Machine learning in trading: theory, models, practice and algo-trading - page 2768
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
Are you serious or are you trolling?
I'm dead serious.
I'm moving the thread towards "trading" rather than mathematical abstractions.
small stop, big takeout ...
nothing else is needed, but 726 rules that were mined for two days, and that's just for sells).
Really a bug on the site, the pictures are not the same, apparently old enlarged.
Trolling of course, well, not much)))).
The question is how long these rules will work. The task is kind of clear, the time of validity of rules should be longer than the period of their production))))))
Really a bug on the site, the pictures are wrong, apparently old enlarged.
I'm fine, zoomed in too... reuploaded just in case...
The question is how long these rules will work. The task is kind of clear, the time of validity of rules should be longer than the period of their production)))))))
trained at 10k test at 55k
Yes and there are a lot of rules, the signal is the sum of the rules, as in Forest so there is a kind of diversification, but the rules are quite different ... very different...
So far the rules are very primitive for the sake of speed, but the speed is not enough even though I wrote in C++ the main engine for them....
And this is despite the fact that I can't feed more than 2k data at a time and this is despite the fact that in the signs one price clause and that's it....
And I want at least 20k data at a time, more complex rules and at least 10 signs, but preferably 50....
So there is room to grow
I'm fine, I'm zoomed in too... I reuploaded it just in case.
trained at 10k test at 55k
Yes and there are a lot of rules, the signal is the sum of the rules, as in Forrest, so there is a kind of diversification, but the rules are quite different ... completely different...
So far the rules are very primitive in favour of speed, but speed is not enough even though I wrote in C++ the main engine for them....
And this is despite the fact that I can't feed more than 2k data at a time and this is despite the fact that in the signs one price clause and that's it....
And I want at least 20k data at a time, more complex rules and at least 10 signs, but preferably 50....
So there is room to grow
Cloze without spread is not very good))))
Klose without a spread is not very good))))))
why the spread?
why the spread?
We open at cloz, it is bid, and close at ask and vice versa. It will be more accurate, though more complicated.
Although it can be made as an external filter.Open on cloz, it is bid, and close on ask and vice versa. It will be more accurate, although more complicated.
In my case, it doesn't matter.
tunnel thinking makes it hard to understand a lot of things. Altshuller called it "inertia of thinking".
ps. Even from one cloze you can synthesise millions of features...
To be more precise, this many.
and that's after discretisation...
tunnel thinking makes it hard to understand a lot of things. Altshuller called it inertia of thinking
ps. even from one cloze you can synthesise millions of features...
To be more precise, this many
and that's after discretisation.
The signs will be the price spread over the entire history of the instrument divided by the price step))))) And their combination. It seems that the number of signs is not the target. There are curses.))))
Zy, only without TRIZ)))))