Neighborhood level Case Shiller

[SummitSeeker]

Calling all tech-savvy people with too much time on their hands, anyone interested in programming the Case-Shiller index methodology? This would allow us to examine price trends in the geographical boundaries of our choosing, rather than lumping King, Pierce, and Snohomish counties together. The index methodology is given here:

http://www2.standardandpoors.com/spf/pd ... gy_Web.pdf

It basically boils down to some matrix math. The hard part will be parsing the sales data (avoiding duplicates, matching sales pairs when the address is entered slightly in each sale) to feed into the algorithm. Plus I think you'll need MLS access to get the data. Someone has already done this for the San Diego market and the results are illuminating:
http://sdhpi.blogspot.com/; code used to make the index:
http://sdhpi.blogspot.com/2008/02/constructing-hpi.html

It seems like something that can be done in Excel, but I don't know visual basic. I'm sure other programming platforms could be used. Anyone want to take up the challenge? Everlasting glory will be your reward.

[deejayoh]

maybe an easier alternative, if you are interested in a particular neighborhood, is to go to redfin and select all previous sales available, and use that data.

couple of caveats -
the best way to look at thedata is probably using $/sqft - but that's the way Radarlogic does it and their index pretty much tracks to case shiller

It can be a bit cumbersome to pull the data in that you can only get 500 listings at a time - but if you're looking at a relatively compact geographich neighborhood, it's possible to do it with 2 or 3 searches if you use price bands as the means of splitting it up.

I've done this for many seattle neighborhoods in the past (Madison Park, West Seattle, Queen Anne, Capitol Hill, Magnolia). It's been a year or so since I last looked at it - but it didn't take too much time.