Custom Search

Sometimes old trading ideas are the best ideas — if you can quantify them with modern analysis and testing procedures. Here, a "classic" chart pattern is defined mathematically and tested to see if it can produce profits.

BY AARON BEHLE AND MARK CONWAY

As an increasingly challenging market has weeded out traders over the past few years, many survivors in search of an edge are revisiting the works of the original technical analysis masters, including Richard Schabacker, J.M. Hurst, W.D. Gann and Harold M. Gartley.

Gartley wrote Profits in the Stock Market in 1935, and what makes the book striking is not that it shows how much technical analysis has advanced since then, but rather, how little it has changed. In many cases, "modern" patterns with catchy names are simply rehashes of price behavior observed long ago by people like Gartley.

One example is a pattern commonly known as the "butterfly," named for its resemblance to a pair of butterfly wings (see Figure 1, right). However, Gartley described this pattern in Profits in the Stock Market as the "Gartley 222," a reference to the page number on which the discussion occurred.

The Gartley 222 can be defined objectively by establishing specific proportions for the four price swings (XA, AB, ÂÑ and CD in Figure 1), or legs, that comprise the pattern, as well as by setting criteria to define the magnitude of the swing ("pivot") highs and lows — points À, Â and C.

Percentage relationships

In his book Profitable Patterns for Stock Trading, analyst Larry Pesavento used certain ratios to define the butterfly pattern, measuring each price swing (from peak-to-trough or trough-to-peak) as a certain percentage of a preceding price swing. Pesavento required these percentages to be Fibonacci ratios: 0.618, 0.786, 1.00, 1.27 and 1.618.

The problem is that if you disregard those patterns whose price swings are not proportional using precise Fibonacci ratios, the Gartley 222 pattern is quite rare. Using a "tolerance percentage" (T%) that expands the range of acceptable priceswing ratios produces more pattern examples and, thus, more trade opportunities. For example, if T% = 10 percent, segment AB (the second price swing) can be between 51.8 percent and 71.8 percent of segment XA (the first price swing), rather than exactly 61.8 percent.

Another criterion that can be applied to the pattern is the "strength" of the pivot points. For example, a pivot high has a strength of 3 when the three bars preceding the high and the three succeeding it are all lower than the high; a pivot high with four preceding and succeeding lower highs would have a strength of 4. Each pivot in the 222 pattern must meet this strength requirement. As the pivot strength increases, so does the length of the pattern and the likely duration of a trade based on it. However, the higher the pivot strength, the fewer patterns that will qualify for trading, and the longer those patterns will be.

Pivot strength can also be measured in percentage terms — e.g., a 2-percent swing from peak to trough on a 60- minute chart, or a 10 percent swing on a daily chart. These parameters should be appropriate to the time frame; price moves on an intraday chart will be proportionally smaller than those on daily or weekly charts. (With all these criteria, finding the pattern by scanning charts is difficult, at best. Accordingly, code for defining the pattern in both the TradeStation and Wealth-Lab analysis programs can be found at www.activetradermag.com/code.htm.) Using objective criteria for defining price patterns allows you to build a consistent strategy for trading them. We will use specific Gartley 222 pattern parameters to enter both long and short trades on different time frames. Back-testing on the Nasdaq 100 stocks over the past several years will provide an indication of the strategy's potential.