,' shows that they meet the definition of an appraisal under state law. Not only should Zillow be licensed to perform appraisals before offering such estimates"
It can be surprising when seemingly innocuous activities are in fact heavily regulated (cf. Stanford needing to remove all their learning videos from their website).
I don't believe the article is correct in the definition of appraisals.
A valuation or an approximation of value by impartial, properly qualified persons; the process of determining the value of an asset or liability, which entails expert opinion rather than express commercial transactions.
When the value of your $100000, $200000, $300000, or more home can be greatly affected by a someone else's declared valuation of your home, its how you get regulations.
this should be. it's a market with few sales at high costs heavily interwinded with the mortgage market.
having a de facto monopoly on what's considered a fair value, rightly or wrongly, opens up bank people and real estate agents to all kind of manipulation.
by manipulating their equation they can help lowball the sales price of an entire area, they can both profit directly out of this, indirectly through stock manipulation and even cause trouble to the whole banking sector devaluing mortgadget properties
But how does getting an appraisal license protect bank people and real estate agents? Isn't an appraiser subject to the same moral shortcomings as any regular person?
And only allowing state licensed appraisers to give value estimates does exactly what you are trying to avoid. It creates a state owned monopoly on housing estimates.
Zillow would lose credibility and it would hurt the core of their business if it was found that they company was purposely deceiving people about the value of their homes so that they could directly profit from it. They might also open themselves up to legal repercussions in that case for just plain old fraud as well.
It's easy to defraud one of the parties in a real estate transaction (usually the bank; they have the money) if the other parties, including the appraiser, collude. The appraiser is supposed to be independent, and the license is a stamp of professional ethics.
If a licensed appraiser acts fraudulently, you have grounds to sue their hinders off.
People are way too tough on Zillow. The article states that the Zestimate is within 5% of the sale price 54% of the time and within 20% of the sale price 90% of the time. That is pretty damn good considering they do not send anyone to look at the outside of homes and they do not go inside homes like an appraiser does. The problem is that people think the value of their homes is equal to what they paid for it plus some percentage gain every year. In reality, the value is just whatever a buyer is willing to pay for it. You can think your house is worth a million dollars, but it is not worth a million dollars if no one will pay that much for it.
The problem is that people think the value of their homes is equal to what they paid for it plus some percentage gain every year.
How is this the problem when you just stated previously that the price is 20% off from reality for 10% of people.
The real problem is a shitty estimate becoming something that people treat as a real estimate. Zillow obviously state this isn't the intention, but if they act in a way that promotes it as an estimate contrary to what they state they are doing, well, people are going to get angry.
At least in the Bay Area, Zillow et al. are decent for getting a decently accurate ballpark figure in mind. A lot of sale prices are clearly put up below market to grab attention, knowing that the sale price is very likely to exceed it by a wide margin.
Here in many parts of the Bay, you'll get laughed at and promptly ignored if you try and use these "zestimates" to try and put downward pressure on the price. Sellers here have many offers to chose from, leading buyers to compete not just on price, but even dropping certain contingencies to push the deal through.
A friend was looking at houses via Redfin. The agent told him on day 1 that he should expect to spend 10%–20% more than the online quote, as a rule of thumb.
Since then I've taken "Zestimates" with ~65,000 to 100,000 grains of salt.
I used to work for Z, and I can confirm that the Zestimate is just that: an estimate. It doesn't know things like whether you did renovations or whether the house next door caught on fire and burned down and things like that. If something about your particular house changed in the last few years, the Zestimate is likely to be way off. That said, it's fairly good as far as comps go, and if you have an average house in the hood, the Zestimate is gonna be pretty close.
Regarding the rule of thumb you mentioned, my wife and I have been looking for a house in the Bay Area, and we've determined that the listing price is just a completely made up number. We made an offer on a house that was 100K over asking price, and the sellers countered with 40K on top of that (so 140K over asking). I was like... what? You listed for X, we offered X+100K, you don't get to counter that. lol
We obviously didn't get the house, and they wound up selling for 175K over asking price (+~25%). We've made several other offers, and sometimes the house goes for 50K over asking and sometimes it goes for 250K over asking. It just seems to be totally random. It's absurd. Instead, we actually rely a lot on the Zestimate and Redfin's estimate to figure out what the "real" value of the house is. Z and R tend to be roughly in the right ballpark for the Bay Area.
> I was like... what? You listed for X, we offered X+100K, you don't get to counter that
Apparently they do, because...
>they wound up selling for 175K over asking price
And, as for this:
>Instead, we actually rely a lot on the Zestimate and Redfin's estimate to figure out what the "real" value of the house is
Let me give you a lesson from econ 101 - the 'real' value of something is whatever someone else is willing to pay for it. If multiple people are willing to pay more than Zillow or Redfin estimate, then the 'real' value is higher.
I'd also like to know how or why Zillow lists houses that are "pre-foreclosure".
I know it's semi-public record, but that certainly doesn't help the homeowner try and recover maximum value for their property if they decide to sell.
On top of that, it's pretty unnerving when you see this pop up for one of your neighbors and good friends. Nobody should have to go through even more humiliation than they already are.
If you ever hover over the URLs, Zillow is using a 3rd-party service for the forclosure potential properties. It's lead gen, and you have to become a customer to get actionable data.
Interesting. I wonder what would her argument be if Zillow simply reported the City/County tax assessors estimates which are usually even lower than Zillow's.
In Illinois (the state from the article) the assessor is calculating the market value:
Most real property in Illinois must be assessed based on its value on the open market, or its “market value.” This value is the amount at which a property would sell in a competitive and open market... [1]
That's not how it worked where I grew up in Upstate New York. The assessed value was a number way below market value, but assessed values among different houses pretty much differ by the same proportions as the actual market values. I do not know why it is the case.
The realtors I talk to hate the Zillow estimate as it typically creates an unrealistic expectation for either the seller or buyer depending on if the algorithm over or under estimates.
Apparently it's rarely correct, though I live in a unique area.
If it's not correct, then the sale price will reflect that and Zillow's reputation will take a hit.
Don't understand why the government needs to get involved in what someone says something is worth, it's their opinion and they're entitled to it.
Edit: It would be funny if Zillow also started putting up alerts about problems with government finances which will likely impact future taxes, such as in Illinois.
>If it's not correct, then the sale price will reflect that and Zillow's reputation will take a hit.
Perhaps with that buyer and seller. I believe, though, that Zillow rewrites historical estimates when they get new data. So you can't really see how wrong they were after a sale.
If an engineer at Zillow manipulated the Zestimate on a home they were trying to buy or sell, would they succeed in driving the price of the home in the desired direction? I think so.
Clearly real estate agents have the same feeling or they wouldn't be suing.
The problem is, you'll never get to establish a sale price if the seller thinks his house is worth 100k over market or a buyer thinks the seller is asking 100k over market because that's what they saw on Zillow.
Further, when zillow gets the estimate of a house wrong it costs them nothing, but there are real costs to Zillow's bad estimates to buyers and sellers.
If Zillow gets the estimate wrong, then people will stop using their estimates, costing them in reputation and perhaps dollars down the line.
There are so many sources of info for house value that I don't see how Zillow could single handedly be manipulating a house's sale price. If it's not selling for what the homeowner thinks it's worth, it's probably not worth that much.
A "civil matter" is where the courts—a part of government—apply law—created by government—to non-crimes disputes, often (as in this case) between private parties.
It absolutely does not indicate the absence of government involvement, even if the government isn't a party to the litigation.
The government's not getting involved, oh you mean the fact that a judge is going to be involved?
Civil courts exist to provide monetary damages if damages can be proved to exist. Yes, they can also compel arbitrary behavior as a part of governing body.
This is independent of any stance on Zillow and a state licensing regime. If the plaintiff can convince that the tool is making her lose money, the judge will award her money.
"but that means < cue completely uninformed legal analogy >" yes you can make a career out of litigation, many people have.
It cant just be the plaintiff proving that Zillow's tool makes her lose money, can it? If that argument was true then you could get sued and lose for reporting on facts that put the company in a bad light, like poor hygiene in a restaurant, or for competing with another company with a better product.
many summary judgements on flaky arguments are made just because the defendant didn't respond. the judgements are collected from you bank in any jurisdiction that will reciprocate
You're right, I shouldn't have phrased it that way. I meant to say I don't think the government should be involved.
Although, I'm not sure why the plaintiff should be awarded anything, even if Zillow causes her to "lose money". Unless Zillow is doing something specifically to her home estimate, then it's merely offering a differing opinion on what they think the home is worth.
"Eyyy buddy, you wanna buy this car? Classic 1991 Honda Civic, only 100k miles on it, and most of the doors still open. I'll give you a deal, $10 000."
"Yeah, I don't think it's worth that much. A hundred dollars, tops."
"Did you just appraise this car? Oh, you done messed up."
muffled sound of approaching police sirens
EDIT: Hah, didn't see you'd used the exact same analogy elsewhere.
Nonsense. For the judge to award your money to a complainant, they need to have a loss and you need to be legally liable for the loss. The fact that they have a loss is not itself a cause of action. The fact that you caused the loss is still not a cause of action.
> Civil courts exist to provide monetary damages if damages can be proved to exist.
Civil courts exist to provide remedy for non-criminal violation of the law; award of damages (actual, statutory, punitive, or otherwise) is one form of civil remedy, but not the only one available.
It's somewhere between comical and meaningless in San Francisco. The operation of the algorithm is painfully apparent and AFAI most heavily weights trailing sale price per square foot and sale price over list, as compared to presumed 'comps,' all three of which factors are frankly piss-poor indicators in this market and city.
As a proxy at a glance for such comps it's at least an indicator of when (as is common) list price is far below expected sale price, and an arguably cruel fiction designed to attract interest (in my tier: from the desperate).
(If you don't live in SF, sales at 30-50% (100s of K) over asking are common again, after a brief cool down at the beginning of this year.)
The estimate for my house dropped by 25% in one month when the market was still rising. I'm assuming that the cause was the house next door which was in poor condition and was sold for about 20% below market value. The buyer put $100k into the house and is now selling it above market value ($200k more). It will be interesting to see what the zestimate for my house will be when the other house sells again. (It will probably sell within a month.)
I have had both our homes (we rent one out) independently estimated and both are worth 25-30% more than the Zestimate. I live in IL like the realtor in the article. Zillow is consistently lower than other sites. The most accurate estimates for our home are on Redfin, which is almost perfectly inline with the appraiser's estimates. Zillow has a pretty overfit model IMO.
This is the flaw here. The seller, real estate agent, banker, and appraiser all have incentives to raise the price of the home. The only person in the transaction who doesn't is the buyer. I've been working on something in this area as a side project, and have been working with the fallout of 2008 in my day job, since 2008... and it continues to this day and the lack of understanding of the incentives here is the main issue. If Zillow wanted to gain customers through their zestimates they would raise the estimate not lower it. After all, more people own homes then are buying one in any given moment. Telling those people how smart they are that they timed the market is a much better way to retain users.
I have different problems. Zillow's estimate is about 15% higher than Redfin for us. But there's also issues with the facts each have about our home. Zillow's estimate also goes up and down by tens of thousands of dollars per week. Some weeks it'll be as much as a $40,000 swing.
Keeping the data correct on the sites is also a chore, and if I don't keep on top of both sites, entire rooms or bathrooms will come and go, square footage will change and so on.
Real estate agents will claim that appraisers don't go off of these sites, but buyers will use them as pricing information so it almost becomes critical for home owners to police these sites to get as favorable an estimate as possible.
At's an algorithm based on available data... it may not be accurate, but that doesn't mean it's too wrong. Also, if you put in expensive features, doesn't mean your house will sell for twice as much in a given neighborhood. Pricing is fairly collective with main factors being property size, home size, bedrooms/bathrooms and some influence on features, namely pool, garage(s) etc.
That doesn't mean it's entirely wrong, unless there are manual tweaks happening to shape specific markets, which may be proven/disproven through discovery. Aside from that, there's the legal aspects in different states and counties in terms of what a licensed/unlicensed seller is allowed to do.
I would say that if the market is more than 20% off in terms of actual sales from the Zestimate, then there may be something up... The Phoenix market seems pretty accurate to within 5-10%, which is pretty reasonable.
Thing is, all appraisers do a bad job. Evaluating a house is completely up to their opinion, and the artificially increase prices just because houses are selling. They drive prices up and down based no their will. Couldn't their job be legitimately done by a computer and a camera? Probably if the technology was there.
This is as ridiculous as suing Nolo for publishing law books without a license.
The only reason why the government needs to have an appraisal license at all is because the government handles the loan for the house. The goal is to mitigate the risk to the government, not to help consumers decide how much they want to pay.
Sounds like an opportunity for arbitrage that would correct the problem. If "house flippers" start using these low prices to make a quick buck, either Zillow or homeowners are surely going to cotton on and things will settle back down to fairness.
But what could a homeowner do about it? If Zillow makes your house look a lot less than what it's worth, so nobody wants to buy it for the proper price, you can't do anything. That's exactly what this lawsuit is about.
They're not even suing for money right now, they just want the price for their house fixed so they have a decent chance at selling it.
You can find estimate calculators from various banks, realtors, etc. I just did a basic google search and checked out estimates from Zillow, Chase, and RE/MAX, and they range from 4% under my home price to 11% over. That's a pretty crazy range of guesses.
My question is: why aren't these other home estimate calculators subject to being sued, then?
(And when I checked out Zillow, I noticed there was a direct link to "I disagree" next to the Zestimate. None of the other calculators had any kind of feedback mechanism.)
Seems like the journalist here forgot to mention that home estimates are pretty common and come from many different sources.
Here in Australia, underestimating by realty companies is the norm. I watched a friend go through buying a house last year, and some houses were advertised at prices that were below reserve [1]. As in, at auction, the bidding reached the advertised price, but the buyer's floor price hadn't been reached yet, so the house was not 'on the market' yet...
[1] In one case, 80k below reserve. You know you've hit the reserve price, because in my state, the auctioneer is legally required to annouce the house is 'now on the market'
Starting at the reserve can spook some bidders and make it hard to start the auction. Starting below the reserve usually gets some momentum going in the bidding, and that tends to carry over to a higher final price.
Advertising a price that's below the reserve is pretty unethical, though it's not illegal (the seller is never locked into a price). People end up wasting time and money (getting reports) for houses that were never actually in their price range in the first place.
I've mentioned it before on HN, but Zillow is vastly overvaluing my home. I paid $43k for it as a foreclosure in 2009. Today it appraises for about $60k (it's a tiny farmhouse on a half acre lot), yet Zillow constantly values it at over $80k. I believe they do this so they can bring up the average price for the houses in my area. Maybe it's good for me, I don't know, but it's dishonest.
Our house is in the hottest part of Portland, OR, with double digit % increases in the last couple years.
Last October when I refinanced, it was appraised at $455k. Zillow had it at $467k-ish.
Not enough to bum me out because I also had a now former neighbor sell recently. Buyers got into a bidding war and he walked off with $15k over asking.
Surely appraisals regulation is just another instance of mortgage regulations. But is it not the over regulated mortgage market that blasted in 2008? Was it because of zestimates? Nope? Well here ya go. Let the people speak their opinion and work on their ML.
I would think that Zillow would mainly be punishing themselves by not having a quality algo for home values, given that it the main draw of the site for many people.
If an algo could predict real estate prices (all of which are unique and based on instantaneous human preferences) with 100% accuracy, any arbitrary human thought would be known.
This month it was $544k, and there's no sign it ever got above $580k in the historic graph. It claims it was $546k on March 31st.
One thing I did notice was a new button underneath the estimate: "I disagree with my Zestimate" which, surprisingly enough, is a lead gen for realtors, not a method to try and get them to change.