[tgrass]

I'm surprised they're aggregating at the zip code level. The neighborhood differences within a zip code seem too great for an average measurement to have much meaning to an individual homebuyer/seller. Case Shiller breaks values down into three tiers of price, which provides a little more insight.

[sp_]

The arithmetic average for home prices is not too useful either. I clicked on a random county in Wyoming that had an average of more than $600K. Turns out that it's like 20 cheap houses (most going for $150K to $200K) and one huge farm for $29.5 million.

[tgrass]

Who needs a median or mode when you have a pretty map, though?

[Joakal]

How can they measure to give more meaningful prices?

[tgrass]

Personally, I'd want it at the neighborhood level. Though there may not be enough data within a reasonable timeframe at that level, it can at least be reasonably compared to a broader index for the city.

I don't think you get much at the zip code level that you wouldn't get at the Metro level.

The probability of a second reduction is also too simple. It makes no account for the market turning positive. That's precisely the error in modelling that failed us on the upswing: the assumption that present trends will continue.

[julesallen]

A pricing heatmap in 10% bands would be pretty good I think. Houses next to each other tend to be more or less the same value with a few edge case scenarios -- the farm as noted in the other reply to this.