Monthly median sales in a particular zip are the result of the particular mix of sales in a zip code with 25 or so samples. This small sample produces noisy estimates.
Any statistic with 25 samples is going to be noisy. I thought you were a statistics guy. If you have trouble with the concept of noise in time-series data, then go ahead and continue to believe that La Jolla’s prices have dropped by 34%. I don’t care.
But if you believe that, then you have to believe that prices have recovered 9% from the bottom in Dec 2006.
