One small detail that is easily missed, housing starts measured on the way down are not the same as those measured on the way up.
If you look at the square footage, expected sales price, etc. of starts being put on the books today, they are far lower in both areas as those at the peak. Thus, prices and sizes return to nominal levels, which actually support further declines in existing values, until evertything catches up to the new constructions.
This at least partially explains the lag seen between housing starts and prices. The starts you observe ‘at the bottom’ are seen as a baseline for where prices are eventually going, and if you consider that prices almost always over-correct at the bottom until they turn, it is easy to deduce that, price wise, we’re nowhere near a bottom.