Using Neural Networks in Trading. - page 11
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
No, no! You show us and we'll discuss it before you turn your intellect up to full speed.
Sorry, no time yet. I'm not in the habit of discussing, I've done it -- checked it -- analysed it, it's better this way.
Sorry for the intrusion.
1. What is "bleaching"?
2. and what is the input? For all indicators are filters and derivatives of price, which is the essence of time wasting.
Here it is written quite normally and not even in bird's eye language. I'm just reading it...
On detailed analysis, it becomes clear that we are dealing with an attempt to map a natural series 1,2,3...n (input data, e.g. candle size) by influencing on it by operator (FR) into a series of real numbers.
It is clear that in this mapping, we get a discrete series of real numbers, where numbers will be clustered around some selected points (see figure), such as 230 or 405, etc. And when trying to build a plot of its density distribution, we will of course get a "hole" distribution, where the probability of the "hole" area in the direction of its increase proportional to the width of the hole.
In general, I tend to think that this way of aligning the initial distribution for the NS input data is not suitable for the case when these data are represented as integers.
In general, I am inclined to think that this way of aligning the initial distribution for the NS input data is not suitable for the case where that data is represented as integers.
So that's where the dog is buried! I don't think it's worth bothering with it.
Come on, it's all for the garbage!
I don't think so either - I didn't really want to have that flat shelf... We'll use our hands to smooth it out if necessary.
firstly, can someone explain to me how the alignment differs from the equation (take all Y to the value obtained by dividing 1/(xn - x0) - uniform PR, if there are holes - approximate it)?
secondly, what then we have a discrete value - just a value measured in whole numbers? no way, it can't be... if you measure the trajectory of a projectile in nanometers it can also be considered a discrete quantity... as no one can measure the fractional part of a nanometre and therefore all values would be integers...
secondly, what do we have then, that a discrete value is just a value measured in integers? come on, there can't be such a thing... if you measure the trajectory of a projectile in nanometers it can also be considered a discrete quantity... as no one can measure the fractional part of a nanometre and therefore all values would be integers...
This is how the greatest discoveries are made in passing!
In his time, the great Newton, thinking about the continuity of motion of a thrown stone and the discrete nature of measuring its trajectory, came to the discovery of differential calculus, which opened up a fundamentally new horizon in science.
firstly, can someone explain to me how alignment differs from equation (take a blunt equation of all Y to the value obtained by dividing 1/(xn - x0) - uniform PR, if where there are holes - approximate it)?
You need to think up such a function (like a*x^2+b*x+c), what would let the input data through it to get other data, but with equal density distribution (it's good for NS), then process them inside NS and apply inverse transformation to output and you will get a forecast for input data! Voila.
So, it's not equating, it's preserving the basic properties while serving the table (alignment) to the NS.
Yeah, I read you gentlemen from the first page and almost fell asleep. What, no one has solved the problem of neural networks in forex?