Machine learning in trading: theory, models, practice and algo-trading - page 1166
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
And in what models can I adjust what I need - the limits of the values of the correct answers?
None, I think... these metrics show the generalizability of the model. You can vary the settings of the model itself, so that the errors change, for example... add regularization or whatever... in short the other way around - the way you want, it doesn't work
I don't understand the problem statement very well.
You know, maybe Gypsy will tell you :) or you can look up the settings on YouTube, I honestly haven't got into it
None, I think... these metrics show the generalizability of the model. You can vary the settings of the model itself, so that the errors change, for example... add regularization or whatever... in short the other way around - the way you want, it doesn't work
I don't understand the problem statement very well.
maybe chick can tell you :) or you can look up the settings on youtube, i honestly didn't get into it
Video of course I have seen.
It's strange that nobody sets such a goal, because for trend strategies it is obvious - 30% of successful entries will easily cover 40-60% of unsuccessful ones. I'm not trying to guess correctly one hundred percent, it's just the metric.
Of course I've already seen the video.
It's strange that no one sets such a task, because for trend strategies it is obvious - 30% of successful entries will easily cover 40-60% of unsuccessful entries, which is what I need from the model. And I don't need 100% of correct guesses, and the metric is customized for this.
What's the logic? 30% of successful entries is better than 60%.
What is the logic? 30% of successful entrances is better than 60% or in what
The logic is that the world is not perfect and you should take what you can get.
The logic is that the world is not perfect and you have to take what you can get.
This is more like a credo :))
it's more like a credo :))
Do you have 50 but 50 inputs in your sample?
My sample is skewed because of the trend TS, and standard metrics are oriented on symmetry, that's why I'm looking for a suitable metric that takes into account this nuance.
Do you have 50 but 50 inputs in your sample?
My sample is skewed because of the trending TS, and standard metrics focus on symmetry, which is why I'm looking for a metric that suits me and takes this nuance into account.
You seem to have a skewed understanding of machine learning mechanisms, not sampling
I'm afraid to even ask what the trending TS has to do with it.
bored, tired of reading, found a terminal and with the help of alglib scribbled MLP-network.... strange that everything works, I even forgot the normalization, but the indicator is adequate.... I suspect that the devil is in the details of the alglib... or i never got used to the numbering of indicator buffers in MT5 (((
What's up there? i don't have mt5
Well well, in theory I teach MLP for 1000 bars, then I draw the whole history on the chart, there should be a huge mistake, especially normalization did not do, but for some reason it draws a very good history....)))
there's already built in preprocessing
That's it, now let's go shake the dust out of the bags and goWell, the theory teaches MLP for 1000 bars, then I draw the whole history on the chart, there should be a huge mistake, especially normalization did not do, but somehow it draws a very good history....)))
i don't understand it (show me the picture)