Machine learning in trading: theory, models, practice and algo-trading - page 223
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
+++
Although I don't use it yet... But it's a titanic job
By the way, today we've released the MetaTrader 5 build 1485 with the updated math library where we've added a few dozens of functions from R plus a set of high-level math operations and a plot-like graphical library.
The total amount of source code in \include\math is already 6617 kb.
In just \include\math\stat alone is the implementation of 461 mathematical functions with good coverage of R capabilities:
So, MQL5 already has a very good basic mathematical functionality. It wasn't available recently, but we have implemented it very quickly.Note that this does not indicate the possibility of regular libraries Alglib and Fuzzy, which are also presented in the source code.
Great! )
It looks like soon supporters of R will have no arguments at all).
And thank you for all this work.
So you confirm that R wasn't originally designed as a language that fully supports algorithmic trading?
In other words, it was not originally designed for trading. It has a different original purpose. So no one ever thought about the effectiveness of solving the problems of trading?
C++ wasn't created for trading. JAVA wasn't made for trading. R wasn't made for trading. The world doesn't revolve around trading.
MQL was created for trading, so does MQL take all the money in algorithmic trading? I doubt it, most likely C/C++ that did not take algorithmic trading needs into account at all.
The point is that it's not the standard set of language functions, but its syntax possibilities and availability of new created necessary libraries for extending functionality.
For me, the task of trading is to gather a bunch of data, process it, train the model and then use this model to make decisions. This is 99% of work in terms of volume and time, R is ideally suited for this kind of work, because it was developed just for convenient work with data.
The remaining 1% of work is to give a trade order. Metaquotes is a closed platform, no program can connect to their servers, so I have an Expert Advisor with a couple dozen lines of code for receiving/receiving data from R and sending trade orders. All tasks are solved very efficiently, there is nothing to complain about.
C++ was not created for trading. JAVA was not created for trading. R wasn't created for trading, the world doesn't revolve around trading.
MQL was created for trading, so does MQL take all the money in algorithmic trading? I doubt it, most likely C/C++ that did not take algorithmic trading needs into account at all.
The point is that it's not the standard set of language functions, but its syntax possibilities and availability of new created necessary libraries for extending functionality.
For me, the task of trading is to gather a bunch of data, process it, train the model and then use this model to make decisions. This is 99% of work in terms of volume and time, R is ideally suited for this kind of work, because it was developed just for convenient work with data.
The remaining 1% of work is to give a trade order. Metaquotes is a closed platform, no program can connect to their servers, so I have an Expert Advisor with a couple dozen lines of code for receiving/receiving data from R and sending trade orders. It works very well, I have no complaints.
Do not call white black, please.
With MQL5, you have an open gate to all public data of the server. And no crutches, like in other languages. Performance and low latency (in terms of data access and trading operations) of the language has long been proven.
Besides, MQL5 is already the fourth generation of trading languages, which we've been developing since 2001. There were MQL, MQL2, MQL4, and all of them were developed as languages of access to the market environment and trade. We have been engaged in automated algorithmic trading for 15 years.
I still really miss the public information about the data exchange protocol between the trading server and the mt5 terminal. This would give me an opportunity to create my own library to connect from mt5 server to R directly.
The MT5 platform is free and available to everyone, I agree, sorry if I was inaccurate.
I still really miss the public information about the data exchange protocol between the trading server and the mt5 terminal. It would give me an opportunity to create my own library to connect from mt5 server to R directly.
Of course, that won't happen.
It's not like we're crazy to let everyone into such a hard-to-build worldwide ecosystem. It's a business.So you're saying that R wasn't originally created as a language that fully supports algorithmic trading?
In other words, it was not originally designed for trading. It has another original purpose. So no one ever thought about the effectiveness of solving the problems of trading?
However, because of its constant expansion and aggregation of a large number of functions, it started solving problems of cointegration, astronomy, nuclear physics, consumer electronics, in parallel with solving statistical problems, and reached the algorithmic trading?
In other words, algotrading in R exists as a "spice"? As if following the logic: - "if R has everything, why shouldn't there be algotrading?"?
In this case, you're contrasting this "tool for all occasions", clearly sharpened for trading, with professional language MQL?
It's somewhat reckless ... (unprofessional). ))
Issuing trade orders is only a small part of trading, and so small.... that there is no point in discussing it.
Trading is first and foremost the brains that formulate trade orders. Trading must necessarily contain a proof that in the future it will be the same, as on historical data. In trading, these brains are called STATISTICS, ECONOMETRICS. Hence R, or rather part of R (its application is much wider than economics), was originally sharpened to solve trading problems.
Considering that you are just too lazy to look at the composition of R, I'm sorry, your ignorance is off the charts. If you are interested in anything on the topic of this branch, on my posts in this branch and you are able to formulate a problem - please. And I am not interested in your education.
1. Issuing trade orders is only a small part of trading, and so small.... that there is no point in discussing it.
2. trading is primarily the brains that formulate trade orders. Trading must necessarily contain a proof that in the future it will be the same, as on historical data. In trading, these brains are called STATISTICS, ECONOMETRICS. Hence R, or rather part of R (its application is much wider than economics), was originally sharpened to solve trading problems.
3 If you consider that you are just too lazy to look at the composition of R, then excuse me, your ignorance is off the charts. If you are interested in something on the topic of this branch, in my posts in this branch and you are able to formulate a problem - please. I am not interested in your education.
1. Have I ever said otherwise? Where did I speak about trade orders? What are you talking about?
2. Well, isn't it ignorant to look for evidence that the future will be the same as the historical data?)) Statistics collect data as a basis for identifying patterns, but cannot serve as "proof" of them. A pattern revealed by statistics is speculative. You can see the pattern, I can't. And vice versa. Therefore, your "proof" based on statistics is an erroneous conclusion and acceptance of a subjective vision as an objective reality.
In trading, statistics is an analysis tool on a par with indicators, patterns and other types of pattern detection. Everyone decides for himself how to use it. For many people statistics is necessary only for evaluation of trading results, for others - for searching of market dynamics repetitions, for others - for checking quality of indicator signals and so on... However, any unambiguous conclusions about the future based on collected statistics are delusional. Therefore, don't "idolize" statistics - their value in predicting recurrences should also be statistically verified. So, - gather statistics on the accuracy of your statistical predictions, and then statistics on those statistics, and so on...)
Now about machine learning: where are the fruits of its effectiveness? Can you write a universal code in R for recognizing classic price formations? I need an algorithm that should accurately find trends, flots, levels, breakdowns, rebounds, corrections, parabolic curves, channels and much more. All this is technical analysis. If R was originally designed for trading, such algorithms should be implemented by default. Where are they? If they are there, what are we arguing about? - R is the best language for trading!
And so, the proverbial machine learning: neuronets must be able to recognize price figures easily, and not inferior to humans in this skill. Can they do it? Where is the proof? Show me a robot on R, which recognizes all the patterns and then I will say that you are right about everything.
3. R I don't know, and my ignorance is definitely there. However, if you prove its efficiency to me in practice (by presenting trading results, or a robot that can solve the above mentioned problems), then I'll be glad to study R.
But while supporters of R saying "why reinvent the velocipedes?" are proposing to use crutches, the adherents of MQL will make their own bicycle on which they will surely overtake their hobbling opponents.)
Tag Konow:
Now about machine learning: where are the fruits of its effectiveness? Can you write a universal R code to detect classic price formations? I need the algorithm to find trends, flotsam, levels, breakdowns, rebounds, corrections, parabolic curves, channels and many other things without errors. All this is technical analysis. If R was originally designed for trading, such algorithms should be implemented by default. Where are they? If they are there, what are we arguing about? - R is the best language for trading!
And so, the proverbial machine learning: neuronets must be able to recognize price figures easily, and not inferior to humans in this skill. Can they do it? Where is the proof? Show me a robot on R that recognizes all the patterns and then I will say that you are right about everything.
3. R I don't know, and my ignorance is definitely there. However, if you prove its efficiency to me in practice (by showing me trading results, or a robot that can solve above-mentioned problems), then I'll be glad to study R.
But while supporters of R saying "why reinvent the velocipedes?" are proposing to use crutches, the adherents of MQL will make their own bicycle on which they will surely outrun their hobbling opponents.)
When you talk, it's like you're delusional.
show me the answers to your points in any language! do you have such answers? what does this have to do with R?
or do you have a grOal on mql?1. C++ was not created for trading. JAVA was not created for trading. R wasn't created for trading. The world doesn't revolve around trading.
MQL was created for trading, so does MQL take all the money in algorithmic trading? I doubt it, most likely C/C++ that did not take algorithmic trading needs into account at all.
The matter is that it is not a standard set of language functions, but its syntax possibilities and availability of new created necessary libraries for extending functionality.
2. For me the task of trading is to collect a lot of data, process it, train the model and then use this model for decision making. This is 99% of work volume and time, R is perfect for such cases, because it is designed specifically for comfortable work with data.
The remaining 1% of work is to give a trade order. Metaquotes is a closed platform, no program can connect to their servers, so I have an Expert Advisor with a couple dozen lines of code for receiving/receiving data from R and sending trade orders. It works very well, I have no complaints.
1. There's no need to argue here. MQL is a developing language and certainly has not become the master of the entire automated trading area. However it expands and enriches its libraries trying to occupy its own position among other languages, as a language most suitable for algorithmic trading. Those guys who point out MQL's shortcomings in order to encourage others to refuse from it, rather than to contribute to its improvement, are simply throwing "sticks in the wheels" of its development. Criticism should be constructive, otherwise it is just malicious and baseless denigration.
Only people who know both languages very well and think objectively can make a real comparison between MQL and R, but local R followers have a clear biased attitude towards MQL, which they can't prove with codes in both languages. They can't prove the dependence of the EA efficiency on the choice of a particular language with the codes, tests and trade statistics. So I'm critical of their claims.
2. If for you trading is a collection of a heap of data and loading of machine power by cleaning up this heap, then the process of such trading looks rather pathetic...