Machine learning in trading: theory, models, practice and algo-trading - page 2609
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
The original intention was to make a feature-free model that matches the tags to any feature
Like, for every returnee, you can pick up the moments it predicts.
but there is still sensitivity to the signs, some work very well with some others mediocrely
The original intention was to make a feature-free model that matches the tags to any feature
Like, for every returnee, you can pick up the moments it predicts.
but there's still sensitivity to the signs, some work very well with some others mediocrely
Maxime, you're on fire, that's a great idea.
connect a third neura for pattern recognition
Maxime, that's a great idea.
connect a third neura for pattern recognition
Yes, if you "free your mind" and don't bang on the door: "price entry, predicting candlestick colour" or something similar, then so many interesting options and directions open up at once.
The original intention was to make a feature-free model that matches the tags to any feature
Like, for every returnee, you can pick up the moments it predicts.
but there is still sensitivity to the signs, some work very well and others mediocrely.
This has a DIRECT impact on the results of this technique...
To be honest, the picture is not very clear without explanations. Well and draw a better heat map or density graph.
Same topic https://www.mql5.com/ru/forum/86386/page2534#comment_26672056
By design, it should look for something - you don't know what, but which is consistent. This is on the topic of automatically generated strategies. So far I give it a "C", maybe there will be some movement. An untried theme.
Do you manually mark bad areas? And if manually, how do you determine the boundaries? Are there many losses or is equity below the straight line?
In general, I come to the conclusion that we can simply look at each hour separately. Maybe there is some sense in it. Boxplots make sense)
And if it doesn't sell for roubles, can we meaningfully not switch it off? Sounds logical)
Sounds like saying that using MO leads to brain atrophy))
))) In general, working out visible actions is of course logical, but it is also carried out without understanding the process. Algorithm fitting is just as random, although it is more stable than on a random plot).
Do you manually mark bad areas? And if manually, how do you determine the boundaries? Are there many losses or is equity below a straight line?
In general, I come to the conclusion that we can simply look at each hour separately. Maybe there is some sense in it. Boxplots make sense)
Trained - run on the data, mark what is bad
on automatic
trained - ran it on the data, noted what was bad
on the machine.
I do not understand, how do you draw the boundaries of bad sectors? Is it a place where a position opened and closed with a loss, or do you take a wider range, such as half of the trades with a loss?