Machine learning in trading: theory, models, practice and algo-trading - page 857
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
The calendar API has been announced, but it's not in MT5 yet.
It would be interesting to use a news feed... I'm not sure if the results would be satisfactory, but just for the sake of curiosity
+ you need to google intensively for new research on non-stationarity. In RL this research is being intensively done at the moment, i.e. the topic is still evolving, so I'm sitting on it for now. The simplest example is effective manifold feedbacks, it's impossible to calculate them analytically or to imagine them speculatively, so only through multiple experiments :)
The calendar is great! And in the meantime, make acquaintances, because we need to understand the key indicators of this calendar, which at least potentially influence the value of the currency, the market expectations. In general, the market is noisy with speculations, but nevertheless the central banks will sell the interests of their states, especially such as the USA, the UK, the EU, China and Russia. And large financial companies understand this.
For example economic indicators to analyze the financial performance of companies more than 50. Realistically used 4-5 key indicators for the company - this is if you are lucky with the economist, if not then you have all 10 and sometimes 20 indicators are calculated. That is why you have to communicate with financiers - every state has its own characteristics and its own set of key parameters for its current state: recession, recovery, recovery ahead of schedule, crisis (recession ahead of schedule), and so on. Otherwise you can get lost in this calendar.
And as for the other two parameters - you can probably cope with them yourself.
But first it is necessary to communicate with experts in fund. analysis. and maybe even at your leisure and read books on the subject to communicate was easier. All the same, it is impossible to be an expert in everything. A long way behind, it means success has become closer, have no doubts and dig deeper :)
Good luck,
But first it is necessary to communicate with experts in fund. analysis. and maybe even at your leisure and read books on the subject to be easier to communicate. After all, it is impossible to be an expert in everything. You've come a long way, it means success is closer, don't doubt and keep digging :)
I am well acquainted with FA, I know analysts with a lifetime of experience.
Yes, another 4 parameter is the assessment of the political situation around the "BASE" base and quoted currency. This parameter is a user parameter (analytical) most likely. That is, it should be somehow manually set the weight for it by some system, I am sure that the financiers have such a point. Political strife, trade wars - all this influences the financial market not worse than reports of central banks of the countries of the issuers of national currencies. Surely they have an analytical news system which exactly delivers a certain risk factor for currency forecasts based on the political relations between the countries. We can see it with the ruble today)). On the one hand some industries will win, other import-intensive industries will lose. Then we look at the volume of their participation in the formation of the budget and figure out the final impact on the budget. It seems that the budget because of the oil / gas will not suffer, but again, the affected industries will need to subsidize. All very interconnected, it is necessary to calculate)))) In general, there are a lot of factors for the relevant (essential) analysis / training of the NS trading system. I understand that it no longer wants to learn from the history of quotes... But if you create a new algorithm, the history of quotes will help check the relationship of these key 4 parameters (conditionally) and the behavior of prices, not in the short term, but in the medium and long term for sure. However, the solution may be even simpler, just sharing my idea. It is up to you to decide.
So your NS trading system is certainly more relevant day by day. We can not wait for stability and 100% annual returns :)
Yes, another 4 parameter is the assessment of the political situation around the "BASE" base and quoted currency. This parameter is a user parameter (analytical) most likely. That is, it should be somehow manually set the weight for it by some system, I am sure that the financiers have such a point. Political strife, trade wars - all this influences the financial market not worse than reports of central banks of the countries of the issuers of national currencies. Surely they have an analytical system of news which exactly delivers a certain risk factor for currency forecasts based on the political relations between the countries. We can see it with the ruble today.) On the one hand some industries will win, other import-intensive industries will lose. Then we look at the volume of their participation in the formation of the budget and figure out the final impact on the budget. It seems that the budget because of the oil / gas will not suffer, but again, the affected industries will need to subsidize. All very interconnected, it is necessary to calculate))))
So your NS trading system is certainly more relevant day by day. We can't wait for stability and a 100% yield :)
The problem is that fundamentalists are doing even worse with macroeconomic models than neuro-networkers are doing with theirs :)))
@geratdc_.
Don't listen to Maxim, he's not in the loop.
You can easily achieve R2 > 0 returnees. It would even be enough to make profit on eurusd m5 by trading on bar opening prices, assuming one or two pips spread. And all this is quite public means, the main thing is to know how to use them.
But where to get such spreads?
For greater accuracy, you need to create your own indicators, and modify neuronics to meet your requirements. Then R2=0.1 is already quite possible, and it is profitable.
Then even more complicated problems begin, but here on this forum it's not accepted to talk about them, it's anti-advertisement.
And those who are smarter than me - even with standard indicators and scaffold.
@geratdc_.
Don't listen to Maxim, he's out of the loop.
Cut the crap, darkie.
monitoring in the studio, you have not lived up to the more complex problems, which are openly written about on the forum (including me writing) :)
all these posts look like stupid excuses, honestly. Would you believe yourself?
And this is taking into account that I do not deny the possibility of a profitable TS on the NS (as well as without it). But I just do not understand all this idle writing without facts.
About trading on m5 opps and taking into account the spread is nonsense, oh nonsense. The impact of the spread is already negligible there.
I don't understand all this idle scribbling.
It's too bad this thread is "empty writing" for you. In this thread, dozens of people have described various algorithms for training models, after which the model result on fronttest starts to correlate with the results on backtest. Obviously, you won't put the working robot in a single piece, you'll have to assemble it in pieces, testing different algorithms and packages.
The conclusion is obvious - you haven't tested the algorithms presented in this thread, or you would have changed your mind a long time ago.
It's too bad this thread is "empty writing" for you. In this thread dozens of people have described different algorithms for learning models, after which the result of the model on fronttest starts to correlate with the results on backtest. Obviously, you won't put the working robot in a single piece, you'll have to assemble it in pieces, testing different algorithms and packages.
The conclusion is obvious - you haven't tested the algorithms presented in this thread, otherwise you would have changed your mind a long time ago.
I don't use other people's algorithms. It's not empty writing when there's an algorithm and a report, and you can make a couple of hints how to better implement TC.
You've sent the tester report where your NS does not make money, although forward and backtest + - look the same. I have these too, there are also those who earn but not always, and I want to always (or almost always). And only then will be able to give out advice.
All I've seen is scraps of reports without detailed explanations, and monitoring Mikhailo. Everything. My research, so to speak, I have described in more or less detail.
Stop pouring water, dark one.
monitoring in the studio, you have not yet lived up to the more complex problems that are openly written about on the forum (including me writing) :)
all these posts look like stupid excuses, honestly. Would you believe yourself?
And this is taking into account that I do not deny the possibility of a profitable TS on the NS (as well as without it). But I just do not understand all this idle writing without facts.
About trading on m5 opps and taking into account the spread is nonsense, oh nonsense. The impact of the spread there is already miserable.
Yes it is. This is the market with low spreads, but in general the dynamics are the same. The main thing is to get to the lot multiplication and then it will be more fun...
Yeah, sure. There's not much to brag about, but in general the dynamics remain the same. The main thing is to get to the multiplication of the lot and then it will be more fun...
I am watching, waiting till the annual range will be at least 2x