Charles Dow's Original Editorials and Their Relevance Today

Archive for December, 2010

Dow Theory Defined

Comments Off on Dow Theory Defined

By Paul Shread, CMT

Dow Theory might just be the most misused and misunderstood of any technical analysis theory, so I thought it would be useful to offer a simple definition of this most important stock market trading system.

First, it’s helpful to understand that Dow Theory was initially an economic theory developed in the pages of the Wall Street Journal by Charles Dow from 1899 to 1902. Dow’s theory was simple: if goods were being produced and moving through the economy, then it should show up in the action of both the Industrials and Transports, the makers and transporters of raw and finished products. If either the Industrials or the Transports weren’t confirming the direction of the other, then it was a warning that conditions might be about to change. And if both were headed substantially south, then the economy was likely to follow.

Simple enough, but a lot of trees have been felled trying to define when a Dow Theory trend change occurs. Dow and his followers, William Peter Hamilton and Robert Rhea, took the general view that a trend change occurred when one or both indexes failed to take out a previous high or low in the prevailing trend and then set a significant low or high in a new direction, signaling a trend change.

Every trend change begins with a correction against the prevailing trend; a trend change occurs when that correction resumes after a break and produces a new extreme in the new direction. So here’s the basic pattern when a bull market turns to bear: a correction, then a rally that fails to eclipse the old highs by one or both indexes, followed by new lows (Dow always used daily closing prices to define new highs and lows). Turning from bear to bull, the pattern would be a rally, a correction that failed to produce new lows in one or both indexes, and then another rally that took out the closing highs of the last rally.

The sticking point among Dow Theorists has been how to define those significant corrections that can mark the start of a new trend. The general rule of thumb has been a one-third to two-thirds correction of the previous major move over a period of three weeks to three months; basically, an intermediate-term correction. The Dow Theorist would then look for a countermove of the same depth and duration, followed by a new extreme in the new direction, before calling a trend change. But what appears to be a hard and fast rule often falls apart in reality: How would you classify a correction that retraced a quarter of the previous move over four months, or retraced most of the previous move in two weeks? That’s where Dow Theory has often been subject to uncertainty, judgment calls and lack of unanimity.

A second issue in calling major turns is Dow’s idea of a “line,” which is an intermediate-term trading range in one or both indexes. A break out of those trading ranges by both indexes is also considered significant, particularly if the breakout is simultaneous.

A Dow Theory Trading System That Has Stood the Test of Time

So with all that as background, here’s the one Dow Theory definition that seems to have performed best over time, that of Martin Pring, author of the basic TA text “Technical Analysis Explained.” Pring focuses on time rather than price; he looks for a correction that retraces a third or more of the time of the previous move, looking for a minimum correction of one month. He ignores price entirely. He also wants both indexes to fail to make new highs or lows after that correction (no “negative divergences” where only one index fails to eclipse a previous extreme while the other one does), and then both indexes confirm the new trend with new closing extremes in the new direction.

With “lines,” Pring also looks for a trading range of at least a month.

How has this performed over time? Using the data in Pring’s book from the 1966 stock market top to the present, that approach has averaged about 13% a year, a few percentage points better than the market as a whole. Over time, that adds up, plus it also has the advantage of sidestepping the steep drawdowns of bear markets. Only once has Pring’s approach been really wrong; it sat out or was short a 32% rally in 1949. It has otherwise been right about 80% to 90% of the time, with losses of 1% to 7% when wrong.

So why does Pring’s system work so well? Here’s one possible explanation: Because trends tend to last longer than anyone anticipates, Pring’s system keeps investors on the right side of the major trend longer and avoids whipsaws. One of the criticisms of Dow Theory is that it is slow to recognize and acknowledge trend changes; that could also be viewed as a strength, as major trend changes are rare occurrences. There have been just three Dow Theory signals in recent years by this system: a buy signal in June 2003, a sell signal in January 2008, and a buy signal in July 2009. Load up on value stocks (like low PE and price to sales and high free cash flow) during bull markets, go fishing and sleep soundly at night. You’ll leave money on the table at the start and end of bull markets, but that sacrifice has paid off over time by keeping you on the side of the major trend as long as possible. Dow Theory follows the major trend; it doesn’t try to call tops or bottoms until some time after they’ve occurred.

If Pring’s system seems too complicated, it could be modified further. Instead of looking for a one-third to two-thirds time retracement of the previous move, simply look for a one-month correction, regardless of how deep it is (remember, use daily closing prices). Even one month could be vague, so call it 18 trading days, a number based on my own unscientific observation; that would eliminate judgment calls entirely. Here’s how that would work when going from bull to bear: New highs, a one-month or longer correction, a one-month or longer rally, and then new closing lows by both indexes. Reverse the process when going from bear to bull. The indexes do not have to do this at the same time; one will often complete a trend change weeks or months ahead of the other.

One example: A few Dow Theorists had the 2010 correction as a Dow Theory sell signal, but Pring did not. Here’s why: The June rally in the Dow and Transports was less than two weeks long, far too short to qualify as an intermediate-term move, so the low in the initial leg down didn’t occur until early July (see charts below). The one-month rally that occurred off the early July low was long enough to qualify as an intermediate-term move, but as the indexes did not go on to set new lows in the August pullback, there was no sell signal. The charts also show why Pring’s system works so well: If you’d called the late June sell-off a Dow Theory sell signal, you would have sold out at 9800 in late June and bought back at 10,700 in late September.

Dow Jones Industrial Average 2010

Dow Jones Industrial Average 2010

Dow Transports 2010

Dow Transports 2010

And finally, one last pet peeve. Dow Theory has been shown to work on much shorter time frames (although perhaps not as consistently), but that doesn’t make it Dow Theory. Every week some article appears in the financial press about how important it is to Dow Theory for both indexes to eclipse X level that was set on a certain date. That’s not Dow Theory, which is about the slow transition of the stock market — and the economy — from bull to bear and bear to bull.

Paul Shread is a Chartered Market Technician and co-author of the book “Dow Theory Unplugged: Charles Dow’s Original Editorials and Their Relevance Today” from W&A Publishing.