Neuro-forecasting of financial series (based on one article) - page 11

New comment
 
Reshetov:

There are two types of MA crossing:

1. Bottom up

2. From top to bottom

Maximising will not work because once a crossing has happened from bottom to top, the next one will only be from top to bottom. I.e. maximise them or not, and the result will be stable 50%/50%.

In this case, it may be more appropriate to take the convergence-divergence of MAs as an input.

this is just a possible option.

 
By doing so, the network will search for the maximum number of signs in the input while increasing the balance. Thus looking for only what really works. Man, this point is too tight in NSH.... There is no possibility to optimize the network for its own function......
 
Reshetov:

Have you ever read Grandpa Krylov's fable called Quartet? No matter how you guys sit down, you're not fit to be musicians.

The two cars are also alternately 50/50 likely to change places.

In this case, I'm talking about convergence-divergence, that is, the change in distance between two lines, not their intersection.

And stop your bullying - it's just unseemly.

 
How did you not realise that the MA is just an example, it is clear that there is no useful information in it..... It's just an example. It's about something else..... I'm just curious what do you think, who wants to try it?
 
nikelodeon:
How did you not realise that MA is just an example, it is clear that there is no useful information there..... It's just an example. It's about something else..... I just wonder what you think, who will try it?

just everybody's too lazy.

but it should work.

 
Well, how to make the optimizer so that the value of the variable tends to zero.... when optimizing......Something like????
 
nikelodeon:
Well, how to make in the optimizer so that the value of the variable tends to zero.... when optimizing......I think at least????

I first normalise all input signals in the range -1, 1

after that I normalize them again with the activation function, so that everything that happens to go off-scale after the first activation is sure to stop in the range {-1,1}

only after that I feed it to the input of the neuron.

 
Reshetov:
So you try all sorts of crap that doesn't contain any useful information. You can also take mersi as an assistant, because in his small-minded opinion it should work.
You're starting to bore me.
 
Reshetov:
Relax
When the boor is around, I can't sleep.
 
Reshetov:

You're some kind of pervert. Why sleep with boorish people?

All normal people sleep with decent whores. At most with his wife.


To fuck boorish people
1...456789101112
New comment