« Zillow Tales: Where Do You Want To Start?
Barbara Corcoran on the State of the Real Estate Housing Market »

Redfin and Zillow: Stats of a Feather

Published by Joseph Ferrara (2) March 1st, 2007 in Zillow Real Estate and zillow. Comments

redfin-zillow.gif

Redfin just released its yearly statistics to show that Redfin agents, on average, get their buyers a better deal. The data showed the “average” Redfin buyer paid 99.233% of the list price while buyers using other brokerages paid 100.233% of the list price. Redfin proudly wears the feather in its cap. But wait. At first blush, an average paid “over list price” for an entire year seems highly suspect. Marlow Harris checked it out and duh–not so.

We post not to condemn or praise Redfin but to condemn, once again, the statistician’s black art. The Redfin sale-to-list-price ratio means ABSOLUTELY NOTHING. Why? Because it uses the group “average” to persuade use of an individual agent, one with a Redfin logo. It is the same sleight-of-hand used by Zillow to make its zestimates appear reliable. Zillow uses the median rabbit, Redfin the average one. The effect, in either case, is to mesmerize and impress.

Those who have exposed Zillow’s median error rate as misleading and useless in the individual case, may be tempted to use “average” statistics to measure value in the Redfin case. Don’t buy the bean counters’ conclusion. Whether an average or a median, the result is the same—-an amalgamation of the best, the worst, and the mediocre. A stellar agent can beat anybody’s pants off but he (or she) will always get lost in the statistician shuffle. It would be better to produce stats on each agent. This group soup approach is for the birds.

If you want to play with statistics, we propose you do it on an “individual” basis. Create a Real Estate Agent (REA) Rating to measure effectiveness. Much like a quarterback rating. You take the raw data—list price, sold price, Days on Market (very important factor overlooked in the report), pre-sale price reductions, seller credits, etc., hire a Zillowist to crunch the numbers and sauce it up and BAM! a REA Rating each agent can add to their business card, along with the alphabet soup of designations. Then we can poke holes in that system.

Thanks for reading our 1,000th post.

Technorati Tags: , , , , ,

Share This Post
  • 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.
  • Great post and I'm right there with you. I did a post a while back on the accuracy of Zillow and other AVMs followed up by an update post just a short while ago.

    David from Zillow left a few comments on my posts saying that my conclusions were incorrect and that the median margin of error was 7.8% rather than what I had concluded. But the data I had (which was supplied by Zillow itself) clearly showed otherwise.

    Seems that they're pulling these numbers out of thin air and that they think the public is too dumb to see otherwise after reviewing the true data.
  • Good question Michael. It depends on what you want to know.
  • So - where do we get reliable information and statistics?
  • The statisticians' bag of tricks never fails to amuse us.
    Company averages do not indicate the ability of a particular agent to save you money on a particular home. We have used the example of zillow. Just because the median error rate is 7.5% is no reason to rely on a zestimate on a particular home. The forest is not the tree.
  • As somebody who takes great delight in crunching numbers, let me say that for the entertainment value, it's a lot of fun. Tell me what outcome you want, and I can probably find you a data set and a methodology that will prove my point.

    Republicans tout the stats that show the tax cuts over the recent years have gone disproportionately to the middle class; Democrats tout the stats that show these same tax cuts have overwhelmingly benefited the rich.

    They're both right.
  • Thanks Mark. It's always great to hear from an expert on these matters.
  • Since I'm a stats junky who hates it when the media jumps all over the latest study that shows cancer risk going up by 10% (from a 0.01% chance of getting that particular cancer to a 0.011% chance...), I have to comment.

    It seems to me that independent of what you're measuring. a 1 percentage point difference between two means, especially when one distribution is a small number of samples (Redfin's sales) and one is large (overall market) means nothing unless you have done a statistical test for significance to determine the probability that the two distributions are really different. I'll bet the significance of the difference is pretty low - in other words a serious statistician would dismiss this difference as meaningless.

    In my opinion, reasonable (though somewhat dishonest) marketing, complete junk stats.

    Cheers,

    Mark B.
    VP Marketing at Movoto.com
  • Was it Mark Twain, "figures don't lie but liars figure"?

    Congratulations on the 1,000th post, I look forward to many more.

    Howard
  • Roberta...LOL.. Zipfins... it's perfect...I think there are a number of things missing from their reports.. DOM would indeed be something they should be posting..
  • Excellent, excellent points Roberta. As you say, pricing the home correctly will sell it quickly & usually at or very close to the list price. Overpricing will allow for easier discounting. And according to the stats, the overprice/discount sale looks better. And where are DOM in the stats?
    Zipfins---clever.
  • Sellsius, Kay and Michael: Zipfins can spin the numbers anyway they like, but no truth is revealed. Too bad days on market are not included in their stats.

    Well-priced homes sell quickly.

    Overpriced homes sell at a discount--sooner or later.

    I have a hunch Redfin buyers are taking advantage of the latter and that Redfin is taking credit for the discount.
  • Statistics can be spun in many ways. If the median looks bad, try the average. Or take a different sample over a different time period. Use a different demographic...The list goes on.

    You should not chose an agent based on the overall company "average" of anything just as you should not use the median error rate to determine if the zestimate you're looking at is reliable.

    Thanks Kay and Michael for your comments.
  • Sleight-of-hand is definitely the right term for these statistics..
    If it wasn't so deceptive it would be amusing... but it is deceptive. While most consumers question any statement that comes with a political overtone they will take these press releases as true news items.

    Redfin and Zip Realty are marginal players with a lot of money behind them. Both started in a hot real estate market and both use business models that are based on that type of market.
    It will be interesting to see how they fare in a down market where they may actually have to do something to earn their discounted fee.

    As for Zillow.. they started with the idea of disposing of traditional real estate agents much like they did travel agents. I believe they have learned that real estate is a bit different.. so they are jumping on the bandwagon to get the agent's money instead.
    They simply can't accomplish what they claim because it is impossible to acccurately decide value on anything but a new development of cookie-cutter homes. Once people start customizing... for good or bad.. their Zestimates will never be accurate.
  • Nice going, guys! It's amazing how much "databs" (that's Data-B.S.) is being tossed around by all of these mee-too'ers! After looking at zillow, trulia, redfin, realty-trac and all of these other former water filter/cellphone/long distance telephone calling card salespeople turned techno gods to save the consumer, it makes the NAR look like the straightest shooter when it comes to data reporting. Go figure! md
blog comments powered by Disqus
Blog Widget by LinkWithin