Very cool experiment.
Hope to see more on this Darks, thank you for posting your great stuff.
Here, i'll repost Mladen's post for your from FF regarding his experiment too. Hope it helps your experiments Darks
mladen
Some more fooling around
___________________________
There is an eternal need to filter out the false signals. Numerous methods have been tested and tried and some are working better and some are working worse. Some can be applied to more things some are specialized.
One method used is the "stepping" :
- step averages
- step stochastic
- step rsi ...
just to name some that were made to work using that method. But all those have one thing in common : you can not apply the same method used for average to stochastic, or rsi, or vice versa. They are, more or less very specialized (methods wise and input parameters values wise too)
Here is a new method that does not need anything of that. It requires only 2 parameters :
- the window (which is the window that is taken for samples that are used to determine the step)
- the sensitivity (the bigger the number, the less sensitive it is - smaller sensitivity values are producing values closer to original values). Less than 0 or equal to 0 means no "stepping"
There are two methods :
- one is "crude" - very step like
- one (the default) is producing much smoother values
Used wpr, rsi, atr, standard deviation, momentum, average (the 4 basic types) just as a test for using same parameters for very different range of values and types of indicators. Example bellow using same parameters (except for the indicator type, of course)
