Machine learning in trading: theory, models, practice and algo-trading - page 1976
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
this "system" is designed for 1 DTs with positive slippages
And where is the system itself, what is compared to what?
At this stage it does not matter. The main thing is that there is a series with the regularity and the system with expected payoff greater than 6. Now we need to achieve this result using MO. There are doubts that this is possible.
On 20 5-digit Ranks the expected payoff is ~2.3, spread is about 5, i.e. the system is losing. I'll take a zigzag for targeting. Then when I finalize the tester I will try to learn without the teacher. And then I will try transfer learning.
It does not matter at this stage. The main thing is that there is a series with regularity and a system with expectation greater than 6. Now we need to achieve this result using MO. There are doubts that it is possible.
Combine the ticks into conditional bars larger than the spread and see if there is anything left, taking into account the spread
Combine ticks into conditional bars larger than the spread and see if there is anything left, taking into account the spread
The bigger is the step, the smaller is the number of profitable deals.
combine ticks into conditional bars larger than the spread and see if there is anything left, taking into account the spread.
Spread is an external value. Although DT does correlate...
it worked )
Why do you need all this trouble? You amaze me. Rstudio has everything you need. Writing scripts in R/Python and debugging them is much easier in Rstudio than in VSCode.
What do you need Python for?
Name the tasks you can't do with R and have to use Python.
In fact now, with the reticulate library, there are no tasks that can't be done in R using Python modules.
Good luck
I added new predictors (all sorts of returns according to my technology, volatility, regression channel for the day), I think it turned out okay - not perfect - plausible.
But I'm on the plus side with my old ones as well, 30 models (100%) in the green zone - nice, I don't even know, maybe the CatBoost learning settings have had some effect.
You would at least show one code or description. Just some trees, leaves... I don't know what you're doing so much time)
some 30 models, some something there...
1.5 months. I did not save the others.
The spread really spoils the picture if the model parameters were first adjusted without it. Very strong influence, it is necessary to rewrite almost logic
Although, it would seem. 15 minutes. What to say about the ticks.
I wish we could update the equity curve for the tester, it would help us understand a lot. And we should also mark some deals on the chart.
You would at least show one code or a description. Otherwise I don't know what you've been doing for so long.)
Some 30 models, some kind of something...
I'm going to quit my job and start writing articles - I need something to eat...
It turned out that I do not possess the skills of sycophancy, hypocrisy and backstabbing - I can't survive in the office jungle ...
Now I'm going to quit my job and start writing articles - I have to eat something to eat...
It turned out that I do not possess the skills of sycophancy, hypocrisy, and backstabbing - I will not survive in the office jungle...
The main thing is not to be too clever in articles, and to write on the fly what is. To formulate my thoughts briefly and clearly, and to deliver them. And without any philosophical digressions)
Type of guys, there is a theme, that's what I did it.
The main thing is not to be too clever in the articles, but to write what is there. To formulate my thoughts briefly and clearly and to deliver them. And without any philosophical digressions)
Type of guys, there's a theme, that's what I did it.
Pay for volumes :)))