Machine learning in trading: theory, models, practice and algo-trading - page 1167
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
Hm, more details from this point please, a couple of months ago I googled Alglib, on the Internet there are only questions how to use, very little information
preprocessing is already there - standardization, normalization, on the machine... how exactly is it implemented - did not look, because the forest has mainly used
It's written in the help on the site
OK, it's tempting that it's that simple, I'll have to try the automatic synthetics to learn NS, what if.... )))
if i did not have overfits, everyone would be happy now)
You seem to have a skewed understanding of machine learning, not sampling
What does the trending TS have to do with it, I'm afraid to even ask
Don't be afraid, there's nothing wrong with asking what you don't understand.
The trend TS with the fact that a good result for it would be 35% of profitable entries and 65 unprofitable ones, at that 35% of them are both buy and sell entries - we have units, and zeros - a barrier to entry. If we estimate the model by the mean value of the correct answers, we can obtain the correct classification value of 70%. At the same time it will be the same 70% in the case when the zeros are correctly classified 60% and the units 10%, as well as in the case when the units are correctly classified 60% and the zeros 10%, the answer is the same, but in fact the content is different.
There's nothing to be afraid of, there's nothing wrong with asking something you don't understand.
Trend TS with the fact that for her a good result would be 35% of profitable entries and 65 unprofitable, while 35% are both buy and sell entries - we have units, and zeros - a ban on the entry. If we estimate the model by the average value of correct answers, then we can get the value of correct classification 70%, at the same time it will be the same 70% in the case when the zeros are correctly classified 60% and the units 10%, as well as in the case when the units are correctly classified 60% and the zeros 10%, the answer is the same, but in fact the content is different.
So there should be three classes: Buy Sell and Neutral. It's OK, there will be no distortion.
It means you need three classes: Buy Sell and Neutral. It's OK, there will be no skew.
Yandex's tool does not know how to unload the model of multiclassification results.... That's why I have to look for other solutions.
The tool from Yandex does not know how to unload the model of multiclassification results.... so I have to look for other solutions.
2 models then, one allows the trade, the other tells which way
2 models then, one allows the trade, the other tells which way
So it is - trade resolution only passes, the side is determined by one of the variables that is in the sample, but the tree rarely sees it, which even surprises me.
I'm trying different metrics with the same data and I'll show the effect of model's choice method (not training).
So it is - trade resolution only passes, the side is determined by one of the variables that are in the sample, but the tree rarely sees it, which even surprises me.
So now I'm trying different metrics with the same data, I'll throw a couple of results later, to show what effect the method of model selection (not training!) has.
All right, throw them in, because few people write, it's boring... and no one reads what I throw in ))
Well, throw, not many people write, it's boring... and no one reads what I throw))
I read your posts, as well as all others, but there is no constructive, only achievements - so you can only rejoice, and there is nothing to discuss, since they do not say the way they went to achieve the result.
I read your posts, as well as all the others, but there is no constructive, only achievements - so you can only rejoice, and there is nothing to discuss, because they do not say the ways in which you went to achieve the result.
It's kind of the opposite there: no achievements, only constructive and explanatory notes.