There are 8 "fuzzy (verb) rule" indicators with genetic algorithm (GA) optimization in the add-on, FuzzyGA1, FuzzyGA2, ... , FuzzyGA8. FuzzyGA1 handles "1 segment" rules like "close rises" or "open drops sharply". FuzzyGA2 handles "2 segment" rules like "close rises, then drops sharply". FuzzyGA3 is for "3 segment" rules like "open rises, then drops, then remains steady", etc. The internal fuzzy logic engine parameters are exposed to the user and can be optimized with the help of GA.

The FuzzyGA parameters differ from the Fuzzy parameters in that they allow us to set certain internal parameters of the fuzzy logic engine. For that reason they may possibly provide better results when used with the genetic algorithm optimizer in the NeuroShell Trader Professional or DayTrader Professional. On the other hand, these internal fuzzy engine parameters are too difficult to set by hand, so the Fuzzy GA set of indicators should not be used without the optimizer.

The Fuzzy Indicator output ranges from zero to one showing how close the current input time series matches your pattern. One means a perfect match; zero means no match at all.

Here are all of the parameters for FuzzyGA Indicators with recommended range settings for the optimizer. The value of N is the number of the FuzzyGA Indicator (e.g. when looking at FuzzyGA2 then N=2 intervals):

Parameters

Input Time Series - the time series under consideration. It can be any price series like Close or any technical indicator like RSI, MACD, etc. There is only one input time series in Fuzzy Indicators.

Code1, Code2, …, CodeN - the search pattern rules. You define the search pattern rules as a sequence of numeric values. Each value represents a specific rule applied to a segment. The value-to-rule translation is as follows:

2 - rises sharply

1 - rises

0 - remains steady

-1 - drops

-2 - drops sharply

Code1 corresponds to the earliest segment, CodeN - to the most recent segment in the input time series.

For example, you want to find the pattern verbally described as "the close rises sharply, then stays steady, then drops sharply". Using the coding scheme above such pattern will be coded as Code1=2, Code2=0, Code3=-2.

If using FuzzyGA Indicators in a trading strategy, let the genetic optimizer find the best rules. Set the range of variation of Code parameters to -2 … 2.

Segment Size - the number of bars in the segment. All segments have the same number of bars. The segments share their edge bars. In other words, the last bar in a previous segment is also the first bar in the segment immediately following that segment. Default segment size value is 5. Minimum setting is 2. If using FuzzyGA Indicators in a trading strategy, let the genetic optimizer find the best segment size. Possible range of variation could be from 2 to 100 bars.

Max Change - the absolute value of the maximum change of the time series in the segment. The maximum change is needed to define the highest "rise sharply" rule or absolute value of the "drops sharply" rule. For example, if Max Change=3 (default value) and Segment Size=5 bars, then the strongest "rise sharply" rule occurs if the 5-bar time series rise is 3 or more points. If the rise is in between 0 and 3 points, then that rise will be classified as different degrees of the "rise sharply", "rise", or "stays steady" rules depending on the actual value.

The same logic applies to drops. Given numbers above, the strongest "drops sharply" rule occurs if the 5-bar time series drop is 3 or more points. If the drop is in between 0 and 3 points, then that drop will be classified as different degrees of  "drop sharply", "drop", or "stays steady" rules depending on the actual drop value.

As another example, let’s examine how a change of 3 in the time series will be considered in a Fuzzy Indicator with two different values of Max Change. If Max Change is set to 3, then an actual change of 3 will be considered the strongest form of "rises sharply". On the other hand, if Max Change is set to 6, then an actual change of 3 in the time series is at most a weak "rises sharply", however it will be considered a strong "rises". This leads us to the following important note:

The Max Change parameter setting is critical to the FuzzyGA Indicator output. Its setting defines what you mean by the "rise sharply", "rise", etc. rules. It should be set to a value most likely anticipated when considering the time series change on the Segment Size bars time scale across the WHOLE time series. Every time you change the Segment Size parameter, the Max Change should be adjusted accordingly.

The easiest way to set the Max Change is to use the FuzzyGA Indicators in a trading strategy with parameter optimization. Let the genetic optimizer find the best Max Change value.

Param1, Param2, Param3 - internal fuzzy logic engine parameters.

Param1: default value=0.2. Possible range of variation: 0.05 - 0.4.

Param2: default value=0.25. Possible range of variation: 0.1 - 0.4.

Param3: default value=5. Possible range of variation: 2 - 50.

Use the genetic optimizer to fine-tune these parameters. We know of no way to set them manually.

Note: Param1, Param2, and Param3 are parameters that require extensive knowledge of how fuzzy logic engines operate internally. As such, a complete description of them is far beyond the scope of this documentation, and would only serve to hamper the understanding of FuzzyGA Indicators for most of our users. However, for those with an interest in further knowledge of fuzzy logic, we recommend that you consult Kosko’s book "Fuzzy Thinking" (see the References section of this help file for a complete reference).  In Chapter 10 you will find a description of how fuzzy rules relate to fuzzy sets. These parameters define the location and shape of fuzzy set triangles.