Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
So many people question how good is the Zillow zestimate valuation and want to understand how the computer algorithm works. I found this helpful description on the Zillow website which you may find useful
We compute this figure by taking zillions of data points - much of this data is public - and entering them into a formula. This formula is built using what our statisticians call "a proprietary algorithm" - big words for "secret formula." Currently, we have data, Zestimates and Rent Zestimates on approximately 100 million homes on Zillow.
What's in this formula?
One eye of newt ... Actually, it's less wizardry than mathematics. When our statisticians developed the model to determine home values, they explored how homes in certain areas were similar (i.e., number of bedrooms and baths, and a myriad of other details) and then looked at the relationships between actual sale prices and those home details. These relationships form a pattern, and they used that pattern to develop a model to estimate a market value for a home.
Who calculates the Zestimate and how do they do it?
We don't have homing pigeons flying from land parcel to land parcel to come up with the Zestimate for a house, but we DO have statisticians who work all day - and some nights - with a huge amount of data. They live and breathe valuation models and tweak algorithms to get us closer to actual market value.
I suppose if you are looking for a VERY broad price level, all of this wizardry is somewhat helpful.
However, if you need to know a value, Zillow (or any of the other 'modeling' services for that matter) just don't have boots on the ground to really make even a reasonable guestimate on what any particular property is worth. There is just no way that a model can incoporate the exogenous factors which truly determine any specific home's individual value.
I am inclined to think that things like Zillow are more dangerous than helpful. Especially when you look at the prices which they throw up. Usually they are quite wide of the mark.
Zillions of data points? No, I don't think so. Even the DOD does not multi-variant nuclear simulations running with that many degrees of freedom. The largest computational arrays in the world are clear
NOT used by Zillow...
Fact is the inputs they have access to are severly limited -- the raw numbers of bedroom counts and such are woefully inadequate to explain why one home might be worth less than half as much as another that has same room count...
It is really a foolish quest that the Zillow "statisticians" are on if you even believe the marketing hype about how they allegedly tweak their models -- S&P bought the C-S index to try and create a tradable hedge for property values. It has been a complete flop. There are too many "non linear factors" involved in pricing residential real estate. Correlatstion between income and employment inna broad sense are most useful, but getting down to town or neighborhood levels the sales of "model home" vs "value in the land" transaction overwhelm any ability for computerized trends to be useful... I suspect even the Schiller and Case themselves know this and that is big reason why they have moved on to other "research". If they are smart they never lookmback at how much they fleeced S&P...
What a coincidence. I never heard of Zillow until last Saturday while visiting friends whose house is on the market. Their house is severely under-zestimated. Mine is over-zestimated by around 20%, as is my whole neighborhood based on comps.
Also, the descriptions were wrong because of remodeling.
I ended up thinking Zillow is interesting in a way, but isn't a realistic tool and can hurt buyers and sellers who take zestimates seriously.
As mentioned, Zillow is decent for getting a really high level idea of what housing costs. Most of the time, you could look at the average price of a neighborhood and compare to another neighborhood and get a rough idea of which has higher values, but thats about it.
On the house by house level, I have seen too many screw ups with houses I have owned or been familiar with for it to be close to accurate. In fact, it took them 3 years to even get my current house located within 2 miles of where it actually is. If they cant get that right, how can anything else be right?
I read somewhere that they store the zillions of data points on 3Tb. I recently purchased 1Tb from Amazon for $70 so it show's what a nonsense the Zillow claims are, and also explains why so many comments from posters about the inaccuracy of the valuations they circulate.
Quote:
Originally Posted by Surbiton
So many people question how good is the Zillow zestimate valuation and want to understand how the computer algorithm works. I found this helpful description on the Zillow website which you may find useful
We compute this figure by taking zillions of data points - much of this data is public - and entering them into a formula. This formula is built using what our statisticians call "a proprietary algorithm" - big words for "secret formula." Currently, we have data, Zestimates and Rent Zestimates on approximately 100 million homes on Zillow.
What's in this formula?
One eye of newt ... Actually, it's less wizardry than mathematics. When our statisticians developed the model to determine home values, they explored how homes in certain areas were similar (i.e., number of bedrooms and baths, and a myriad of other details) and then looked at the relationships between actual sale prices and those home details. These relationships form a pattern, and they used that pattern to develop a model to estimate a market value for a home.
Who calculates the Zestimate and how do they do it?
We don't have homing pigeons flying from land parcel to land parcel to come up with the Zestimate for a house, but we DO have statisticians who work all day - and some nights - with a huge amount of data. They live and breathe valuation models and tweak algorithms to get us closer to actual market value.
It lists an investment property of mine in Baltimore at ~230k. The neighbor's home is zestimated at ~205k. Both homes are identical and attached to one another. We paid 87k for ours and they paid 300k during the boom. Where they get these numbers from I have no idea.
I am sure the models are cool and may actually work under good circumstances, the problem is the data going in is too erratic to get a goos number out. Many times they use assessor data, which is subject to how the assessor assesses...sometimes off of a faked up, indexed, or old number.
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.
Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
