There are various tools used for detecting accumulation or distribution in stock indices... this thread looks at an approach that at least I have not seen before.
A little background in how this accumulation-distribution is constructed:
1) For the index of interest, all of its components' high-low-close prices are needed, as well as each component's closing volume
2) Using each component's high-low-close (HLC) daily prices, the price change from the prior day's close is determined for each component. For example, if a stock's high is $10.00, its low is $9.00, and its close for the day in question is $9.50, and the prior day's close was $9.75, then the high price change is 10.00-9.75 (+0.25), the low price change is 9.00-9.75 (-0.75), and the close is 9.50-9.75 (-0.25).
3) The calculations from step (2) are then multiplied by each component's respective closing volume and added up. So essentially the Price-Change-Volume ($PCV) result is the index theoretical high $PCV, theoretical low $PCV, and closing $PCV, for all of the index's components.
4) From the raw HLC $PCV sums for all of the index components, a method developed by Marc Chaiken is then utilized, the Chaiken Money Flow recipe is something like this, with High, Low, and Close being actual price in Marc's method:
The difference from the Chaiken accumulation-distribution methodology used in this "new" tool, is the "volume" variable from Marc's recipe is not used since it is already included in the High, Low, and Close variables as described in steps (2) and (3).
The end result, which I am calling the HLC $PCV accumulation-distribution tool works pretty well for detecting distribution in particular. The tool's accumulation detection prowess seems to be not as effective as its distribution properties, but there are many other great indicators for measuring accumulation... distribution as measured by money flows seems to be more difficult in detecting.
The HLC $PCV method has been used in experiments with broad based and more narrow based small cap and large cap indices. I have found the large cap indices provide the best distribution signals and normally the more components the index has, the better the signals.
The first chart is the HLC $PCV indicator of the SPX components. My daily files of SPX components with the needed raw data only goes back to August 2004, but over that short time, there are few examples of price tops where the HLC $PCV tool gave clear signals of distribution.
The HLC $PCV cumulative line must be considered over time frames of less than three or four months... as one can see lower highs in the displayed time frame. In December 2006, this tool clearly began exhibiting distribution tendencies, but then an interesting event took place over the past week or so.. the HLC $PCV indicator took off to the upside after posting a double bottom.
From the limited data, new recovery multi-week highs normally suggest the buyers have taken over control from the sellers and negate the previous divergence that suggested distribution was underway. It is difficult to tell from the chart, but the HLC $PCV tool has taken out its early December high.
Next is the broader based Russell 1000 (RUI) index HLC $PCV indicator. A clean break of its early December high has occurred, essentially trumping the distribution behavior observed through the month of December. Note the weakness of the HLC $PCV tool leading into the May 2006 price high.
The NDX HLC $PCV variant surprisingly works OK as well. Normally more narrow based indices are more impacted using price-change volume techniques, due to the effect one or two issues can have on the composite data. Like the SPX HLC $PCV variant, it is difficult to discern from the chart, but the NDX HLC $PCV line has also exceeded its late November-early December 2006 double top.
In January 2006, the NDX HLC $PCV tool displayed a spike high coinciding with a price spike high, one of the very few times price and the HLC $PCV have topped in concert... verifying once again no single tool in perfect all of the time!
There are many other tools under development using the HLC $PCV data, and one shorter term tool using ratios, is suggesting a short term price high may not be too far away. However, if the intermediate term cumulative HLC $PCV line holds up well during any near term price weakness, the probability is good there is more price appreciation likely over the next several weeks.