Forex Trading System- Pattern frequencies

OVERVIEW

The ability to identify various point and figure (P&F) patterns and to look up their rela-tive frequencies of occurrence can be advantageous when calculating possible subse-quent column values.

Again we will employ our 7,000,000+ closes database as our statistical sample. Wewill use a reversal amount of three boxes and a seven-column pattern. We will limit themaximum number of figures (Xs or Os) in each column to nine (that is, quantitiesgreater than nine are rounded down to nine). The minimum number of figures in eachcolumn is set at three because of the reversal algorithm mechanism. Thus there areseven possibilities for the number of Xs or Os in each column.

Each pattern consists of seven columns. This generates 823,543 possible pat-terns. In our initial computer tests, we discovered that the lateral congestion pat-terns like 3333333, 3343333, 3334333, and 3333433 dominated the top of thefrequency count. Therefore, we found it necessary to impose a few conditions to fil-ter these lateral patterns.

First, we mandated that the center column have the greatest number of figuresamong all the columns. This eliminates the redundancy of shifting the pattern onecolumn in either direction. Second, the fourth column must have a minimum of sixXs or Os. These two conditions reduced the number of possible patterns nearly byhalf to 470,596.

BOX SIZE = ONE PIP

In Table 17.1, patterns are ranked by frequency of occurrence.
Using the one-pip box size, the nine most frequently encountered P&F patterns withseven columns with our filtering constraints are shown in Figure 17.1.

INVERSE PATTERNS

While compiling these frequency counts, we also combined each unique pattern with itsconverse pattern (that is, we substituted Os for Xs and vice versa). For example, thetwo patterns shown in Figure 17.4 are both logged as 3347365.

USAGE

The tables and pattern diagrams in this chapter were compiled more or less as generalinformation. They can, however, be used to estimate the number of Xs or Os in the sub-sequent columns.For example, assume we are using a three-pip box size and a three-box reversalamount and we have a five-column pattern that consists of 33363 as in Figure 17.5.

Using the third frequency table (Table 17.3), we locate all occurrences of the pat-
tern 33363 in the first five columns. (See Table 17.4.)By summing the frequencies, we can calculate the percentages of likelihood for eachtwo-column sequel to the 33363 pattern: 33 = 36%, 34 = 34%, 44 = 18%, and 53 = 12%.