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
That's what I'm asking, can it not?) was curious about this behaviour myself. Not a single trade to america and europe
You have a purchased black box.
This is the disadvantage of it, that you cannot modify it yourself.
Look at its documentation, maybe there is such a possibility.
You have no one to ask)) We know neither the device, nor the official pre-sale description, nor the docs that are issued to customers.
I don't have them either) the authors are selling a black box without docks for now.
I am intrigued. Will there be an answer to my question?
Does it take into account the market spread during training, or a manually set constant spread?
Is there anywhere you can see a graph of the balance and the TALES of one pair, with one order in the market, with a fixed lot?
Ivan Butko:
What are the input signals to the adder in this neural network? How many are there? What kind of neural network, simple one layer weight training or something more complicated? What can you say without knowing anything about what you are asking. Can you lay out what it trains?
What are the input signals to the adder in this neural network? How many are there? What kind of neural network, simple one layer weight training or something more complicated? What can you say without knowing anything about what you are asking. Can you lay out what it trains?
There's parameter "depth", there you set the number of bars from zero, how many are taken into account, max - 20. And you can choose "classification" or "regression" and "prediction of the next bar" or "prediction of the trend's value". And what, where, and how, as said above - black box.
There is a parameter "depth", where you set the number of bars from zero, how many are taken into account, max - 20. And you can choose "classification" or "regression", and "forecast the next bar" and "forecast the value of the trend". And what, where, and how, as already said above - black box.
Excuse me, but I assume that under the guise of "black box" there is a simple neural network like the one described in the article https://www.mql5.com/ru/articles/497
For example, you may open or close bars or something else and run it in the tester. You can introduce the filter for entry by time, during the day. It's not difficult to try to check it, maybe you'll get the same result.
And here, in the third post, is what you are asking for. Max drawdown 170 pips in 2 years. And on the real I have several training in one.
I have trained a neural network several times. Each of them has a different profitability graph, but each of them opens trades only at night and early in the morning.
Does it mean that the neural network has found a working strategy? Or is it better to train it so that it trades all day long?
buy for 25 kopecks without even asking how it trades? hmmm... well, this is ridiculous.
the chart shows that it's flat, it trades with no stops\and long stops. One rare pullout will eat up all the profits and half the deposit.
It is obvious that in the training there are filters that only allow trading at low volatility or by time. It may be just a limiter, not necessarily at the level of NS.
Doesn't it even work in the backtester if you ask such questions?
It looks like a normal self-trainer with some degrees of freedom, i.e. it is trained in a pseudo-random way, so the results vary from training to training, this is normal.
In the video that was uploaded in the other thread, you can also see that it trains a set of models and then fits the best one using the test period, as in genetic optimization. I.e. it is also some kind of fitting, but NS do not train differently.
And why bother creating a bunch of topics with the same thing? It's just advertising