Machine learning in trading: theory, models, practice and algo-trading - page 3060
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
Some code that doesn't work for me again. If you want substantive discussions, post reproducible results.
The code works and is reproducible.
The code is working and reproducible.
Yes, it is working - we figured it out - I had the wrong version of R.....
How to make minute quotes for 10 years from it and load it into MT5 so that it would be possible, can you tell me?
The code is working and reproducible.
Nobody ever doubted it ;))))
He installs a new version of R for each library, and then speculates whether the developers are fools or not.... )))
It's funny. And sad.... And disgusting...
we continue to chew on cauzal in our spare time.
through forceful brain injections
The first part, where the aunts compare different lerners.
The first part, where the aunts compare different lerners.
I'm looking at it, and so far I've only got this idea: the so-called effect is essentially an error on a delayed sample.
In other words, it's some kind of justification for everything going wrong. But I don't understand, where is the way to identify the exact cause.....
And what do you see the point of this research for trading?
And what did you see as the point of this research for trading?
you will have to translate their marketing definitions into normal human language to figure out how to screw them in.
Roughly speaking: there is a group of trains with a tritment in the form of a trained model, let's say, there is a test (control group) without tritment. All other conclusions and uplift of the model is done according to the proposed methods. maybe this is not quite a correct analogy.
Look at it in a simpler way: you do any tritment (cause) and then analyse the effects through all sorts of randomised tests. You get a causal analysis.
you'll have to translate their marketing definitions into normal human language to figure out how to screw it in.
Roughly speaking: there is a group of trainees with a tritment in the form of a trained model, let's say, there is a test (control group) without tritment. All other conclusions and uplift of the model are made according to the proposed methods. maybe this is not quite a correct analogy.
Look at it in a simpler way: you do any tritment (cause) and then analyse the effects through all sorts of randomised tests. You get a causal analysis.
Perhaps I really didn't understand the purpose of all this.... But it seemed to me that the goal is to detect the influence of a new factor, or it can be thought of as an outlier of the past value of the predictor, on the indicator (price or something else - regression is mostly in examples). Then the task should be to detect these outliers when the chronology of events remains unchanged (you cannot randomise sample lines for time series). And, it turns out that this is a rare event, or a one-time change. Then it is enough to look at the changes in the distribution of the predictor index over a fixed time window. Those predictors, which have such a change, are the cause (or maybe not - here I did not understand their idea how to determine the cause or effect), and if these changes at different parts of the test more often lead to the effect "the model does not work", then we need to make the model work more carefully with these predictors....
Perhaps I really didn't understand the purpose of all this.... But it seemed to me that the goal is to detect the influence of a new factor or can be thought of as an outlier of the past value of the predictor on the indicator (price or something else - regression is mostly in the examples). Then the task should be to detect these outliers when the chronology of events remains unchanged (you cannot randomise sample lines for time series). And, it turns out that this is a rare event, or a one-time change. Then it is enough to look at the changes in the distribution of the predictor index over a fixed time window. Those predictors, which have such a change, are the cause (or maybe not - here I did not understand their idea how to determine the cause or effect), and if these changes at different parts of the test more often lead to the effect "the model does not work", then we need to make the model work more carefully with these predictors....
Predictors and tritment are different things. Causal works with outcomes, to determine ATE (average treatment effect) if no traits are involved, or CATE (conditional ... ...) if there are covariates (traits), taking into account some external influence (e.g. an advertising campaign, which is the tritment). This is to determine if there was an effect on the control group. Then, given that effect, you can analyse and improve your model or some metric.
... you're the one trying to figure out how to apply something you don't know how to apply.
take a free short introductory course