FXLearning

As I know, the DML implementation of the Dr.J.Ehlers version published in the "Cybernetics..." is quite the same, maybe with the only exception of the 7 bars initialization code segment. I was able to implement it by using a linear compensation slope for Alpha with a length (in bars) that is a parameter for the user (the initialization period is an 'Input' variable).

The smaller the ITrend "Alpha" parameter (meaning the higher the cutt-off frequency), the wider the difference between the ITrend start value vs. the median price will be. This is the time segment where the initialization code is required.
I don't know the MetaTrader formula language so I cannot comment if their version is the same as Dr.Ehlers function.
"TheDML" is providing 2 versions for the ITrend: one based on the "Rocket Science...." algorithm (adaptive SMA by the CyberCycle Period) and the version explained in "Cybernetics..." which is based on an adaptive median filter algorithm (by the CyberCycle period).
The lag between the 2 version is quite different because of the different algorithms that were used. The latest is smoother and has less lag, but needs a longer initialization period.

What I perceive as the most beneficial advantage of this package is that the vast majority of the functions are ADAPTIVE.
The user can call any function with parameters that are indicator values instead of scalar values (fixed value numbers).
They accept even other DLL's function calls output.
This is one of the most versatile packages I know.
I hope the info I've prtovided was of use.

Regards
Guara