Forex Trading System- The measured moved

OVERVIEW

The concept of a measured move is simple in theory. Essentially, it is the estimate of thelength of a price movement in one direction following a specific price pattern or chartevent. However, in practice numerous factors influence this phenomenon: the underly-ing financial security being analyzed, the box size and reversal amount used to generatethe swing data, the chart pattern, the number of waves in the pattern, and the locationof the key inflection points in the swing data.

In Figure 23.1, we see a bull wave of 11 pips followed by a bear wave of 7 pips andthen another (incomplete) bull wave.

The goal of the measured move principle is to estimate the height of the third wave
based on the percentage of retracement between the first and second waves or someother mathematical relationship. The measured move concept can also be applied tomore complex price formations and multiwave cycles (which we will examine later).

TESTING APPROACH

In this study, we take a purely statistical approach to estimate the measured move,drawing from the 7,000,000+ quotes in our historical currency database. Raw streamingtick data was converted to swing data using our standard reversal algorithm with a con-stant box size of one pip in the EURUSD currency pair.

In the initial run of this analysis, we limit ourselves to three input variables:

1.Reversal amount in the swing algorithm.2.Height of the bull wave.3.Height of the bear wave.
CLUSTER CHARTSOur first step is to examine the data by displaying it in its most pristine form using clus-ter charts. In Figures 23.2, 23.3, and 23.4, the x-axis (the independent variable) repre-sents the percent of retracement between wave 1 and wave 2, or
The y-axis represents the dependent variable and is calculated as the percent of re-
tracement between wave 3 and wave 1.

To the untrained eye, cluster diagrams may appear confusing. This is true also tothe trained eye sometimes. There appears to be no clear linear relationship between thedependent and independent variables. However, the crucial point in the three clustercharts is to note how the cluster patterns change when the reversal amount is increased(the examples use three-, five-, and nine-box reversal amounts).

AVERAGE RETRACEMENT

To coerce the cluster charts into displaying more revealing information, we averagethe y-axis values for each corresponding integer x-axis value. (See Figures 23.5through 23.7.)
By plotting only the y-axis mean values, we have eliminated the original confusiondisplayed in the corresponding cluster charts. The straight diagonal line in each linechart represents the results of an ordinary least squares (OLS) linear regression. (SeeTable 23.1.)
To forecast the height of the third wave with this information, we use the followinglinear formula:
y = Ax + B
wherex =wave 2/wave 1
y =wave 3/wave 1
A =slope
B =intercept

There is one drawback to using the average y-axis mean values as opposed to usingall the data points (as in the cluster diagrams). When using all the cluster data points,the coefficient of correlation drops below 50 percent in all three cases, although the es-timated slopes and intercepts are extremely close to those in the table using just the av-erage values.

DECILE DIVISIONS

Tables 23.2, 23.3, and 23.4 were calculated in order to enhance the validity of theforecast of the measured move by a statistical method referred to as partitioning or
grouping. Each table corresponds to a different reversal amount. The x-axis (wave 2/wave 1) has been divided into 10 partitions called deciles. For example, the fourth decile(40 percent) means that the partition includes x-axis values ranging from 30% <>

PRACTICAL EXAMPLES

Example 1: Three-Box Reversal Amount, 11-Pip Bull Wave, 7-Pip Bear WaveIn this example, a three-box reversal amount was used to create the swing data and wehave isolated an 11-pip bull wave (the impulse wave) followed by a 7-pip bear wave (thecorrective or retracement wave). Our objective is to estimate the height of the thirdwave and its likelihood of occurrence. (See Figure 23.8.)
Step 1:Divide the height of wave 2 by the height of wave 1:

Step 2:Multiply by 100:
0.6363 +100 = 63.63%
Step 3:Round up to the nearest multiple of 10:
63.63 becomes 70
Step 4:Locate the corresponding row (70) in the appropriate reversal amount table
(Table 23.2, three-box):
Row 70
65.33
20.53
17,922
Step 5:Multiply the height of wave 1 by the measured move percentage in that row
(y-value, column 2):
11 pips +65.33% = 7.1863 pips
The estimated length of wave 3 (the measured move) is 7.19 pips.Given these input parameters, we estimate that the third wave will rally to a point of
11.19 pips. (See Figure 23.9.)
Next we must calculate the level of confidence that this point will be reached.
Step 6: Locate the standard deviation table, the third column in the corresponding
row, or:

Step 7:Multiply the height of wave 1 times the percentage of standard deviation:
11 pips +20.53% = 2.2583 pips
Step 8:Add the calculated deviation value to the forecast estimate to obtain the up-
per confidence boundary:
Upper confidence boundary = 11.19 pips + 2.26 pips = 13.45 pips
Step 9:Subtract the calculated deviation value from the forecast estimate to obtain
the lower confidence boundary:
Lower confidence boundary = 11.19 pips – 2.26 pips = 8.93 pips
Placing upper and lower boundaries above and below the forecast estimate asshown in Figure 23.10 allows us to make the following statement about this particularmeasured move:
There is a 68 percent likelihood using a normal distribution that the peak of thethird wave will fall between 13.45 pips and 8.93 pips.Example 2: Five-Box Reversal Amount, 16-Pip Bull Wave, 8-Pip Bear WaveCalculate retracement of wave 2:

Retrieve corresponding row for 50 percent from the five-box reversal table:
Row 50
57.69
16.77
2,828
Multiply height of wave 1 times measured move percentage:
16 pips 57.69% = 9.2304 pips
Multiply height of wave 1 times standard deviation percentage:
16 pips +16.77% = 2.6832 pips
The results are shown in Figure 23.11.There is a 68 percent likelihood that the peak of the third wave will fall between
19.91 pips and 14.55 pips.Example 3: Nine-Box Reversal Amount, 14-Pip Bull Wave, 10-Pip Bear WaveCalculate retracement of wave 2:

Round up (71.42 percent becomes 80 percent) and retrieve corresponding row from
appropriate table:
Row 80
87.14
31.02
2,057
Multiply height of wave 1 times measured move percentage:
14 pips 87.14% = 12.1996 pips
Multiply height of wave 1 times standard deviation percentage:
14 pips +31.02% = 4.3428 pips
The results are shown in Figure 23.12.There is a 68 percent likelihood that the peak of the third wave will fall between
20.54 pips and 11.86 pips.
Abs oslutesM omns
145
FIGURE 23.12dhW8.oE 1 dacC8WcE GCcO TsHPuEHbE rsnHuWRCEa

CONCLUSION

From this initial study of the measured move, we have learned that it is possible to esti-mate the height of the third wave in a three-wave cycle when the heights of the first twowaves are known. However, the magnitude of the standard deviation of the estimate of-ten lessens the value of a discrete forecast. In subsequent chapters, we attempt to iso-late chart patterns and multiple-wave formations where the magnitude of the standarddeviation is noticeably less, thus enhancing the efficiency of a discrete forecast.