In response to the question about how we calculate data, we base the data solely on the raw numbers we get from the MLS. Therefore if a listing had it’s price reduced several times, then it is delisted, relisted at lower price, then sold, that would actually show up as 2 separate entries. That would not hide the fact that a home had its price reduced, but it would throw off the percentages since a home would be counted twice and therefore the price reduced/long days on market listing would have less of an impact. That was one reason why we chose medians and not average b/c medians are affected less by outlying data and manipulation (provided there is a reasonable amount of data). We try to “clean” up data that is obviously wrong but some things are harder to detect programmatically that it may seem without also discarding “good” data. Overly ambitious cleaning of the data can also create errors in the opposite direction as well.

As for the manipulation of the market, that is always possible no matter what we try but it seems to happen less frequently than it may appear. For example, the median days on Redfin for Menlo Park is 62 days for houses so it wouldn’t seem to be that much manipulation going on. Admittedly, median days in Palo Alto is much shorter but talking to our agents down here, Palo Alto is still a ridiculously hot market so 24 days seems reasonable for a median there.

John Kim
Product Manager
Redfin

% of listing price is somewhat useful if we combine it with the trend in listing prices; we can find that at
http://www.housingtracker.net/...

Listing prices in 8/2007 vs 8/2008
25th percentile: $449,962 $299,950
Median, 50th : $598,569 $499,700
75th Percentile: $790,750 $798,500

so there does appear to be a high end, hanging tough... while the price slope becomes precipitous. The Real Bay Area, where prices only go up, seems to get smaller and smaller...

"Days on Market" and "% of listing price" are essentially useless statistics imo, because they are all too easily manipulated.

Hi Ellie,

A questions about how you record price reductions in your neighborhood pages:

When there is a reduction in the price of a house that occurs at the same moment that the realtor pays the $25 fee and changes the MLS number, is that recorded as a price reduction in your system or does the system miss that it is actually the same house? Here in Palo Alto and Menlo Park, the routine is this: the moment that the seller is about to accept an underbid, the realtor closes the listing, opens it up again under a new MLS number, and then adjusts the price to match the offer that they are about the accept. That way, the statistics for the listing is "1 day on the market, sold at 100% of the listing price, no price reductions -- see! Palo Alto and Menlo are immune to market forces"

You should base your statistics on initial price, not the adjusted price and then leave it to the analysts to decide exactly how to discount unrealistic sellers. During the boom no one had a problem reporting huge final to initial price gaps. The sellers were similarly "unrealistic" if you will, setting the initial price at a level far below what they would be willing to accept.

Pop, we based this on final list price, not initial. WE do it that way because a lot of initial list prices are unrealistic, and the buyers don't even begin to negotiate until they come down to reasonable levels. But you can get at the number of price reductions in an neighborhood in our neighborhood pages- for example, the fourth graph on the Telegraph Hill page shows the price reduction trend there:
http://www.redfin.com/neighbor...

I'm not sure if agents are using the trick you described, but what we hear from SF Redfin agents is that many homes are getting multiple offers, and homes are indeed selling above list in many cases.

Comparing sale price to asking means nothing except that agents are becoming better at setting the price. Case Schiller data shows house values declining in some areas of SF and increases in others. In declining areas, in might make sense to wait to buy - no big rush.

Could the final list price be deceptive and skewing the figures here? What seems prevalent in some markets is that if a home or condo goes through multiple price reductions to finally sell, the last price is logged in the MLS and other data sources as the List Price. The home finally dropped enough to reach reasonable market value, and got an offer at approximately 100% Final v. List Price - but the Final v. Original List may be more like 75% or a worse percentage. If Redfin did the homework and the "List" numbers above don't include any price reductions, my apologies and thank you for providing the real numbers!

Reading between the lines on some sales, it also seems that selling Agents might be able to pull a "fast one" on the MLS with the following trick: an offer comes in below the Final List which the seller accepts, but because Agents (and sellers presumably) want to appear to be smart and effective pricing strategists, they re-list the home for just a day at a fraction below the offer accepted. This is just a lay-person's guess at how some homes/Agents were able to claim "sold at or above asking." If this isn't possible, please do correct me!