Machine learning in trading: theory, models, practice and algo-trading - page 2237
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
Broadcast to the tablet from a normal computer, a mouse radio from the computer + keyboard
Or on the same laptop, too, you can either install a remote desktop or broadcast as a tablet.
A computer on i7 or I9 needeverything has already been written a long time ago, in 5 lines you can do....
Get rid of it without punt mql , all it can do is open/close trades
Then try it and tell me if it's worth it or not.
I also registered the same way the old-fashioned way. A lot of glitches.
I registered through Fb.
Then try it and tell me if it's worth it or not.
)))) cool idea...
I already tried it
)))) cool idea...
I've already tried it.
The author in the video claims that this approach is not yet programmed in python and probably R.
The author in the video claims that this approach is not yet programmed in python and probably R.
what difference does it make what to optimize with?
Isn't the main thing WHAT to optimize?
What difference does it make what to optimize with?
Isn't the main thing WHAT to optimize?
It matters what to optimize and with what.
Boosting and trees have the ability to remember everything, so their approach to learning is not ideal.
And what to optimize is just as important - and here I observe a different efficiency of learning depending on the point where the signal is received - perhaps there is a reference point, where you can most likely determine the future events.
I also want to share the problem that if the binary classification is to trade/not to trade and if the signal is not formalized enough and can be formed often, then we come to a situation when after training there are new signals that appear in the intervals between a conditionally open position and its closing point when preparing data for training, ie if we missed an entry, the signal may reappear before the previous entry.
I got good models but I can't reproduce them because of this effect.
What's important and what to optimize and with what.
doesn't matter...
I don't use typical targets anymore, all these nice pictures with them, it's just a fitting on the story...
never mind...
I don't use the typical taggers anymore, all these beautiful pictures with them, it's just a fitting on stories...
What do you use now?
Final version of functions of mql5 Expert Advisor with python program.
There are two functions in the advisor, one updates the time in the file and the second reads the actual trading signal in the file, which is formed in the python program.
The python program in the "not_actual" status reads the current time, calculates the actual signal and writes it into the file.
This glue is not very fast, but it works independently. It works in the tester, on the demo, I have not tried it on the real. If there are, questions or ideas as it is possible to improve, write, and that theme as stalled...
Why ready-made does not suit? here and here. You actually need only part which is responsible for communication between MKL and Python (ZeroMQ).
Good luck