OpenCl and the tools for it. Reviews and impressions. - page 19
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I did it easier: I look for ready-made solutions (wavelets, neuronetworks, etc.) I create a .dll (do not forget that it is convenient to have an interface with forms, with keybars and bootstraps :)) When I work with Mt4 or Mt5 it is very convenient and saves time for checking ideas, instead of porting codes or reading scientific literature
Renat, I understand that your company has to make money, and not only fulfilling "whims of users", but still I will stick to my opinion: a person familiar with programming will find it difficult to write complex programs in mql5 because of restrictions resulting in the inability to directly port from C++ (no work with multidimensional arrays, no work with pointers). A person who is not familiar with programming at all (doctor, humanitarian ....) will not be able to learn mql5 independently, even in terms of editing code created with the help of mql5 wizard. For the first ones, the desire to program in mql5 will help to overcome difficulties, but those who are not familiar with programming need ready-made implementations in the form of libraries with detailed comments on each part of the code. I hope that the mql5 kodobase will soon contain all the necessary libraries for scientific and "near-scientific" market research, then the computing power of MT5 will be in demand.
good luck!
Renat, I understand perfectly well that your company must make money, not only fulfill "whims of users", but still I'll stick to my opinion: a person familiar with programming will have hard time writing complex programs in mql5 because of limitations, resulting in impossibility of direct porting to C++ (no work with multidimensional arrays, no work with pointers)...
A person who is not familiar with programming (doctor, scholar ....) cannot study mql5 without assistance, even in terms of editing code created with the help of mql5 Wizard...
I don't know about porting for humanities people or doctors, but for trading, mql5 is pretty solidly shaped. For me, switching to mql5 didn't cause any problems. I have some not quite familiar shorthand, but that's only because I mostly write in Java.
The documentation for dummies may cause some problems because many of the items do not have ready-made examples. But again, for me it did not cause any difficulties, because I browsed CodeBase, read articles and understood it. I.e. additional troubleshooting in this case is only for my good.
Well, if there is OOP, the libraries of ready-made classes will gradually grow. It is unlikely to happen tomorrow, but it will take some time before everyone finally upgrades hardware, only because this hardware tends to age faster than software, and moves to MT5. In other words, the situation will only get better as they upgrade.
...
Well, as long as there is OOP, the libraries of ready-made classes will gradually grow. It will unlikely happen tomorrow, but it will take some time until everyone finally upgrades hardware, only because this hardware tends to age faster than software, and moves to MT5. In other words, the situation will only get better as they upgrade.
It is clear that they will move anyway, they are not going anywhere. The question is different. They will become more divided in the trading community into professionals and artisans. No longer will be the "middle class", who secretly makes things as they can.
They will either order for money or trade by hand. And I am sure that such stratification will not add popularity to the platform in the long run.
It is clear that they will switch anyway, they will not go anywhere. The question is different. The traders' community will become more stratified between the professionals and the humanitarians.
There will no longer be the "middle class", who are now quietly making crafts as best they can.
They will either order for money or trade in pens. And I am sure this stratification will not add to the popularity of the platform in the long run.
And to hell with stratification. It does not affect the trading results, so it does not matter.
The average traders will not go anywhere. And even if they do it will not matter.
The popularity of the platform grows when it is developed and refined for end users, i.e. traders, rather than for developers or gamer hardware owners.
1... And that's it! You're gonna want it soon. Where's the money? :)
2. Nah, I don't think so. I really think it's a matter of overshoot power. Try either reducing the vicinity or increasing the discreteness. So that there's a couple or three million left (not 214).
I've tried it both ways. I cannot reduce the number of input parameters to disable genetics.
As they say, if the mountain won't come to Mohammed...
I would have to check 5 input parameters at a time (11^5 = 161061 passes). I need 8 inputs with a sample rate of 11 in each. Then three to disconnect and three to reconnect, and again to wander around. That's also a way out. If we don't check forward stability at all and wait until developers deign to condescend to trackers, the result will be much more pitiful. So we will have to take advantage of what we have, albeit with some twists, i.e. we will split the process of additional forward testing into two stages.
In trading it is better to overdo it than underdo it.
I've tried it both ways. I can't reduce the number of input parameters to disable genetics, but I can't do it.
As they say, if the mountain won't come to Mohammed...
I would have to check 5 input parameters at a time (11^5 = 161061 passes). I need 8 inputs with a sample rate of 11 in each. Then three to disconnect and three to reconnect, and again to wander around. That's also a way out. If we don't check forward stability at all and wait until developers deign to condescend to trackers, the result will be much more pitiful. So we will have to use what we have, albeit with some inconsistencies, i.e. break the process of additional forward testing into two stages.
In trading, it is better to overdo it than underdo it.
My thinking in such cases is as follows.
I have a decision space (area) and need to check it. I.e. whether in a bounded vicinity there is a maximal value of parameters leading to sinking. At the same time overshooting is impossible due to technical reasons. Problem.
Resource - there is a fanatical genetic algorithm that can only look for maxima.
Solution : I reverse the trade - I look for maxima on -f(x1,x2,...xn). If many combinations of parameters leading to positive -f(...) are found, it means that the initial strategy has a hole, and our GA-fanatic has successfully found the largest holes.
// Yura. Admit it, you are too lazy to just solve the problem. Fighting is more interesting... ;)
My thinking in such cases is as follows.
I have a solution space (domain), I need to check if it is not empty. That is, there is no maximum value of parameters in the bounded vicinity, which leads to a drain. At the same time overshooting is impossible due to technical reasons. Problem.
Resource - there is a fanatical genetic algorithm that can only look for maxima.
Solution : I reverse the trade - I look for maxima on -f(x1,x2,...xn). If many combinations of parameters leading to positive -f(...) are found, it means that the initial strategy is broken, and the largest holes have been successfully identified by our GA-fanatic.
It is also a solution, but only if the hole minima are in the neighborhood of an extremum. And GA does indeed often hang around the vicinity. I will have to write an external program to search for holes in the vicinity based on all optimization results. So far I cannot think of any suitable idea to solve this task with MT*.
So far I don't see any direct correlation between the number -f(x1,x2,...xn) and the quality of forward tests. This will have to be checked further.
Yura. Well, confess, you are lazy just to solve the problem. It is more interesting to fight. ;)
And it's useless to argue, especially with the greatest revolutionary of Technical Progress of all Times and Nations. He came up with the postulates that:
1. The main task of the gigantic network is to SPEAK. (All other tasks are either secondary or not worthy of a giant network. ) Not very smart and obviously technologically backward traders somehow thought that a computer network must solve their necessary tasks, and even as effectively as possible? Dream on).
2. Genetic algorithm is an excessive whim, a relic of the past, and all tasks in the XXI century should be solved by a complete brute force, but limited to a million passes.
In other words, thousands of super cool and sophisticated excavators are united into one team, and a children's shovel for a sandbox is attached to all of them instead of a bucket, thus making a technical revolution.
It is useless to fight with such figures. The more excavators are working, the more they pretend to do something. And if the work is boiling, the mankind should be grateful. After all, thousands of excavators in one team simultaneously and harmoniously digging the same trench a handful at a time, and it is a spectacle that fascinates the progressive minds. And here the ungrateful mankind instead of bowing down to the ground to the innovator, also expresses its dissatisfaction for some reason.
And it is useless to fight, especially with the Greatest Revolutionary of Technical Progress of All Times and Nations. He has, after all, come up with the postulates that:
And you are the Greatest Bending Sticks of All Times and All Nations.