Discussion of high-frequency trading on MT5 - page 74
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 just want to say that some samples or source versions can be posted here for people to have substantive conversations. It won't help me (I'm not a coder), but it's nice for people, and the discussion will be more fun. Although ... it's just my opinion.
Sorry, but I'm not a pro in the field of AI, so I can not explain in detail the theoretical background, let alone the algorithms - maybe if I'm tortured ...((( But it, neuro training on patterns and predictions, at least in that engine, which I have, works and it's enough for me. And the arbitrage algorithm, which I promised to tell you about, when implemented with unrealistic abstraction, is reduced to the selection of the best price and combination of ready-made signals from neuro-advisors and in isolation from them have no sense at all.
As you like, I personally do not need it, I have chewed over this topic in detail, and those who superficially understand the logic of neural networks, such information with a high probability will only harm, confuse and draw into the abyss of eternal simulations.
I am only occasionally unobtrusively interested in the initial processing of vectors, before feeding them to neurons. If you only normalise the price in the hope that the multilayer Perspectron will give birth to a grail, then don't continue... It's as likely as if a hurricane flies over a landfill and picks up a Boeing by accident)))))))
Anyway I join Negoich the substantive conversation is best left alone.
As you like, I personally do not need it, I have chewed over this topic in detail, and those who superficially understand the logic of neural networks, such information with a high probability will only harm, confuse and draw into the abyss of eternal simulations.
I am only occasionally unobtrusively interested in the initial processing of vectors, before feeding them to neurons. If you only normalise the price in the hope that the multilayer Perspectron will give birth to a grail, then don't continue... It's as likely as if a hurricane flies over a landfill and picks up a Boeing by accident)))))))
In general I join Negoich the substantive conversation is best left alone.
But Boeing has been flying for a long time and does not need our proof of its feasibility. However you, and not only, armed with outdated technical information "for housewives" and spare parts from the junkyard, are trying to assemble it, and after the failure, you conclude that it is impossible in principle...))) That would be just ridiculous, but similar, unfortunately, occurs in reality, when talented and persistent loners in makeshift conditions, with limited opportunities, try to solve impossible tasks, and after years of failures fail, go down and continue their "scientific and technical" activities, already at the dump...(((
In general, I support Heroix as well as you and I believe that every trader should develop his own trading strategies and create trading robots by himself. But IMHO, you have to use the most convenient and advanced tool for that. Such a tool should contain a set of ready, professionally developed components and should allow traders not to "re-invent the wheel", but to solve their problems efficiently by focusing on specifics of trading and details of their strategy. I already have such a tool, and I'm not the only one... /*advertising removed*/
But Boeing has been flying for a long time and does not need our proof of its construction. But you and not only you, armed with outdated technical information "for housewives" and spare parts from the dump, are trying to build it, and after a failure, you conclude that it is impossible in principle...)) That would be just ridiculous, but similar, unfortunately, occurs in reality, when talented and persistent loners in makeshift conditions, with limited opportunities, try to solve impossible tasks, and after years of failures fail, go down and continue their "scientific and technical" activities, already at the dump...(((
In fact, I support Heroix as much as you do and I believe that every trader should implement his own trading strategies and create trading robots of his own using them. But IMHO, you have to use the most convenient and advanced tool for that. Such a tool should contain a set of ready, professionally developed components and should allow traders not to "re-invent the wheel", but to solve their problems efficiently by focusing on specifics of trading and details of their strategy. I already have such a tool, and I'm not the only one... /*advertising removed*/.
From ready-made for NS, I've used STATICTICA, NeuroSolutions, Neuroshell and a lot of other packages, and also wrote custom stuff in C++. I was very fond of it at one time, and for good reason. I've made 100500 of my own configurations of the architecture and teaching methods. I simulated and visualized vectors in the process of training 5000 neuron sets. I can reasonably assert that custom hierarchical, multi-level NS is the future. You shouldn't be so harsh, belittling the level of technology I've used in this regard. We can if you want to intellectually pohichtovat on this topic, maybe really suddenly it turns out that your software, this is what is looking for and can not find the company Google, then happily I reversed an opinion))))
First, tell me about network architecture(s) and learning algorithms, what is it? Multilayer Perspectrons, radial, Kohonen, probabilistic, Hopfield, etc... maybe your exclusive configuration and learning algorithm for it. For example I did up to 5 levels of hierarchy, when network of ~100 neurons is combined into a block that recognizes and reflects on its input data context, each block is trained individually, there are many blocks, you can train them in infinite number of ways, sometimes you get unexpected results, at first you get excited, but gradually you get used to it and try to build some META algorithm, that gives more predictable results, then you get stuck with advanced multilevel filtering, which in turn leads to unexpected results.
In general I'm not interested in black boxes, as I'm sure many others here are. If you say that your software is not homemade and not for housewives, then explain why so.
This is a forum for programmers, do not forget)))) I for example do not like to test "something" not knowing what to expect from it and how it is structured roughly, and the purchase is generally out of the question.
I have used STATICTICA, NeuroSolutions, Neuroshell and many other ready-made NS packages, and I have also written custom ideas in C++ myself. I was very fond of it at one time, and for good reason. I've made 100500 of my own configurations of the architecture and teaching methods. I simulated and visualized vectors in the process of training 5000 neuron sets. I can reasonably assert that custom hierarchical, multi-level NS is the future. You shouldn't be so harsh, belittling the level of technology I've used in this regard. We can if you want to intellectually pohichtovat on this topic, maybe really suddenly it turns out that your software, this is what is looking for and can not find the company Google, then happily I reversed an opinion))))
First, tell me about network architecture(s) and learning algorithms, what is it? Multilayer Perspectrons, radial, Kohonen, probabilistic, Hopfield, etc... maybe your exclusive configuration and learning algorithm for it. For example I did up to 5 levels of hierarchy, when network of ~100 neurons is combined into a block that recognizes and reflects on its input data context, each block is trained individually, there are many blocks, you can train them in infinite number of ways, sometimes you get unexpected results, at first you get excited, but gradually you get used to it and try to build some META algorithm, that gives more predictable results, and then you get stuck with advanced multilevel filtering, which in turn leads to unexpected results.
In general I'm not interested in black boxes, as I'm sure many others here are. If you say that your software is not homemade and not for housewives, then explain why so.
This is a forum for programmers, do not forget)))) For example I don't like to test "something" without knowing what to expect from it and how it is set up approximately, while the purchase is out of the question.
In my previous comment I said that I'm not going to discuss about AI, because I do not consider myself a professional in this field, although I read, probably the same as you "bearded", but publicly available information... ala 20-40 years later and dabbled in all sorts of universal entertainment and mass-training softinami... I advise you too to stop wasting your time and believe that most of the public and free stuff is "last year's snow"...))
In my previous comment I said that I won't discuss about AI because I don't consider myself a professional in this field, although I probably read the same "bearded" but publicly available information... ala 20-40 years later and dabbled in various, universally entertaining and mass-training soft utilities... I advise you too to stop wasting your time and believe that most of the public and free stuff is "last year's snow"...))
I didn't mention info, so you don't have enough data to judge "beardiness")) Well, about custom C++ sets on the topic of NS I'm completely silent, it is unlikely that even Google tested exactly what I tested, because the variants of such sets are infinite and coincide exactly in the architecture of NS, training, etc. is unreal. In general, I don't understand what beardiness we're talking about. Maybe a scalar product is bearded? Or a sigmoid? Ahhhh I get it, bearded is probably a perseptron, single and multi-layer. Although no... probably just bearded whatever I use, not bearded whatever you use. Well, then it's ridiculous.
I'm not trying to engage you in a theoretical discussion on neurons and engage in pissing measures in that context. But all your words will turn out to be nonsense if you don't specify NOT rotated algorithms, at least as a class, excluding commercial tricks.
How do I know that your EA Generator is even a neural network? Maybe it's MAs. Anyway, what's the fun of selling a cat in a sack with no demo and no hint of an algorithm. You now have to either rehabilitate yourself in a marketing sense, or change the brand. You have completely confused your target audience dear. It is also possible to divorce, but it is a delicate art. So far you're not good at it, sorry.
I did not mention about info, so you do not have enough data to judge about "beardiness"))) Well, about custom C++ sets on NS topic I'm even silent, it is unlikely that even Google tested exactly what I tested, because the variants of such sets are infinite and coincide exactly in the architecture of NS, training and so on is unreal. In general, I don't understand what beardiness we're talking about. Maybe a scalar product is bearded? Or a sigmoid? Ahhhh I get it, bearded is probably a perseptron, single and multi-layer. Although no... probably just bearded whatever I used, not bearded whatever you used. Well, then it's ridiculous.
I'm not trying to engage you in a theoretical discussion on neurons and engage in pissing measures in that context. But all your words will turn out to be nonsense if you don't specify NOT rotated algorithms, at least as a class, excluding commercial tricks.
How do I know that your EA Generator is even a neural network? Maybe it's MAs. Anyway, what's the fun of selling a cat in a sack with no demo and no hint of an algorithm. You now have to either rehabilitate yourself in a marketing sense, or change the brand. You have completely confused your target audience dear. You can also divorce, but it is a delicate art. So far you are not good at it, sorry.
I'm not selling anything, I'm not baiting anyone, I do not give in to baiting, I only teach trading robots, may be live ones, but the engine is not ready yet, so I can not help yet - no offense ...)) _http://www.rbcdaily.ru/cnews/562949980476018
lohhft & Alex_Bondar, if you don't immediately stop advertising a paid product sold as a completely unsafe exe file, I will ban you both forever.
Everything about this product has been removed from the thread.