« Now You Can Buy “Ticket” Insurance | Main | Mutual Fund Expenses Do Matter »
A Little Zillow Talk
By JLP | February 14, 2007
It’s funny how everything seems to happen in groups.
My most recent issue of Fortune has a cover story about Zillow. It’s an interesting article.
Then, today’s Wall Street Journal had an article titled How Good Are Zillow’s Estimates? (free) Here’s what they found:
The Journal looked at transaction prices recorded for 1,000 recent home sales in seven states, using data from First American Real Estate Solutions, a data provider in Santa Ana, Calif., and compared those prices with Zillow estimates, which didn’t yet reflect the sales. The median difference between the Zillow estimate and the actual price was 7.8%. (That was close to the 7.2% median “margin of error” reported by Zillow itself on all transactions involving homes whose value it has estimated.)
The estimates were about equally split between ones that were too high and those below the mark.
Zillow came within 5% of the price in a third of the transactions studied by The Journal. It was more than 25% off target on 11% of them. In 34 of the 1,000 transactions, Zillow was off by more than 50%.
I think this just proves that there is some art to appraising real estate. I looked up my house the other day and found that they simply used the same value as the county uses for tax purposes. I guess that value is better than nothing, but not nearly as good as market value. After all, I’m interested in knowing how much I can sell my house for.
What’s your opinion of Zillow?
Topics: Housing Market | 11 Comments »
February 14th, 2007 at 5:35 pm
How does a 7.8% Median compare to most real estate agents across the country? I’d say it’s fairly resonable…
February 14th, 2007 at 5:48 pm
It’s a good starting off point. I have also criticized zillow’s accuracy, but in reality, how much would we know without them? Obviously, upgrades, and a fast moving market are not easily calculated by the zillow mega computers.
February 14th, 2007 at 6:00 pm
Median? That’s a useless measure. Give us the mean, or better yet, a real statistic, like standard deviation.
February 14th, 2007 at 6:20 pm
I claimed my home on Zillow a few days ago and found that it was as high as $570k since I bought it (didn’t check exactly when). I can assure you that right now our home would sell for only about $430k max. In the summer, since we have a nice pool and yard, we could sell it for about 450k.
Right now, Zillow is saying it’s worth $500k. I put a “make me move” price of $575k. Heck, if I make almost $200k profit on our home, they can have it!
February 14th, 2007 at 6:28 pm
Kurt –
Hi, it’s David from Zillow. A median is actually the correct statistical way to interpret trends in a non-homogeneous data set like house values. The median ensures that outliers (like a few sales of very pricey houses) don’t skew the metric (as would happen if you used the mean). I agree that standard deviation could be an interesting additional statistic — but note that that would tell you how similar homes are in the area, and not what’s happening with their actual values — which is where our focus is.
February 14th, 2007 at 6:35 pm
JLP –
Just noticed your comment about tax values. We show those in areas where we don’t have enough data to calculate Zestimates. You should have noticed text that said: “Zestimate: None Available”.
We’re not suggesting that the tax values reflect market value (they don’t) — but merely provide the data as a means of comparing the relative value from one house to the next.
February 14th, 2007 at 7:34 pm
It’s nice to see somebody from Zillow out there on the blogs!
I gasped when I saw the $1.3 million estimate for the 50 year old house I’m renting in the San Francisco area…I hope it would not sell for that much! Maybe it would, I don’t really know.
February 15th, 2007 at 8:17 am
Still don’t trust it. I posted about this last week:
http://retiringearly.blogspot.com/2007/02/i-still-dont-trust-zillow.html
February 15th, 2007 at 4:49 pm
Dave,
Not to get into a huge discussion here: the median seems like the best thing to use when optimizing your data algorithms, but not the right figure when people are looking up the projected value of something. If I tell you I have a “median” of $10 in my pocket, what you would pay for it? Impossible to say.
Anyway, I understand that you have metrics for building your network, and I don’t disagree with that, but when used in a way other than “best fit” the analysis should be changed.
February 15th, 2007 at 4:52 pm
Dave,
Rereading your comment, I think we got our signals crossed: I’m not talking about what the zestimate represents, but rather the analysis of the error. In that case, I think a drastically wrong estimates SHOULD affect your accuracy rating more than a slightly-worse-than-the-median estimate. Likewise when you get it spot-on, that should improve your score.
January 3rd, 2008 at 4:56 pm
I’ve actually sold one of my clients homes through Zillow. He was researching his homes value and it just so happened to be that my listing was a couple blocks away from his home. He called me up and we had a transaction in play within a week. Thanks Zillow.