50.00

46.00

42.00

38.00

34.00

30.00

26.00

22.00

19.00

Part III: Bollinger Bands on Their Own dates: 01/12/00-01/08/01 box: 0 rev: 3 last price: 31.88

. . .X.

X X

□ ~

3X □

□X

x xp

X. X X

X X

□x

X □X

Ox . .

XQX X

7

" ~X9 -

50.00

46.00

42.00

38.00

34.00

30.00

26.00

22.00

1-19.00

Figure 11.8 Modern point-and-figure chart, IBM, one year.

might represent a quarter point or a half point. At higher prices the box size is increased so that each square on the grid, or box, might now denote a half point or a full point. For a $10 stock each box might be a point, or for an $80 stock each box might represent a point and a half. The ChartCraft approach, originally developed by Abe Cohen, is the most widely accepted. Table 11.1 presents the ChartCraft box-size recommendations.

In order to switch from a negative swing to a positive swing, using the ChartCraft system, a three-box threshold is employed. This allows for a small enough box size that vital detail is not lost

Chapter 11: Five-Point Patterns

Table 11.1 ChartCraft Recommended Box Sizes for Stocks

at the same time a large enough filter is employed. So with the ChartCraft method, for a $10 stock a 1 A-point reversal is needed to change swing direction * 3. For a $70 stock a 6-point reversal is required to change swing direction (2 * 3).

The main problem with this approach is variability-abrupt, large changes at transition prices. For example, a $19 stock, with its half-point boxes, reverses swings with a 1 -point move, whereas a $20 stock, with its full-point boxes, requires a 3-point swing to reverse. Normally reversals get smaller in percentage terms as price rises, but there are places where higher prices beget higher percentage reversal values due to transitions in box sizes. Using our example, a $19 stock uses a 7.8 percent reversal, whereas a $20 stock uses a 15 percent reversal. You have to rally all the way to $40 before you get back to a 7.5 percent reversal.

A simple method of smoothly specifying box size, Bollinger Boxes, was developed in order to avoid the problems caused by the traditional rules. To create Bollinger Boxes, all of the historical methods used to specify box size from Wheelan to Cohen were tabulated. Then the rule sets were plotted, with price on the x axis and percent box size on the axis. For each set of rules this process produced a stepped line to which a curve was fit (Figure 11.9). The formula for that curve was noted and the procedure repeated for each known box-size methodology. These procedures revealed an ideal box size that can be simplified to 17 percent of the square root of the most recent price (see Table 11.2).

As a control, the square root rule (SRR) was used. The earliest mention of the SRR is in Burton Cranes 1959 book, The Sophisticated Investor, where he cites Fred Macauleys writings in the New York Times-Annalist magazine as the original source. The SRR suggests that volatility is a function of the square root of price;

Part III: Bollinger Bands on Their Own

Table 11.2 Sample Box Sizes Using Simplified Bollinger Boxes (0.17*lastA0.5)

Price

Reversal

$4.5

$8 $18 $69

8% 6% 4% 2%

for an equal move in the market, stocks will rally such that the square roots of their initial prices change by a similar amount. This rule produces large percentage gains for low-priced issues and large point gains for high-priced issues. From this perspective, low-price stocks are more volatile than high-price stocks. This is an intuitively correct idea. On average we expect that low-price stocks will experience greater percentage increases and decreases than high-price stocks.

There was relatively little variation between the historical methods that were plotted, and the fits to the SRR were near perfect.

25.0% w -

20.0%

-CCBox%

- Simple BB %

5.0%

0.0%

10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190

Price in Points

Figure 11.9 Curve fit for Cohens point-and-figure box-size rules.