Serious Zillow issue

[BuyerWillEPB]

Lately, I have noticed an issue with the the way Zillow charts their Zestimates.

Whenever there is a sale to the upside, their Zestimate adjusts immediately up to meet that sale, and then continues on from that newly recorded base.

However, when there is a sale (or two, or three) to the downside, it appears that the computer algorithm completely ignores them and continues on using the last high sale point.

I posted an image that demonstrates this on my personal blog.

Click here.

Sorry, but the image came out a little difficult to see. If you want to check these exact Zillow charts, the addresses are:

4950 Monongahela St. 92117

5111 Jamestown Rd. 92117

[The OC Scam]

I have noticed that here in the OC also!!

Our property has gained 5 percent in the last few months with no high sales but only lower prices??

The comparable homes section only reflects higher sale from last summer 2006. Another thing I don’t understand is the stating higher priced home sales from Feb 2007 which are bigger homes in a better part of our zipcode?

Engineered for price gain only?

[The OC Scam]

I have noticed that here in the OC also!!

Our property has gained 5 percent in the last few months with no high sales but only lower prices??

The comparable homes section only reflects higher sale from last summer 2006. Another thing I don’t understand is the stating higher priced home sales from Feb 2007 which are bigger homes in a better part of our zipcode?

Engineered for price gain only?

[The OC Scam]

I have noticed that here in the OC also!!

Our property has gained 5 percent in the last few months with no high sales but only lower prices??

The comparable homes section only reflects higher sale from last summer 2006. Another thing I don’t understand is the stating higher priced home sales from Feb 2007 which are bigger homes in a better part of our zipcode?

Engineered for price gain only?

[bsrsharma]

I have seen these problems with Zestimates

1. They appear sticky on the down trend since the prices are dropping faster than their algorithm can assimilate new prices into the system

2. Some data are just plain wrong. I see many sharp upswings and downswings where either they have bad data or the algorithm is choking on non-existing data (like using 0 for input data when data isn’t available)

3. In some cases I have seen it shows sharp upswings but gradual downward slopes, giving an impression that they are eager to increase the zestimates but reluctant to lower them.

I think, on the whole, it is still an immature model.

[bsrsharma]

I have seen these problems with Zestimates

1. They appear sticky on the down trend since the prices are dropping faster than their algorithm can assimilate new prices into the system

2. Some data are just plain wrong. I see many sharp upswings and downswings where either they have bad data or the algorithm is choking on non-existing data (like using 0 for input data when data isn’t available)

3. In some cases I have seen it shows sharp upswings but gradual downward slopes, giving an impression that they are eager to increase the zestimates but reluctant to lower them.

I think, on the whole, it is still an immature model.

[bsrsharma]

I have seen these problems with Zestimates

1. They appear sticky on the down trend since the prices are dropping faster than their algorithm can assimilate new prices into the system

2. Some data are just plain wrong. I see many sharp upswings and downswings where either they have bad data or the algorithm is choking on non-existing data (like using 0 for input data when data isn’t available)

3. In some cases I have seen it shows sharp upswings but gradual downward slopes, giving an impression that they are eager to increase the zestimates but reluctant to lower them.

I think, on the whole, it is still an immature model.

[Author]

Posts