Inventory dropping to 2006 levels says Miller's numbers...

You're so hilarious, steveF. A bit pitiful in your desperation. Hey, man, wait about 10 years and it would go back to 2007 prices again. :) Don't worry about it, just wait it out, unless you're desperate to unload right now.

i think this research is pretty weak

Comparing Month-to-Month Inventory Levels....

Really....do we need to go over this for the 1,287 time....

see this explanation posted on Curbed as to why Miller's statistical methods are suspect:

"Miller's methodology is completely fucked up, making any comparison with past statistics meaningless.

Refer to the following thread on Streeteasy:

http://www.streeteasy.com/nyc/talk/discussion/12284-jonathan-miller-and-questionable-inventory-statistics-part-ii

Apparently, he has adjusted his historical inventory numbers by adding back - get this - deals that subsequently closed but had not previously been counted in inventory (i.e., listed).

So historical inventory equals (A) listed inventory + (B) subsequent closed transactions not previously listed, whereas current inventory numbers only equal (A) since, of course, closings are not yet available so (B) is still a mystery.

Therefore it is an apples-to-oranges comparison and completely fucking useless. Either historical inventory is OVERSTATED in the comparison or current inventory is UNDERSTATED"

beat - no. if you read the thread, you'll see that the adjustments that you describe are a theory proposed by a poster and subsequently disclaimed (as in, "i don't know how it's done") in her next post.

angler and Miller himself note the change in the data series in the link below. I have not seen an actual explanation of the new methodology, just this statement that there is a new one.
http://www.streeteasy.com/nyc/talk/discussion/11718-are-the-miller-samuel-inventory-figures-correct

Sidelines - thanks for that explanation. But doesn't the mere concept of "backtesting" mean that, until the current inventory levels are also backtested, to compare them to adjusted historicals is still an apples-oranges exercise.

Exactly, beatyerputz, but they're still yapping away on the other thread.

Since all quarterly snapshots are later revised upward, any pattern will still hold true once you look back past recent unrevised quarters. Seems simple enough.

I think the angler explanation is the thing to focus on. Miller had two data series, the one he was publishing and another that he was tracking behind the scenes. At some point (Dec 08), he decided that the behind-the-scenes one was more accurate and started to publish that instead. That's the same thing as saying that Dec 08 was the point where the backtesting of the original data series/methodology broke down. Looking at the differences between the two series, one could reasonably suggest that he was a little late in reaching this conclusion, but for whatever reason (clean break/fresh start at year end?) he did at the end of 2008.

The key point is that the idea that the historicals were adjusted is not accurate. The historicals were tossed out and a new series substituted in their place. The current numbers derive from the new series that he has been tracking (but not making public) for years, so they represent the current application of this admittedly mysterious new methodology that has not been described to the market. The current numbers are apples-to-oranges with the old (i.e., pre-2008, as originally reported) numbers, but apples-to-apples with the historical data series Miller now uses, for example in his charts on market trends. See my post at the very bottom (for the moment) of the thread referenced in Curbed for more.

"Since all quarterly snapshots are later revised upward"

But they aren't. Your post suggests some kind of rolling or periodic update to previously published historical numbers. What happened was a one time shift from one data series to another, so every number changed once and at the same time. Whether there will be another change based on yet another new methodology in the future is another question, but there isn't and hasn't been a running series of updates.

Okay, so here's the next problem.

Those of us who have been watching inventory as calculated by urbandigs (which has used a consistent methodology throughout) have seen inventory jump from around 5500/5600 in late 2007 to approx 10,400 now (a little less 90% increase).

If Miller's methodology were also consistent, I would expect to see a similar % increase from end of 2007 to present, but his numbers show around a 40% increase. Still a big number, but way off of the increase demonstrated by another (consistent) methodology.

All this to say, until Miller demonstrates how his numbers are consistent throughout his chosen time period, they remain highly suspect and should not be trusted.

beat - yes. difference between old Miller voodoo and new Miller voodoo (as well as what new voodoo actually is) needs to be explained before any Miller voodoo can be trusted.

Blah blah, its actually BEARISH news, not bullish news like the idiot OP claims. Miller specifically says June 2006 was a fluke. Secondly, these crazy Miller-bashers, if 100% correct, can only mean that inventory is actually a lot HIGHER than miller currently states, which would mean shiznet is hell of whack, yo.

"Miller had two data series, the one he was publishing and another that he was tracking behind the scenes."

Translation: Miller keeps 2 sets of books: the real one's and the one's he publishes to further an agenda for public consumption.

"Blah blah, its actually BEARISH news, not bullish news like the idiot OP claims. Miller specifically says June 2006 was a fluke. Secondly, these crazy Miller-bashers, if 100% correct, can only mean that inventory is actually a lot HIGHER than miller currently states, which would mean shiznet is hell of whack, yo."

can't blame a bull for trying...