Machine learning in trading: theory, models, practice and algo-trading - page 1446
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
So vagabonds, I don't get it. Did I hire you to raise this branch, AAA?
Hi all :-) I haven't been here for a long time and again I see that the only good tool on the market is on the second page. How do you feel, how do you trade?
I've got good results, but my balance curve stubbornly refuses to go up, but I'm not a timid man.
I see that some people have got neural network software, which does not need any input data, just select neurons in the layer and voila, the grail is ready. If it is really a forward looking picture, it is quite interesting, but I want to upset that even with such a good balance curve it would be difficult to repeat it in a real account even if I have a bridge with MT4, and if I do not have it, I pity the owner of this program, he cannot keep up with automatic machines with his hands. I've never seen such a big crash with an automaton, or there's no connection, or I forgot to put a checkmark, or the fifth, or the tenth. I then some shit always begins to happen as soon as vosstanovlyuyutsya to the original dept, immediately breaking off what is it. I've got used to it, but I get the feeling that someone interferes with me when I'm trading in tact with the market. I have already got used to it, but I have a nagging feeling that someone is bothering me when I am trading in tune with the market. I understand that these might be coincidences and accidents, but they always appear at the moments when I am not expecting them. It's about the real difference between trading and trolling for a nice forward. Beautiful forward test is not half of the work, it is possible to say a quarter. For example, today I spent all day on my computer. I may have had to build a hundred or two models, but it was useless. The sadness :-( But I have taken a certain risk today. The models have been built and already loaded into automatic mode without any forward. As there is, I have optimized them to the very close of Friday and we'll see how this week will be. But I have a feeling, that there will be some work, because my approach is very tricky :-)
And by the way,Ivan Butko, what is the price of opium for the people these days? How much did you get it for?
You know, I was thinking, I would really like to take a software, customized for neural network trading, at least this one, which I'm talking about and put my data in it, it would be interesting to look at the result, what is the software itself and the data itself, how much they work? But alas, this is all a dream because Maxim, and most of those present, want a ton of them and where to get them if an average of 2-3 signals per day. 20 days of trading in a month, total 60 signals multiplied by a year turns out about 720-800 entries and that for the year, that's it ....
But ifIvan Butko will dig in the classifier, I'm ready to support informationally. I wonder what the beast is!!!!
And by the way,Ivan Butko, what is the price of opium for the people these days? How much did you get it for?
You know, I was thinking, I would really like to take a software, customized for neural network trading, at least this one, which I'm talking about and put my data in it, it would be interesting to look at the result, what is the software itself and the data itself, how much they work? But alas, this is all a dream because Maxim, and most of those present, want a ton of them and where to get them if an average of 2-3 signals per day. 20 days of trading in a month, total 60 signals multiplied by a year turns out about 720-800 records and that for the year, that's it ....
But ifIvan Butko will dig in the classifier, I'm ready to support informationally. I wonder what this beast is!!!!
Took it for 25,000.
I'm not talking about it yet. But if I can't make money on it, I'll tell you the secret.
At this point, I want to correct myself, I was wrong in my previous posts: the more neurons and layers, the more promising the result. Figured out how to apply them correctly, even if there is extra overtraining (fitting).
Besides, I want to inform you that even 3 neurons can train the network to trade positively for a few months (forward). Apparently, it (neuronet) really needs to be trained smart.
Took for 25 000.
While I do not distribute. But if I can not make money on it, I will reveal the secret.
At the moment I want to correct myself, I was wrong in previous posts: the more neurons and layers, the more promising the result. Figured out how to apply them correctly, even if there is extra overtraining (fitting).
Besides, I want to inform you that even 3 neurons can train the network to trade positively for a few months (forward). Apparently it (a neuronet) really needs to be trained.
I hope it's not for the Bakinskiye? Ivan, let me tell you a secret in the field of MO, it is not the architecture of the network, layers or number of neurons that is important. In NS, it is very important what data you feed to the network. On the basis of what information it draws a conclusion about buying or selling, and I'm not talking about the output, I'm talking about the inputs. What do you feed as an input? I am just curious. You don't have to tell us your secrets, you just mentioned that you choose the number of neurons and layers and that it decides everything by itself. I can tell you that a regular Perspectron can handle the market, if you give it the right data for the output.
And in general, I admit that I was interested in the classification module, which spoke about it would be interesting to see screenshots of the screen forms with parameters, what you can and what you can't. The price has already been announced.
I hope it's not for the Bakinskiye? Ivan, let me tell you a secret in the field of MO, it is not the architecture of the network, layers or number of neurons that is important. In NS, it is very important what data you feed to the network. On the basis of what information it draws a conclusion about buying or selling, and I'm not talking about the output, I'm talking about the inputs. What do you feed as an input? I am just curious. You don't have to tell us your secrets, you just mentioned that you choose the number of neurons and layers and it just decides everything by itself. I can tell you that a regular Perspectron can handle the market, if you give it the right data for the output.
And in general, I admit that I was interested in the classification module, which was spoken about it would be interesting to see screenshots of the screen forms with parameters, what you can and what you can't. I have already announced the price, so sell the goods, as they say.
What do you mean "sell"? It's not my product, I only have a license.
If you know neural network better than me, then why are you asking for a walkthrough?)
I just shared that neural networks really work. I.e., a kind of incentive for prorammists in this thread.
And it's too early to say, I'm on the real, in fact, did not trade. When I get my hands on it, I will choose the best results, check the results, start, and then it will be seen, whether they work or not in real trade (but forwards don't lie too).
What do you mean sell? It's not my product, I only have a license.
If you know neural network better than me, then why are you asking for a walkthrough?)
I just shared that neural networks really work. I.e., a kind of incentive for prorammists in this thread.
And it's too early to say, I'm on the real, in fact, did not trade. I'll choose best results, check results, start, and then I'll see if they work or not in real trading (but forwards don't lie too).
Ivan, you misunderstood me. I meant advertise the product, of course I was not going to buy it from you, if I bought it, then only at the offices, would help them, so to speak ..... That's all, and about my knowledge of NS, experience is sufficient only and yes today Optimizer Reshetova again worked hard all day. Consider about 100 models today has built by the method of reference vectors, which is considered a costly production. I'm just really interested in the classifications module and want to test my data in an alternative software. Tell me, Ivan, what is the result of optimization? Is it somehow integrated in MT4? Or is the order transmission channel organized there? I want to share my results with you, I'd rather have previews with parameters or at least tell me the name of the product.
Ivan, you misunderstood me. I meant advertise the product, of course I was not going to buy it from you, if I bought it, then only at the offices, would help them, so to speak ..... That's all, and about my knowledge of NS, experience is sufficient only and yes today Optimizer Reshetova again worked hard all day. Consider about 100 models today has built by the method of reference vectors, which is considered a hard costly production. I'm just really interested in the classifications module and want to test my data in an alternative software. Tell me, Ivan, what is the result of optimization? Is it somehow integrated in MT4? Or is the order transmission channel organized there? I don't know, I'd rather have parameters of the product, or at least tell me the product's name or maybe there is a link to a resource?
They did not pay me for advertising))).
They didn't pay me for advertising)))
Anyway, since you said A, then say B, I'm curious about it out of curiosity. I have a dream, you know. All asked to spin my data on their systems, but they vidiliti my sample does not fit. It's too small. Anyway, if you take the classifier training, I can send you the data for analysis. Prepared. The result will be interesting of course.....Why pull. Enclosed are two current files, which today ran JPrediction, if you can build a model during this week, we will run it in a week on new data, and there you will see if the software is worth it or not, money down the drain. What do you think?
Anyway, since you say A, you say B, I'm curious about it out of curiosity. I have a dream, you know. All asked to spin my data on their systems, but they see my sample does not fit. It's too small. Anyway, if you take the classifier training, I can send you the data for analysis. Prepared. The result will be interesting of course.....Why pull. Enclosed are two current files, which today ran JPrediction, if you can build a model during this week, we will run it in a week on new data, and there you will see if the software is worth it or not, money down the drain. What do you think?
I have already said
I already said.
You know Ivan, you're acting like a sissy, why did you say that? Just to show off. Like, look what I did, but I won't tell you anything else. Is that it? You know, this has always been a men's club with balls of steel and no room for chicks. Remember that, if you start talking, talk all the way, people share their experiences to help save years of searching, but not in your case. I got it. Thank you.