Well....if nobody wants to play Conkers with m...
Image by law_keven via Flickr

I know… most sellers at some point take a look at the ‘Zestimate’ that Zillow produces for their property.  And there is almost always one of two reactions…

  • Those people are on crack, my house is worth way more than that” (if the Zestimate is lower than expected)
  • Wow, I didn’t know it was worth THAT much… Cool” (if the Zestimate comes in higher than expected)

The bottom line is that if it is low, it must be wrong, and if it is high, then it must be right.  It is pretty rare for someone to see a low valuation and think it could be right, or see a high valuation and think it could be wrong…

But let me tell you a truism… the Zestimate is wrong.  Blanket statement.  It could be high.  It could be low.  It isn’t dead on.

Here are the current levels of accuracy for the Atlanta Statistical Area…

Within 5% of the true value, 25% of the time.
Within 10% of the true value, 45% of the time.
Within 20% of the true value, 66% of the time.
Median error, 11.6%.

This is lifted straight from Zillow’s accuracy.  For the Atlanta MSA, you are within 5% once out of 4 times… So, are you telling people that the just gave them a $400,000 valuation, and they have a 25% chance of the house being worth $380k – $420k?  And a 10% chance of it being worth either $$360k – $380k or $420k – $440k?  Or that there is a 10.5% chance of it being worth either $320k – $360k or $440k – $480k?  Or a 34% chance that it is worth less than $320k or more than $480k?

Maybe you will let them know that about half of the time your $400,000 valuation is worth $446,400 or $353,600… since the median error is 11.6%

The thing is that the Zestimate could rise or fall by a significant percentage and still be within the range of error..

So, the thing is that 75% of the time, Zestimates are off by more than 5%.

Now, I am not going to argue with Zillow’s national or regional data.  It is likely to be much closer than the data for a specific property.  The folks there argue (and I can’t refute them in any way) that Zestimates are as likely to be high as low and therefore cancel each other out and that for larger samples, they are quite accurate.

in fact, everyone that gives you an estimate of market value is doing exactly that… giving you an estimate of market value.  Market value can’t be determined until there is a buyer…  That was the previous post…

Reblog this post [with Zemanta]