Hard or Soft Landing for Real Estate?

There is no shortage of punditry on how this real-estate bubble will end.  Every week it seems more data pours in supporting both a hard and soft landing.  Here are four ETF indicators that represent the current confusion...

Homebuilder ETFs showing definite weakness:

ITB:  iShares Dow Jones US Home Construction

XHB: SPDR Homebuilders

Broader real-estate index ETFs showing sideways movement with some upside strength:

IYR:  iShares Dow Jones U.S. Real Estate Index Fund

ICF:  iShares Cohen & Steers Realty Majors Index Fund

Any contributions regarding other indicators, or the weaknesses of these indicators is welcome.

Comments

[TN]

In the patrick thread that you mentioned that you used a neural net to signal short opportunities? What are you using as your inputs and your training data for the NN?

[randolfe]

I use quite a few, depending upon what I'm trying to predict. To generalize, I start with some standard stuff like:

* correlations to indices (with usually a lag of 4 for a daily model)

* open & close ratios to other stocks if it's a pair-wise fit

* momentum stuff like standard MACD, but usually some variable oscillations based upon my fundamentals analyses

* simple historical volatility stuff

* usually a few XY r-squared correlations with lag to rotational sector indices or sometimes just based upon my sentiment about macroeconomic direction

I usually try to predict simple direction first with a 75% or greater accuracy. Then I try to predict some correlative changes; usually simple correl and rsq. Finally, I use those NNs as inputs into specific prediction models.

I usually use a committee of NNs so I can account for both feed-forward and time-delay style topologies, since different types can become better depending upon how quickly the underlying data changes.