I am glad to learn that Bloomberg is doing that, although I cannot really understand why they would not have added a Summation Index if they already calculate a McClellan A-D Oscillator for a group of stocks.

Another outfit, High Growth Stock Investor, runs a service that allows users to perform a variety of studies on a given set of issues, such as a sector index's components, or some other list of stocks.  Included among these studies are the McClellan A-D and Volume Oscillators and Summation Indices.  For information on their service, see http://www.highgrowthstock.com/.  

There are a handful of big pitfalls to understand when trying to approach such an enterprise.  The first is coming up with the list of components to a sector index, which is not too hard lately but was quite a chore when we first started studying this topic in the mid-1990s.  Once you have the current list, you can put that in to your database and output the results pretty easily, provided that you have an accurate database with all split adjustments done and no bad data points to pollute the calculations. 

But a funny thing about index components is that they are frequently changing, so if you are going to build a historical data set then you will need the dates of all past changes in components, and you will need to be able to hunt down data on the stocks that have been kicked out.  Some of those stocks may no longer trade, and in some cases the symbol gets recycled so that finding the historical data on an old company's stocks is rather hard and you may end up getting data for the new company now using that symbol.

Even if you do go through all that trouble, you will need to devote somebody's time to keeping up with future composition changes, and manage the adjustments to your database.  The more indices/ETFs you follow, the more of a full time job that can become.

We have built spreadsheet files to do this for the 30 DJIA stocks and the Nasdaq 100.  We publish the breadth stats for the Nasdaq 100 in our Daily Edition every day, including its Summation Index, along side the same indicator data for the NYSE and Nasdaq.  It took a lot of manhours to pull off creating these files.  Changes in the DJIA have been relatively few in number, so we were able to create that dataset all the way back to 1988.  The Nasdaq 100 was harder to hunt down all the changes for, partly because they change out around 10-15 issues each December.  That is on top of the changes due to mergers and delistings.  Making split adjustments is something we also have to worry about, so we had to build in split-detection formulas to help us notice when prices change by more than 20%, just so that we can go and check to see if a true split has occurred.  This is easier than reading the list of adjustments published each day by the Nasdaq just to see if any of the issues mentioned are among those included in the Nasdaq 100. 

If you care about tracking up and down volume, it gets even stickier because typically the historical volume figures for a stock are adjusted at the same time as prices in case of a stock split.  As a result, building a database for up and down volume from historical data is bound to have errors.
 
The chief value of A-D numbers is as a liquidity indicator.  There is no way that the vast majority of stocks can go up day after day for a protracted period unless the market and the economy are flush with liquidity.  When there is so much money out there that even the most undeserving stocks can go up, you have a great condition for investing.  Conversely, when you see prices of the major averages going up but the A-D Line going down, you know that you have a problem that is going to cause a lot of damage at some point.  The 1998 to 2002 period reinforced that lesson, because we saw persistently weak overall A-D numbers which foreshadowed the serious price damage to be revealed later in the major indices.

One of the biggest downfalls of exploring breadth data for a single sector is that you lose the great inherent property of breadth data, i.e. you want to be able to see the effects on a diverse group of stocks.  Focusing on a single sector misses out on what the rest of the market is doing, and therefore loses you the benefit of measuring the effects of liquidity, which is the whole reason why breadth statistics are so useful.  Also, single sectors can often exhibit excessive homogeneity, in that the stocks all tend to act like each other all the time.  In gold stocks, we often see all of them up one day, then all down the next.  Chip stocks are this way too.  In a more vague and diffuse group like "technology" or "basic materials", you may see more differentiation, but the homogeneity effect is still there.

As a result of these difficulties, we have pretty much abandoned our efforts to create several different mechanisms to track the sectors we were interested in.  Maybe somebody else can figure out a way to do it without as much "manual labor", but in our view the payoff was not worth the price of the effort.  That was not the conclusion we expected to reach at the beginning of our exploration, but as with many technical analysis questions, one must go fishing in the data just to find out of there is any utility there.  For those willing to put in the effort at exploring multiple avenues of analysis, the reward is that every once in a while you run across something really useful. 

I hope these insights are helpful.

 

Tom McClellan
Editor
McClellan Market Report