Hacker News Reader

Comment by h2odragon | Hacker News Reader

Zillow's estimates of home values in my area have consistently been 150% of actual prices, as far as I've seen over the past years. Not that I've looked that closely and given that this is a hard area for them with low sales and a broad range of values for reasons not easy for their algorithms to quantify.

However, I always took that to be intentional, sort of a "see how valuable your home is / how high the values are where you're looking at"; hype that cost them nothing but made everyone feel better. Did the person that set that "glamor bonus" fudge factor get fired or just wasn't allowed to talk to the team deciding offer prices?

View on HN

The Zestimate history for a condo I used to own is hilarious. There have been several sales of this property over 5 years, all at about the same price.

If you follow the Zestimate through time, it climbs rapidly until there is a sale, then it goes "oh shit" and plummets down to the sale price, which is barely different from the last time it was sold. This cycle repeats each time the property is sold.

Zillow is absolutely certain that this property should be rapidly increasing in value, no matter how many times it sells at about the same price it sold for last time.

When we sold the property, the agent told us it was in an awkward spot: too expensive for single people, not quite big enough for high-earning families. So it's stuck in price limbo as all the neighboring condos skyrocket.

Based on this experience, I remain convinced that there are micro-structural factors the fancy data science pricing models just can't seem to capture yet.

It varies widely by Area how accurate the zestimate is. In some places Redfin is much higher than Zillow, in others places it’s reversed. Sometimes they’re too high, sometimes they’re too low. I presume the algorithm does best when the prices are fairly flat, but in this market real estate prices changed dramatically and it seems pretty impossible for an algorithm to figure it out without modeling the entire economy and movement patterns of the population.

Another anecdote: I’m in Oakland, where condo prices went down slightly during the pandemic overall, yet Zillow’s estimates are always much higher than Redfin’s (which seem to be in the same ballpark as the sale prices).

maybe Redfin actually incorporates "price at end of closing" into their models, instead of say .. "wishful drinking" prices from Zillow and crew?

Odd. My house is valued at $330k on zillow, would sell for $420K+ in a few days.

Mine is pretty accurate ($2.5M), based on recent sales and a couple of solicitations from realtor friends.

FWIW, we have had a high frequency of home sales in the past year, so that might help. I just looked up my previous house (sold 5 years ago), and Zillow is saying $740K, which I am pretty sure is massively overvalued, however the street it is on, and general area, have a fairly low turnover, so less recent data.

Humble brag

Not if you bought a home a decade or two ago and house prices went up around you with no action of your own.

I drive an old Honda Civic and my neighbors all have Tesla, Porsche, etc.

I didn't mean it that way, and it's all relative really. There are a ton of houses above and below my price point.

No, you're rich.

I'm working on my second million!

...I gave up on my first.

Mine is pretty accurate as well ($44.8M), based on recent mentions in Architectural Digest and a couple of solicitations from fellow G650 owners.

There isn't exactly a large market for homes at that price point. Plus at that price, a thorough DD is an order. So not unexpected that your home's value would be accurate.

What's "a thorough DD"? I assume it's building on the previous joke but I don't recognize the abbreviation.

Due diligence.

It varies a lot based on location and what's sold locally. If there aren't a lot of similar properties in the area that have sold recently it will be much less accurate.

Reply on news.ycombinator.com