
I recently introduced the Relative Dollar Volume Flow indicator and took a look at dollar inflows and outflows in the Dow 30 Industrial Stocks. The indicator has been useful in detecting weakening dollar inflows during topping processes and in identifying market bottoms. More information on my research into Dollar Flows can be found on the Trader Performance page of my personal site.
First off, just a few notes of explanation. The basic logic behind the indicator is that we look at every trade placed during a market day and see if that trade occurred on an uptick or on a downtick. We record for each trade the price of the trade times the volume of that trade. If the trade occurred on an uptick, it is considered an indication of buying interest and is added to a cumulative daily total. If the trade occurred on a downtick, it is considered a sign of selling interest and is subtracted from the cumulative daily total. That total at the end of the day represents Dollar Volume Flow. I divide this value by the day's share volume to obtain the *Relative* Dollar Volume Flow. That enables us to compare dollar flows for lightly traded stocks with those for more actively traded issues. Basically we're seeing the proportion of total volume that is represented by buying vs. selling interest.
Because the Dollar Volume Flow is calculated by multiplying the size of trades by their execution price, large trades greatly affect the total. That means that the Dollar Volume Flow is sensitive to the buying vs. selling sentiment of the largest market participants.
My final calculation is that I subtract the prior 200 day moving average of Relative Dollar Volume Flow from each daily value. This creates a zero mean, in which zero represents average Relative Dollar Volume Flow. Accordingly, I call this *Adjusted* Relative Dollar Volume Flow. It tells us whether dollar volume is above or below average for each trading day.
In the chart above, I have taken 40 highly weighted S&P 500 stocks and calculated their Adjusted Relative Dollar Volume Flows going back to January, 2006. The 4o stocks are evenly divided among eight market sectors: materials, industrials, consumer discretionary, consumer staples, energy, health care, financial, and technology. A 10-day moving average for Adjusted Relative Dollar Volume Flow for the 40 stocks is plotted against the daily closes for the S&P 500 Index (SPY).
What we see is that markets tend to continue rising as Adjusted Relative Dollar Volume Flows rise and top out following peaks in those flows. This pattern enables us to see divergences between price and flow prior to market declines. At market bottoms, we see negative Adjusted Relative Dollar Volume Flow figures that quickly transition to positive values as institutions snap up stocks at bargain prices. This is how we know a bottom has been put into place. Market declines tend to continue until they attract significant buying interest from the largest market participants. We see that clearly off the July market bottom in the chart.
Note that, as emphasized earlier, we are not yet seeing signs of aggressive buying despite the recent market decline. Across the 40 stocks highly weighted within the eight major sectors, dollar volume flows remain subnormal. In absolute terms, we are still seeing, on balance, positive flow numbers. These flow values, however, are well below their 200 day moving average, which has tended to correspond to price weakness.
So, to address the WTF question: Where's the Flow? It hasn't yet returned to the S&P 500 market. Accordingly, it appears to be premature to call a market bottom.
Next stop on the research agenda: We'll look at flows sector by sector to see which ones are holding up best and worst during this period of market weakness. There, we'll find at least one sector that *has* regained its positive dollar flows.


6 comments:
Excellent analysis, Brett. One quick query: attention has recently been drawn to the growing use of "dark pools", large institutional trades executed outside of the public market structure; it has been suggested that this is mostly problematic for indicators based on bid-ask analysis, but I wonder if there might not be some potential skewing of tick-based indicators as well - ie. it is not clear to me that such trades are necessarily in synch with the uptick-downtick movement of the public auction.
Great post as usual Doc.
I wonder to what extent dark pool trading would distort this method of analysis. From what I understand among the participants these days in dark pools are executives with large stock option grants.
Dr.Brett:
Your research never ceases to intrigue me. Keep up the great work.
I am not sure if this is the proper forum but I have started blogging again. The new new site is www.thetradingdigest.com/blog
Hopefully you will stop in like you did at the other. The site is still rather new but will be up to standards this week.
Dave Johnson
Steve Haughey makes a good point.
I was never too satisfied that a size trade occuring on an up or down tick signified the 'aggressive side' We do it that way because we have no reasonable alternative.
Now , with much of the real volume occurring AWAY from the displayed market, tick references are almost unjustifiable. Stocks tick so fast and trades post late, so really, it's almost ludicrous.
Not to diminish the good Docs efforts, but I think the data is bound to be flawed. You might be able to see something of value in it over very large samples though.
Hello and thanks to all who have commented. I'm on the road, so I have limited opportunity to respond.
I agree that, if we get to the point where a significant portion of volume is transacted away from the exchanges, that could impact the flow analysis. I don't think we're there yet.
The acid test for whether the flow data are useful is whether or not they add predictive value to attempts to model future price changes in stocks or averages. If the flow data provide historical patterns to trade above and beyond those achievable with current measures, then they are valuable.
Although some of the predictive modeling research with these data is proprietary, I will be posting some of the results. Those results suggest to me that the flow data do indeed capture market sentiment in a valuable way.
Brett
Hi Dave,
Thanks for the link and best of luck with the site!
Brett
Post a Comment