Machine learning in trading: theory, models, practice and algo-trading - page 1031
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
How many times have I written, and you're still like little kids. In order to teach the network to bounce we need to work out the conditions when we will do it. That is, we need a point in time when we feed the network with data and it gives us the result. If we are talking about a rebound from a level, the task comes down to one thing. We need to classify this level using the NS. If the NS says "1" or "yes" to the level we want to determine, then we believe that this level is support and a bounce will occur, or rather we build our trade on a bounce from this level, if the network says "no" to this level, there will be no bounce and the trend will continue. When you form a task for the NS and do not know which way to approach it, try to break your task into a number of small subtasks, until you begin to receive an answer to the mini-task as "Yes" or "No". Then you can do inverse transformations by getting the value of the net. But the point of the trade is either down or up or fence "I don't know". It is very important to formulate a question to the NS so that it would be clear to the NS itself, and these are usually simple questions.
Does anyone know what a "simple question" is ?????
Right is a question the answer to which can be either yes or no. From the answers to simple questions you can form more generalized structures and make more global conclusions, getting the results of several NS.
As for the levels, I wrote above that the task comes down to determining what kind of level "Support" or "Resistance". Having defined the level correctly, all of your subsequent trades will bring you profit. The important thing is to correctly identify what kind of level it is. I gave you examples above.
Many of you fail in the NC, because forming the task you make mistakes over and over again. Therefore attempts do not bring success, because you fundamentally make mistakes. This is all from not knowing what you can do and what you can't do. Ask me about it and I will not be able to answer this question right away. Because it is developed with years of experience. But the gross errors are immediately visible.
If you want to make a good TS on levels, you can do it together and openly. Although to whom I say, you have already fucked up with regression. Don't fuck up now with levels....
The support and resistance lines are the high and low of the previous bar of a larger period. You can also put MA(3) on the high and low of a larger period and use them as lines.
It's easier to trade within the trend than to rely on these lines...
The support and resistance lines are the high and low of the previous bar of a larger period. You can also put MA(3) on the high and low of a larger period and use them as lines.
It is easier to trade inside the trend than to rely on these lines.
))
) Well, a trend is a body of a candle of a larger period. And we never know beforehand where it has begun and where it has finished.
))
Well, a trend is a body of a candle of a larger period.) And it is never clear in advance where it has already begun, and where it has already finished.
This is why it cannot be predicted and there is no need to.
Support/resistance levels are formed from large volumes in the stock market
yes
It is also worth noting that levels are not a specific number, they are a distribution at a specific price
Yes, to make it even simpler, the zone.
I think this is a non-trivial task.
The devil's not so bad...
In general @Mihail Marchukajtes says it correctly.
I don't know, I haven't heard any deep thoughts from him
I also think @Yuriy Asaulenko's approach is correct. Applying MO to ready made strategy is simple and hearty.
When constructing a model of an OC for trading, it is necessary to follow the strategy. And by separate "blocks" of models to solve specific subtasks, such as finding levels.I'm trading with my hands, despite the fact that the levels are counted by a neural network or rather an ensemble of neurons, but that's not the point
I also think that @Aleksey Vyazmikin's suggestion to build a team was unfairly left out.
What kind of suggestion did I miss?
And it's necessary to share experience and expertise. And for this you need a structured communication on the themes. Which of course in this branch is not possible.
Shared as much as I could when I was a seeker, the old-timers remember, but everyone has his own idea of the market and therefore his own way ...
I decided to create a team to work on the development of a trading robot. Someone said let's optimize the stochastic and asked me to perform this task, of course I wouldn't do that, do you think I was lazy?
Okay, now we've talked about it again and forgotten)))
What exactly are you suggesting?
By Michael's correct words I meant this comment: https://www.mql5.com/ru/forum/86386/page1031#comment_8177887
I do not know what he proposes (from his own bell towers) will not work, not counting the common phrases like solve the big divide into small
I do not follow his advice and would not, again purely from personal experience.
What's the offer, I missed it.
I've shared as much as I could when I was a seeker, the old-timers remember, but everyone has his own idea of the market and therefore his own way...
I've decided to create a team to work on creating a trading robot, someone said let's optimize the stochastic and gave me the task, of course I wouldn't do that, do you think I'm lazy?
You can read about the proposal here.
The concept of the team does not provide a specific manager, but offers a mechanism of self-management, through which you can influence the development of research areas, especially if a person is actively involved in the implementation of mini-projects.
Regarding Alex's suggestion in this thread: https://www.mql5.com/ru/forum/261479. Yes, he beat me to it).
He started by saying that he decided to put together a team for a specific task.
And I saw it as an opportunity to create a community of commerce and application of IOs there. Like the Open Data Science community. They are based in slack chat. It's very convenient for me.
Of course, we can discuss trading with the help of the MO, but they are not very fond of it.
I propose an alternative community, based on discord'a, for example. Servers there for free. Ability to create rooms, voice communication, user groups to customize access.
Flood and talk on general topics in a common channel, separate themes for technologies and projects. That's then everyone can find what he likes and participate where it is interesting, rather than where someone needs help in any endeavors.
Why haven't they created one? Here they are, feel free to join in. We'll figure it out as we go along.
https://discord.gg/ZsN2cf
as soon as the channels on this site are finished, it will be possible to move back hereAleksey Terentev:
For example, on the same levels.
I have thoughts about realizing it on neural networks, plus using new methods for building neural network models, which have very interesting results
you. But what's the use of me going on about it in this branch, talking about my experiences, sharing links and experiences. My posts will go into the annals of the forum, as well as this entire branch.
very interesting.
Well, why didn't you create one? We'll figure it out as we go along.
https://discord.gg/ZsN2cf
As soon as the channels on this site are finished it will be possible to move back herehttps://discord.gg/p4SKkhV