Wednesday, May 6, 2009

Very Narrow Range

Tuesday’s SPY range was the lowest in over 3 months. Since 1999 there has been a bit of a downside bias following narrow range days. The narrower the range in relation to recent ranges, the more bearish the inclination has been. Below is a table showing the 10-day returns following a day that was the narrowest range in X days:

Of course this morning the excitement over the jobs report seems to be trumping any narrow-range tendencies so far.


Douglas said...

OT - Falling Volume:
Bear Market Rally
New Bull Market

I have been crunching the data on a new model that I am developing.

Going back to 1962 on the S&P 500 I have taken all the instances where (i) the index is 2% or more below the 200 D MA, (ii) the index is above the 50 D MA, (iii) volume is declining based on 10 day volume moving averages, (iv) volume is declining based on the 30 D MA of the volume.

Prior to the current rally I find 28 periods of time when this occurred. One time in 1966 it was at the start of a bull run and there was one day in March 2003 also.

Apart from that - looking at the charts - I would define the other 26 periods of time when this occured (usually, like now for several days in a row) - to be bear market rallys.

The reason why I post this comment is to see if it provokes any ideas or comments from yourself or others that may help my modeling.

I am intending to program this excel model further to look at (i) more moving averages (on volume), (ii) timing (I noticed some interesting peaks, troughs and sudden turns in volume) and (iii) the time from when these conditions took place to when the market would then reach a lower level (e.g. what was the maximum wait in these situations before the market turned down or reached a lower level than when these conditions were taking place).

Apologies for the long off topic comment but if anyone else who is interested or who crunches the data has any ideas this might all.


Daniel said...

Much Food for Thought there, Douglas. And it is far from “off topic”-- much of Rob's work is about volume, and when and where it becomes key.

Several of his recent posts have sounded alarm signals that I consider to be of Long-Intermediate timeframe significance, which is where your model seems to be focused.

Also, much of his work is about adjusting and nuancing parameter-sets, depending on where a datapoint is sitting, with regard to key MAs. He often shows how that can skew perceptions and results to the bullish or bearish end of a results table.

His recent longer-term volume work, (taken along with some fundamental valuation considerations), seems to be hinting that we may be facing an 18-month sideways market.

One that can't go down too far because of cash on the sidelines. One that can't go up too far because we still have not yet hit mean-reversion to Bear-ending P/E levels. (cf. the work of Hester and Hussman).

One good way to build your model would be to test it along the way, like Rob does. Devise a simple algorhythm which captures the essence of one component, make it into a trading system, and see how it does... Then do the same with another, etc.

Let us all know when you perfect it, so we all can get rich. Heck, Douglas, to know beforehand whether a Move like this one is to turn out to be the Big Heartbreaker, or a genuine First Upleg in the Bull Market of 3/6/2009 -- [ ? ]... to know that is like knowing the ending of a detective mystery film beforehand.

It sure makes reading the clues easier. So good luck.


Douglas said...

Thanks Daniel

The reason why I am trying to do this is because the other models I have give very different results based on whether we are in a bull market or bear market or at the start / end of a bull market.

So, I am looking at a few different indicators that might give a clue as to where we are in context. Volume is one of them - based on an idea I picked up on a different blog.

Interestingly volume seems to have picked up a bit in the last couple of days - sadly I've not had much time yet to develop this model further.