Greetings !
Recently, a whole horde of trading systems have appeared that use AI
Such systems will always be closed to us; this is beneficial to the manufacturer.
Thus, you will never be able to change the logic inside, only make minor adjustments to the algorithm.
Moreover, all systems that use AI are usually very heavily over-optimized.
How to determine whether it is AI or not? Just look into those eyes !
In theory, such systems should not use indicators, only a Neural Network.
Such a neural network is extremely difficult to optimize and train, because This requires huge PC resources.
For example, according to statistics, the complete history for the last 3 years for gold takes about 3GB for 1 processor thread.
Thus, we get about 48GB of RAM for 16 threads.
And this is only 3 years of historical quotes.
The neural network should be optimized over a period of 6-8 years, or better yet, all 20 years.
This will require about 80-100GB of RAM just to store all historical ticks in the MT5 memory.
These are huge resources, especially in terms of time. This will take forever and a huge electric bill.
Therefore, many people use simple indicators inside and do not show them in the system parameters, pretending that it is AI.
As a rule, such systems already in the very name deceive their end user, because... This is not artificial intelligence at all.
Our systems never use AI, we use forward optimization.
Most systems on the market are over-optimized, it's like walking on the edge of a knife.
If the market changes, everything immediately collapses and there are huge losses in productivity and deposits.
So be careful when you see the word AI in the title or description.
For MT5 this does not work and will not work, because... it is simply impossible to train, the history and quotes are too huge for such manipulations.
For all our systems we use Forward Optimization ( ZENVO DIVO RAZOR GOLD WAY ).
This is a preliminary optimization until a certain date, for example until December 2022.
At the same time, we use an algorithm that does not require all ticks inside the candle.
We only need 4 points of the M1 candle, these are the opening of the candle, the closing and the maximum and minimum points of the candle.
This test mode is called 1OHLC. This allows you to significantly speed up any optimization and does not require huge resources.
But your algorithm inside the system must be able to work with this.
The settings you get with this optimization were generated over the last 12 years until the December 2022 mark, I call this the Foundation.
Foundation is the foundation for the settings that were obtained during optimization over 12 years of history.
If these settings show excellent performance after December 2022, this means that the algorithm is Really Working in the Market.
Such vulnerabilities are extremely difficult to find on the market, because... all simple algorithms usually do not work after foundation.
As practice shows, AI in this matter is a completely closed code and algorithm.
You will never know how it works inside and the worst thing is that you will not be able to optimize your process and settings on your own.
