Machine learning in trading: theory, models, practice and algo-trading - page 1176
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
There are articles on the MOE, where everything is scientifically sound, and here more or beginners can ask questions, or discuss some ideas to try. In general, the article should be written when I am already confident in the results of their actions, I am so far away from it.
And in my opinion, uncertainty is a common feature, as for this branch of the forum, and for articles on the MO.
Manifestations of confidence are observed only in trolling, otherwise the fragile structure, unfortunately, crumbles at the slightest pressure.
And one of the reasons for this, I think, is precisely in excessive scientism, and another is that we, as beginners, are always trying to start from scratch and question fundamental concepts.
It seems to me that instead of these extremes, one should just take ready-made MO models with examples and write ready-made Expert Advisors, test them, monitor them and give reviews in articles.
This is why I suggested my own engine once again and if I am ready to help, I am willing to connect and help.
And in my opinion, uncertainty is a common trait, both for this branch of the forum and for articles on MO.
Manifestations of confidence are observed only in trolling, otherwise the fragile construct unfortunately crumbles at the slightest pressure.
And one of the reasons for this, I think, is precisely in excessive scientism, and another is that we, as beginners, are always trying to start from scratch and question fundamental concepts.
It seems to me that instead of these extremes, one should just take ready-made MO models with examples and write ready-made Expert Advisors, test them, monitor them and give reviews in articles.
This is why I suggested my own engine once again, and I'm ready to help if I need it.
I partly agree with your thoughts, but MO models for trading have peculiarity - non-stationarity, or rather very small samples, unable to describe all possible phenomena, and I do not know such examples by analogy with other directions, that is why I have to make something of my own.
I will not refuse to help, because I'm very poorly versed in model building, and a little bit of Catbust settings saw that the results may very much depend on the settings of the model, which gives more questions than answers. And I have one question today - what is the maximum that can be survived from my sample with predictors, as it is not clear whether I need to make more predictors or they are already enough for the minimum.
Help to connect the catbust in the form of a python to the terminal - certainly need, thank you!
I partly agree with your thoughts, but MO models for trading have peculiarity - non-stationarity, or rather very small samples, unable to describe all possible phenomena, and I do not know such examples by analogy with other directions, that is why I have to make something of my own.
I will not refuse to help, because I'm very poorly versed in model building, and a little bit of Catbust settings saw that the results may very much depend on the settings of the model, which gives more questions than answers. And I have one question today - what is the maximum that can be survived from my sample with predictors, as it is not clear whether I need to make more predictors or they are already enough for the minimum.
Help in connecting ketbust in the form of python to the terminal - will certainly be needed, thanks!
I think that your question is in fact the same as for any other EA, which data to use and which settings to use to get the best result. The answers to these questions probably should be sought through development, debugging, testing and optimization, as usual.
That's what I do :) I just thought that suddenly someone has great skills in this business of working with models...
That's what I do :) I just thought that suddenly someone has great skills in this case is to work with models ...
R and Python libraries give unlimited space in this direction, so once again offered my engine, if necessary, I'm ready to connect and help.
Can I see the source code of engine? I wonder how it is made.
Great tests, thanks
Is there any information on the differential of errors of the train\test? Just take any one Accuracy or logloss there, the most common
for example, something like this
right trace left test:
I'm interested in the model's ability to generalize, and what kind of stuff is there to fight overfits. I see you've quickly mastered the device. Finally, substantive talks :))
It looks complicated; it was already said somewhere above that one of the signs of low overfits is exactly the similarity of graphs for equity on Lerning and Quizzo; the same logic is applied to classification/regression; equity is the consequence.
alas, this problem has no solution:
1. or write in a third-party language(platform) TC, but you get problems:
a) no historical data
b) no tester
c) There is no test on a demo account
-) there may be problems with the platform support, as an example - I googled Alglib, very little information on it on the web, all only on the developer's website, there is no support
all this you need to fix it with .dll, integrations and other crutches
2. or write everything in MQL, you won't have any problems a.b.c.. You will either look for ready-made solutions and mathematical apparatus in Kodobase or write all logic (of the mathematical apparatus) from scratch with MQL capabilities.
If you write your code for yourself, this is the most practical solution. You can't use .dll for Market.
Many development and analysis systems allow you to create a .dll as an example - Matlab
In Matlab, the output is always ready to hand, one line of code and you have a ready chart, all variables are visible, you can change the variables... in one word, Matlab is a ready development environment for a mathematical apparatus.
At first glance it looks as you describe but after about 3 or 5 years of making crutches and liners for RAD systems you soon realize that, unfortunately, "there's no such thing as a free lunch". The situation with RAD systems is like a martingale that "sweeps garbage under the carpet", they simplify template modeling, but greatly complicate customization, the work on which is the essence of the profession. That's why a potential muppet learns on RAD as a beginner for 2-3 years, because it's easier and more effective to learn, and then smoothly passes to his software.
What is worth in a third-party language:
1. upload the history to CSV,
2. Make a tester (it is only and no more than a cycle),
3. On the demo account, you can test, say, through file exchange with the terminal. If you do it through RAM-Disk, the performance is like exchanging through memory - Gigabytes per second.
If the system succeeds, which it won't the first time, it will save a lot of time on modeling. And later on how to get it into the terminal is a soluble question.
For example for some reason the results of the same strategies and data obtained with different testers are sometimes much different, but someone is right (the closest to the truth).
Can I see the source code of the engine? I wonder how it is made.
The engine is integrated into a large project, there are many megabytes of source code in several languages, some interpreters, in addition to python and p, there are also scripting java and pascal.
And if you are interested in the principle and example of Python code execution, which I use, I proposed it here a long time ago.
https://www.mql5.com/ru/forum/86386/page553#comment_6302133