Machine learning in trading: theory, models, practice and algo-trading - page 2660
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
through the transaction IDs, there are markings there.
What is a transaction identifier?
a note whether or not a trade was opened by hand.
what other options are there? )mark whether or not it was opened by hand.
what other options are there? )Take a look at the announcement of the new MetaTrader 5 build 3360 Platform Beta release : Float in OpenCL and maths functions, activation and loss methods for machine learning, please.
We have already done a lot of work both on the introduction of built-in data types vector, vector (as well as templated vector<double> vector<float>), matrix, matrixf, complex, and on the large expansion of mathematical functions.
Already now MQL5 has become convenient for native and fast writing of neural networks and matrix operations. And with extended support in OpenCL.
We have also done preparatory work to implement a powerful machine learning engine natively in MQL5 language. This will allow us to write full-fledged ML systems directly in the platform.
Take a look at the announcement of the new release of MetaTrader 5 build 3360 Platform Beta: Float in OpenCL and maths functions, activation and loss methods for machine learning, please.
We have already done a lot of work on both the introduction of built-in data types vector, vector (as well as templated vector<double> vector<float>), matrix, matrixf, complex, and a large expansion of mathematical functions.
Already now MQL5 has become convenient for native and fast writing of neural networks and matrix operations. And with extended support in OpenCL.
We have also done preparatory work to implement a powerful machine learning engine natively in MQL5 language. This will allow us to write full-fledged ML systems directly in the platform.
About 10 years ago, everyone would have gasped, now they're like, well, well, well
.
Availability of native matrices and vectors with a large set of standard mathematical functions is a necessary foundation for a breakthrough in algo-trading capabilities. And we have done it.
Import of onnx models is also in preliminary development, though without 100% support. Unfortunately, there are few packages that can 100% fully import any onnx file.
It is certainly more convenient and productive to do research and development in existing systems right now. But execution can be ported through onnx models without the use of third party libraries.
With the addition of ML engine, it will be possible to do research and development directly in MQL5. This is the next leap.
Availability of native matrices and vectors with a large set of in-house mathematical functions is a necessary foundation for a leap in algo-trading capabilities. And we have done it.
Import of onnx models is also in preliminary development, though without 100% support. Unfortunately, few packages can fully import any onnx file 100%.
It is certainly more convenient and productive to do research and development in existing systems right now. But the performance can be transferred through onnx models without using third-party libraries.
With the addition of ML engine, it will be possible to do research and development directly in MQL5. This is the next leap.
About 10 years ago, everyone would have gasped, now they're like, well, well, well
.
10 years ago, they would not have gasped.
No one has ever been able to depict the Fourier transform properly, because complex numbers are used there.
They do it and moan that the end of the indicator is distorted.
They have opened the way, from the point of view of mathematical research.
It is normal, even super.
does anyone have a trading robot fish in python?
For ML robot...