How do you practically assess the contribution of a "specific" input to the NS? - page 7
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
I suggested to this NS coryphaeus that I feed the NS input ZZ and teach it how to recognise U-turns.
I would have gone too.
I suggested to a coryphaeus here that we should recognise the make of a car by its tread pattern. But the coryphaeus has gone into the doldrums, not even an eye.
I would have gone too.
Maybe someone can explain it to me.
We take the NS and draw samples of the letter "a" in different handwriting. We teach them to recognise the letter "a". And in kotir - what do we teach? You have to mark with pens what we teach with "head and shoulders" or something.
And we input into the NS what we believe may have an impact on further price movement. Last year's snow? -Nope, not good, my mood - not good, the current position relative to yesterday's highs/minimums - maybe, the slope angle of the regression line plotted for some period - maybe, etc. There is a lot of room for imagination. And of course it is absolutely useless to give it all in a bare form, we must take some ratios, series, normalize them, etc.
We cannot find anything on this subject in the literature, even on the financial markets. Everything is behind the scenes. As an example some 10-20 latest closing prices, and that's it.
I started a discussion on this very subject at one time https://www.mql5.com/ru/forum/114902. I learned something there.
Why?
And to the input of the NO we feed that which we think can influence the further price movement. Last year's snow? -Nope, not good, my mood doesn't do, the current position relative to yesterday's maximums/minimums - maybe, the slope angle of the regression line drawn for some period - maybe, etc. There is a lot of room for imagination. And of course it is absolutely useless to give it all in a bare form, we must take some ratios, series, normalize them, etc.
You will not find anything on this topic in the literature, even in the NS literature on the financial markets. Everything is in the background. As an example, some 10-20 latest closing prices, and that's it.
That's why I started a discussion on this subject at the time https://www.mql5.com/ru/forum/114902. I learned something there.
Confirmed my suspicions. The usual TA nonsense. Without crossing that it seemed, count it as a pattern and learn it.
Well firstly, I did not say that this is my entrance) This is an average of what I have seen, mine is not given out so easily. Secondly, why claptrap? Thirdly, what is not rubbish?
Econometrics is not nonsense, so faa thinks.
faa, do you think that all possible trading systems are reduced to regression and pattern recognition?
I am not against the tests themselves, they are more or less correct, because they are taken from statistics (by the way, the testing methodology with confidence intervals itself has long been questioned).
I am concerned about another thing: who determines their adequacy when applied to a given regression?
In this sense, econometrics seems similar to chemistry: there are a large number of "tests", for each of which one must still decide whether they are adequate in relation to the given task.
Thirdly, what's not a nibble?
Secondly, why is it rubbish? Thirdly, what is not rubbish?
I suggested to one NS coryphaeus that I feed ZZ into the NS input and teach it to recognise U-turns.
It's a piece of cake. The results are fantastic.
The trouble is, ZZ draws.