- Find a Real Estate Professional
- Alabama Realtors®
- Alabama Mortgage Lenders
- Alabama Home Improvement Pros
- Alabama Other Real Estate Services
- Review an Agent, Lender or Pro
- Marketing on Zillow
- Real Estate Agent Advertising
- Join the Professional Directory
- More
Answers (7)
Best Answer
"I just want to know what factors are considered and if there is a way to update those factors under consideration." -
1) Last sold price adjusted for sold date
2) tax assessed value adjusted for local property tax rules
3) Sqft
4) lot size
5) # bedrooms
6) # bathrooms
7) # stories (maybe, it doesn't seem to be in the database for most)
8) year built
9) covered parking.
They do not use ANY "comparables" in the estimates at all but create the comparable list from the estimates instead
They model the county records to the "non-distressed recently sold" in the "vicinity", larger radius for rural, smaller distances for dense urban. They then test the model against the recently sold, throwing out those that don't fit well, and then recalculate the model prior to using the model for extrapolation of an estimate. City boundaries are ignored for the modeling. The models do not use sold data in a different county.
It is not a linear model, so one should not expect it to behave in any linear fashion.
You can change many of the data items, but the owner/agent provided data is phased in over about 6 to 8 weeks, so you will not see a step-change with the change of one of the data items.
If you need to correct the last sold information, or tax assessed value, you will need to flag it, or use the "report a problem" button as only Zillow can correct those items. The last sold price has a major impact on the estimates.
Multiple models have to be run for each county area, as if there is data missing, any assumption on that data would create problems, so they model the non-distressed sold units with various data items missing for the extrapolation of estimates where data is missing.
1) Last sold price adjusted for sold date
2) tax assessed value adjusted for local property tax rules
3) Sqft
4) lot size
5) # bedrooms
6) # bathrooms
7) # stories (maybe, it doesn't seem to be in the database for most)
8) year built
9) covered parking.
They do not use ANY "comparables" in the estimates at all but create the comparable list from the estimates instead
They model the county records to the "non-distressed recently sold" in the "vicinity", larger radius for rural, smaller distances for dense urban. They then test the model against the recently sold, throwing out those that don't fit well, and then recalculate the model prior to using the model for extrapolation of an estimate. City boundaries are ignored for the modeling. The models do not use sold data in a different county.
It is not a linear model, so one should not expect it to behave in any linear fashion.
You can change many of the data items, but the owner/agent provided data is phased in over about 6 to 8 weeks, so you will not see a step-change with the change of one of the data items.
If you need to correct the last sold information, or tax assessed value, you will need to flag it, or use the "report a problem" button as only Zillow can correct those items. The last sold price has a major impact on the estimates.
Multiple models have to be run for each county area, as if there is data missing, any assumption on that data would create problems, so they model the non-distressed sold units with various data items missing for the extrapolation of estimates where data is missing.
"but by doing a multivariate analysis of some sort based on recent sales that aims to predict what a house with a certain combination of features would sell for?" -
Correct; but it is not really a "prediction", but more of a "statistical analysis". The tolerance margin is too wide to be used for predictions.
"If a big factor is prior sale price then a house that was bought at the bottom of the last cycle will have a lower estimate just because the owners kept it a long time?" -
Not quite right; it can be higher or lower; mostly depending on "condition" compared to "average", and "condition when purchased", and "condition now as compared to when purchased".
Although the "terms of use" state we won't "reverse engineer" their methods; and the method of "adjusting for inflation" the prior sold prices is "proprietary"; it is of much interest to me. There are several different options they may be using to adjust prior sold prices. One may be the U.S. consumer price index for housing. It could be on a "regional" basis, as the data is available at several regional levels. Another may be Case-Shiller housing indexes, though they don't cover all areas of the U.S. Another possibility is Zillow's Housing index by County. But Zillow doesn't offer that in non-disclosure areas, and this seems to be a bit circular as the indexes are created by taking the median of the estimates in the area. Zillow's indexes have an exceptionally high correlation to Case-Shiller's indexes when they both offer indexes for the same region. But Case-Shiller's is a lot more susceptible to changes in foreclosure saturation, thus impacted by two dynamics instead of one. The indexes always lag behind the market by several months, so they may not be that useful for recent sales, but they are useful for bringing old prices forward. The only trouble is that Zillow's indexes only go back about 14 years, which is not helpful for older sold data. Case Shiller goes back about 100 years. And the CPI goes back sufficiently also.
Your guess is as good as mine at which inflation adjustment factors they are using to adjust sold prices for the time the properties were sold.
Correct; but it is not really a "prediction", but more of a "statistical analysis". The tolerance margin is too wide to be used for predictions.
"If a big factor is prior sale price then a house that was bought at the bottom of the last cycle will have a lower estimate just because the owners kept it a long time?" -
Not quite right; it can be higher or lower; mostly depending on "condition" compared to "average", and "condition when purchased", and "condition now as compared to when purchased".
Although the "terms of use" state we won't "reverse engineer" their methods; and the method of "adjusting for inflation" the prior sold prices is "proprietary"; it is of much interest to me. There are several different options they may be using to adjust prior sold prices. One may be the U.S. consumer price index for housing. It could be on a "regional" basis, as the data is available at several regional levels. Another may be Case-Shiller housing indexes, though they don't cover all areas of the U.S. Another possibility is Zillow's Housing index by County. But Zillow doesn't offer that in non-disclosure areas, and this seems to be a bit circular as the indexes are created by taking the median of the estimates in the area. Zillow's indexes have an exceptionally high correlation to Case-Shiller's indexes when they both offer indexes for the same region. But Case-Shiller's is a lot more susceptible to changes in foreclosure saturation, thus impacted by two dynamics instead of one. The indexes always lag behind the market by several months, so they may not be that useful for recent sales, but they are useful for bringing old prices forward. The only trouble is that Zillow's indexes only go back about 14 years, which is not helpful for older sold data. Case Shiller goes back about 100 years. And the CPI goes back sufficiently also.
Your guess is as good as mine at which inflation adjustment factors they are using to adjust sold prices for the time the properties were sold.
- commins
- Contributions:4
Thank you to everyone for taking the time to answer. If I understand Asian Orange's explanation it makes sense. I don't know how "comparables" are used by other sources but Asian Orange am I correct in saying Zillow does it not by "comps" but by doing a multivariate analysis of some sort based on recent sales that aims to predict what a house with a certain combination of features would sell for? If so I can understand how depending on how you weight certain factors and how they combine you might get a sudden change not shared by nearby homes. If a big factor is prior sale price then a house that was bought at the bottom of the last cycle will have a lower estimate just because the owners kept it a long time?
Is this roughly correct?
Is this roughly correct?
- Allison Fishwick, "Allison Fishwick"
- Contributions:92
Zillow's Zestimates are not based on the interior of your home, the condition or the updates. They are based off limited facts of comparable homes sold in your area and public record. Here is a link to help understand more about Zillow's accuracy:
http://www.zillow.com/howto/DataCoverageZestimateAccuracy.htm
http://www.zillow.com/howto/DataCoverageZestimateAccuracy.htm
- commins
- Contributions:4
If the estimates depend on comps then one house value is independent on the others by definition. If you cannot explain a $75,000 drop in value in a 30 day period, then what validity do any of your figures have? There has to be something more than throwing darts for these calculations to have any meaning. I just want to know what factors are considered and if there is a way to update those factors under consideration.
- Sharon Lewis, "Sharon Lewis"
- Contributions:3923
If you really want to know what your home is worth in todays market, hire an appraiser or contact a local realtor and ask them to do a comparative market analysis on your home, or a quick cma on it. Zillow, as fun as it is, cannot go into each and every home to see your updates so that is why they call it a 'zestimate' like guesstimate . I hope that helps.
Because they each fluctuate independently and the calculations of one have almost nothing to do with a neighbor's that you might consider identical.
The tolerance margin is too wide to be paying any attention to short term changes in the estimate anyway.
The tolerance margin is too wide to be paying any attention to short term changes in the estimate anyway.
why did the zestimate on my house recently plunge relative to the homes around it?
Stating a discriminatory preference in an advertisement for housing is illegal. If you think this content is discriminatory or otherwise inappropriate and feel it should be removed from Zillow, please let us know by completing the information above.
We will review this content. Thanks for helping make the site more useful to everyone. To learn more, read Zillow's Good Neighbor Policy.