![MQL5 - Language of trade strategies built-in the MetaTrader 5 client terminal](https://c.mql5.com/i/registerlandings/logo-2.png)
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
By winding up "MotorHead" on the alarm clock, blowing the minds of those who don't have a lecture at eight in the morning, I was somehow positioning myself. Now I don't even know what to do.
No - it's not a method. And even my girlfriend won't let me...// could have been more original! )))
Eh, eh... where are you guys who hate heavy rock...
Expert Advisor according to a signal scheme (+1, -1, 0).
The higher the Smoke parameter, the smoother the curve on the optimisation interval.
Expert Advisor according to a signal scheme (+1, -1, 0).
The greater the Smoke parameter, the smoother the curve on the optimization interval.
Generally speaking, I have figured it out.
The perceptron inputs are fed with signals in the form (Open[i] - Open[i + shift]) / Open[i + shift], i.e. Momentum with the shift period divided by 100
The results are calculated as the sum of the products of the weights on the values of the inputs.
The results are normalised by dividing them by the maximum absolute value of the inputs.
They are then compared with the threshold value. If the normalized value of the perceptron output is below the threshold, we smoke bamboo on the fence. If the value is positive, it is a signal to buy. A negative value is a sell signal.
The most useful thing here is normalization of the perceptron output for comparison with the threshold value. I haven't used it in my EAs yet, but I should try it.
Yura, I take it that my suggestion didn't impress????
I don't know what sentence I'm talking about. If the message is in private, then there was some silly question about whether it is necessary to use perceptrons in TC. Since I do not give personal advice on non-private topics, I did not read further and deleted the message.
The question may be answered publicly: perceptrons are not necessary at all, because the method is more universal and suitable for any TS. But not every TS:
1. fits well to the story
2. Even if it fits the history, it will not necessarily give a satisfactory profit on OOS.
3. If it gives a profit on OOS only from the N-th attempt, and in most cases it fails because the probability of getting a profit in the future is approximately equal:
p = number of successful attempts / number of all attempts
Such are the pies.
You should have read to the end. There was not only a question, but also a suggestion to improve DG..... toAschetO!!!!!
You should have read to the end. There was not only a question, but also a suggestion to refine DG..... inAschetO!!!!!
It's not a suggestion for improvement, it's called "getting madder and madder". I don't need to be polluted by such "rationalisation suggestions".
What you are suggesting, i.e. dragging all the settings into the programme, is of no fucking use to anyone. I intentionally made the program in such a way that it would give the maximum result with the minimum of intervention of crank handles, i.e. so that it would set up all routine things automatically.
If you get a buzz out of tweaking and rebuilding everything, you don't need any program: optimize and test everything manually from morning till night and from morning till night - there are plenty of tweaks for every taste in strategy tester.
It's not a suggestion for improvement, it's called "getting madder and madder". I don't need to be polluted by such "rationalisation suggestions".
What you are suggesting, i.e. dragging all the settings into the programme, is of no fucking use to anyone. I intentionally made the program in such a way that it would give the maximum result with the minimum of intervention of crank handles, i.e. so that it would set up all routine things automatically.
If you enjoy tuning and rebuilding everything, then you don't need any program: optimize and test everything manually from morning till night and from night till morning - there is a sea of settings for every taste in the strategy tester.
If so, comment on how the test range is selected?????
If so, comment on how the test range is chosen?????
See https://www.mql5.com/ru/forum/131739 - all long ago commented on
If I personally repeat the same thing to everyone, I will waste my whole life, and the retarded will not get it anyway.