Zigzag similarity module (Pattern Recognition software)


This is an experimental module. It analyzes some zigzag and finds similar patterns of that same zigzag indicator in the past. 

Look at this chart:


These are the steps to do:

1) set the initial value of zigzag critical change to 5%. The program varies this value analyzing different zigzags with different critical change parameters;

2) set the order (the amount of turning points to be analyzed);

3) choose zigzags from the list that provide the most similar patterns in the past for these last four turning points. In our example the program has found the most similar pattern in the year 2000.

 Vary the order; you will be able to find many interesting things. For example, if you set the order = 7, the program will find you zigzag that provides the best coincidence with the last seven turning points (the most recent turning points). Zigzag with the best coincidence  for these 7 turning points is shown in the first line of the list. In our example, it is the zigzag with the critical change value of 2.4%, and the most similar pattern corresponding to our 7 turning points took place in the year 2004:

In brief, you give the program two things: a starting critical change value for zigzag (the program is playing with this number, and what you get at the end may be some different number, close to your initial one) and the order of turning points (this value is fixed). When zigzag with the best coincidence of turning points is found, you have a hint for the next possible market move.

However, sometimes zigzag with the best coincidence of several  latest turning points does not look good outside these points, and it is risky to follow the hints provided by such a zigzag. To help you in this situation, the program analyzes a bunch of zigzags with different critical change values. We recommend to browse them all and find zigzag that provides the most similar pattern in the past together with good general appearance:

For our example, I believe that zigzag with critical change 3.3% fits better.

When you work with this module, I recommend to vary order parameter and preview all zigzags with different critical change parameter one by one. There are no formal criteria yet that allow to evaluate what zigzag is better, so the only thing that we can use now is visual fitness.

In "Options" tab you can define the range of varying of zigzag critical change values as well as type of the algorithm used for zigzag variations (Normal, Draft, etc.; it depends on the number of analyzed zigzags):


You can also play with "Force" parameter:

It se's  the last turning point on the latest highest high or lowest low even if this turning point is not confirmed by opposite movement. It works this way: "if we assume that this highest high is a real turning point, then ...".

We search for the analogy in the past; when it is found, it can serve as some hint regarding the future price movement.