Distribution Days Are Important – For A Fade

From Richard Russell in Dow Theory Letters:The recent market action has been characterized by many days of divergence when the Industrials rise and the Transports fall or vice-versa. Robert Rhea wrote that two or more successive days of divergence usually suggests distribution. And speaking of distribution, the action of the last 30 days has been characterized by many “distribution days” when the market declined on rising volume (that is rising volume over the volume of the preceding day).

The latest count of distribution days over the past two weeks is ominous and highly unusual. There were 6 distribution days for the Nasdaq and the S&P 500 and the NYSE Composite. There were 5 for the D-J Industrial Average. That’s far too many distribution days for a healthy market.

This is one of those pieces of fabled market lore that is very easily tested. And as is often the case, there is little truth to what the textbooks suggest.

Let’s just take the past 12 years, when we’ve witnessed nearly every imaginable type of market. Over that time, there have been 154 days which registered 5 or more distribution days over the prior two weeks, and 147 days that registered none. Supposedly, the former should be apocalyptic signs of a terrible market, and the latter the best possible environment.

Well, over the next two weeks, the days that had a lot of distribution over the past two weeks returned an average of +0.2% with 54% of them being positive. The “best” days that had no distribution over the prior two weeks returned an average of -0.2% over the next two weeks, with 56% of them positive. Not much of an edge, right?

Let’s look at the data since the October 2007 peak. Of the 33 days that showed 5 or more distribution days over the prior two weeks, the next couple of weeks returned an average of +0.5% with a 58% win rate (the next month showed even better results).

How about those good times with no distribution over the past two weeks? Well, over the ensuing month those returned an average of -6.4% with a 21% win rate.

Please do me a favor and take the concept of “accumulation” and “distribution” days with a huge grain of salt. Especially nowadays when volume figures are so suspect.