Using neural networks in trading - page 35
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
Painting ...
I see you have NSDT as a friend of MT. Obviously through the datafeed of the clot. Is that correct? If yes, I have a question - how stable this connection works? Are quotes transmitted to silk, does silk have time to work "its business", aren't silk trade commands lost in MT?
I see you have NSDT being friendly with MT. Obviously through the datafeed of the clot. Is that correct? If yes, I have a question - how stable this connection works? Are quotes sent to silk, does silk have time to work "its business", aren't silk's trade commands lost in MT?
No, datafeed is mine. Stable. I do not send anything to MT4 (commands to open orders)
I stepped away from the nets. Just showed that this can happen. In the strategy that I pointed out, the result is not sustainable. I'm not sure what the result will be, and I'm not sure what the result will be.
I personally think using nets for trading is difficult (purely personal). You can certainly achieve some interesting moments. But the basic postulate buy cheaper, sell dearer (personally I have not managed to do it). The systems follow the market, and in a sharp market (like now) even if I use ticks, little good will come out.
Again, it does not mean that someone may have better results. You don't judge by yourself.
Good luck to you all ))))
No, the datafeed is mine. It's stable. I don't send anything to MT4 (commands to open orders)
I'm away from the nets. I just showed that it may be. In the strategy that I have shown, the result is not stable. I also use only the "face" of the seal. The network is built in a neurosolver.
I personally think using nets for trading is difficult (purely personal). You can certainly achieve some interesting moments. But the basic postulate buy cheaper, sell dearer (personally I have not managed to do it). The systems follow the market, and in a sharp market (like now) even if I use ticks, little good will come out.
Again, it does not mean that someone may have better results. One does not judge by oneself.
Good luck to you all ))))
All. what. you. listed. is. my. reason. for. giving. up. TA. Initially a successful TA and then it starts to go stale. And in the beginning of fading it is impossible to determine whether the drawdown is natural or the TS is dead and we will learn about it when we sell the deposit. One hassle.
It's these circumstances that made me switch (back) to econometrics. In it, the main question is whether one can trust what one has constructed. Specifically, on the numbers. It's all in lieu of faith.
All. that. you. listed. is. my. reason. for. giving. up. TA. At the beginning the TS is successful, and then it starts to fade. And at the beginning of fading it is impossible to determine whether the drawdown is natural or the TS has already died and we will learn about it when we sell the deposit. One hassle.
It's these circumstances that made me switch (back) to econometrics. In it, the main question is whether one can trust what one has constructed. Specifically, on the numbers. It's all in lieu of faith.
I use this network when the market is calmer and more trendy or something. You just have to understand what you're pitching in and what you want to get. Lack of clarity in the head generates various illusions. Or the sleep of the mind breeds monsters.
In my case everything works, I am not satisfied with the timeliness of the signal. I'm not a black box, I can see the whole structure of the network, I can "tune out" neurons if I want, I can accelerate or slow training, I can make training online by mistake, or by time, limit number of instances for training. Again, the information presented has some element of adaptation and goes through filters. (By the way I can use the network to identify which tools are more related, and then use them as evidence)
To sum up: I've heard a lot here about how great it is to use econometrics, but I haven't seen a single example of use at all.
In what way?
How?
Well, as an option we feed the prepared data containing a set of instruments (for example, all currencies possible) to the output some kind of target function. For this nerve is great in terms of usability (to optimize, to play with tools), so we look who makes the biggest contribution to the prediction. From there we can get twisted, build our own network, feed it with data we've picked up and move on... But then again, I honestly gave up on that a long time ago. I'm more interested in high and low. In this case nets must be set in "yes-no" mode. But in this mode they are too weak for time series (again, imho).
As an option we feed the prepared data with a set of tools (for example, all currencies that are possible) to the output some target function. For this nerve is just perfect in terms of usability (to optimize, to play with tools), so we look who makes the greatest contribution to the prediction. From there we can get twisted, build our own network, feed it with data we've picked up and move on... But then again, I honestly gave up on that a long time ago. I'm more interested in high and low. In this case nets must be set in "yes-no" mode. But in this mode they are too weak for time series (again - imho).
I'll build a multiple regression and use partial correlation coefficients to determine the same thing. Why build a network?
Probably in order to make more adequate forecasts. It's kind of a prediction thing.
Probably in order to make more adequate forecasts. It's kind of a prediction thing.
Well, it's hard to say whether the NS or the regression is a better predictor
even a linear regression can show better results than NS