# How do you search by comparing highest Zestimate to the listed sales price?

September 07 2011 - Groveport
• Be a Good Neighbor. Be respectful and on-topic. No spam or self-promotion! See our Good Neighbor Policy.

And again, don't get caught up to much in what a machine estimate tells you that something "might" be worth....

The modeling is done from the "recently sold" 3 times a week.  And after you buy, it will be a "recently sold", so it will become part of the modeling, so if you bought at list price with a much higher Zestimate, shortly after purchase the Zestimate will be almost equal to the purchase price.

This has upset a lot of people that don't understand the modeling.

There is no "buy a high Zestimate" in order to flip it by convincing others it is worth the Zestimate.  It just doesn't work that way.  Nor can you tell the motivation of a seller that way.

It is a good way to do a quick check for recently renovated or fixers, but that is about it.
September 08 2011
at the better fit --> *as* the better fit
September 08 2011
One also needs to remember, that the estimates are a statistical modeling method, thus subject to probability theory.

For my city, here is the distribution of % estimate deviance from list price:

Web address for full size image:
http://photos3.zillow.com/is/image/i0/i9/i306/IS5khsy3s0hear.jpg

For the pink normal distribution curve, the Average, Median and Mode functions were used to approximate the Mean; In this case, the Median was chosen at the better fit.  And the Standard Deviation function was used to determine and approximate Standard Deviation.  In this case, it was divided by 1.6 as it resulted in a better visual fit as well as provided a better R^2 correlation of determination.  So the Mean= -0.8% and the Std Dev = 10.6% with a resulting R^2 = 76.03%

For the cyan normal distribution curve, the Magnitude, Mean, and Standard Deviation were calculated with Excel's iterative solver using a "least squares curve fit" method.  (As the error function is non linear, it cannot be solved for directly with Linear Regression, which would be a closed form least square curve fit solution).  The Mean = -2.01% and the Std Dev = 9.97% with a resulting R^2 = 77.18%.

All this to say that either the list price or the estimate may be statistically off just due to probability, and thus will tell you nothing about the quality of the property or any "deal" one may be getting.

Still, you can see that their are sharp peaks outside the curve that would indicate either the "fixers" or the "recently renovated"...  You need to manually filter for actual conditions.

And remember, there is no such thing as "free equity".  The amount something is worth is only the amount paid, no matter what estimating techniques were used.

September 08 2011
Obviously, I was interested in "all" the present listings for the chart.  You can easily filter for "in your price range".

I always set a minimum price (usually \$10k) to try to get rid of trash advertisements and mistakes such as mislabeled rentals.  I also try to avoid accidental empty lots or properties that have no info by setting a minimum of 1 bed, 1 bath, and 300 sqft.

On the home search results, you get several pages; and most indicate a list price and an estimate.  I just blocked and copied the data into Excel; and had Excel divide, and then Excel do the sort.  I also turn off the images in the browser before doing the search to avoid bringing the pictures into Excel.  Even then, there are "images" that need to be removed; and though they can be deleted manually, it is easier to copy the column to a new worksheet without copying the images, and then deleting the original worksheet.

The address and link for each property stay with the data.  But it doesn't come into Excel as columns, all the data is in one column; and the number of items per record is not consistent, so it takes a bit of creativity to get the data in the proper columns and to delete the excess rows.  Of course, one could copy the HTML into a text editor and use macros to edit out the unnecessary stuff, but this is a bit time consuming too.

It was much easier with the "list" mode search results; but Zillow discontinued that as less than 2% of the people were using it, and of those that were, only a very small percentage were actually opening home details pages, so it not only was more maintenance work for the programmers, but it appeared it was primarily being used for people to "steal" the data, which is against Zilllow's agreements with their data providers.
September 08 2011
Pasadenan..  This is good stuff but where do you get it, and is there anyway to actually identify the addresses of those top green dots in my price range?

Thanks for passing on the knowledge. Best.
September 08 2011
There is no direct method to do that; but there are some indirect methods to accomplish that.

But what you will find is that most of those are "fixers".

Here is the present comparison chart for the listings in my city:

http://photos1.zillow.com/is/image/i0/i8/i9980/IS1sed4hq4y5vvn.jpg

Enlarged; for the ≤\$1 million listings, ±40% difference range:

http://photos2.zillow.com/is/image/i0/i8/i9980/IS5obw0qw5m3z7.jpg
September 07 2011

##### How do you search by comparing highest Zestimate to the listed sales price?
6
September 08 2011
3
June 14 2011
Guide
Guide
Guide
Guide
Guide
##### Be A Good Neighbor

Zillow Advice depends on each member to keep it a safe, fun, and positive place. If you see abuse, flag it. More on our Good Neighbor Policy.

##### Homes for Sale
1. 125 Shaffer Dr, Groveport, OH 43125

For Sale: \$75,000

• Beds: 3
• Sqft: 1243
• Baths: 1.0
• Lot: 7405
2. 6184 Dietz Dr # 605, Canal Winchester, OH 43110

For Sale: \$279,990

• Beds: 3
• Sqft: 1955
• Baths: 2.0
• Lot: 13024
3. 2553 Bridlewood Ct, Obetz, OH 43207

For Sale: \$104,000

• Beds: 4
• Sqft: 1746
• Baths: 2.5
• Lot: 10497