Zigzag indicator and neural nets - page 2

New comment
 
klot:
It works quite well. I mean, a neural network that has this indicator on its inputs is pretty much sunk. But that is not the main thing. The main thing is to find out the algorithm to calculate the probability of Peak or Trough.

Do you have any idea how it can be calculated?

We need to train the network to recognize Peak and Trough right away. Not zigzag input. I.e. history should be divided into peaks and troughs at once (do not trust Conechan) and manually
 

It should be several NSs (like a committee of sorts), because data from different TFs needs to be fed in. Only then the system will become really unsinkable (although, frankly, something tells me that NS can hardly learn to do it well - due to clustering of significant price levels).

Smaller TFs should be additional confirmations for forecasts with NS built for older ones. Respectively the zigzags should be built on different TFs.

But first of all - determine the trading TF, i.e. the TF which signals will be used for trading. Let us say the trading one - 4 hours, and additional confirming ones - 1 hour and 15 minutes. Let's train three NS, each of which gives its own forecast according to the current situation (its current phase), and then we compare those three forecasts. As soon as they become close enough - this is the signal for a rollover on the trading TF.

 
Mathemat:
See nen's corner on Onyx, what's there about the zigzag and other related
stuff like that. Do you need a link or do you know how to find it yourself?

Give me the address, if you don't mind. Thanks.
 

Welcome to http://onix-trade.net/forum/index.php?s=0e28d6cce0220c68b09d61715e75205f&showforum=54

There have also been thoughts of engaging a zigzag to feed signals to a neural network. But I've never messed with neural networks and don't plan to yet.

 
Indeed thoughts are material :)
It turns out that so many people are interested in this issue, I honestly did not even expect it.

But first of all - determine the trading TF, i.e. the TF, on the signals of which we are going to trade. Let us say the trade one - 4 hours, and additional confirming ones - 1 hour and 15 minutes. Let's train three NS, each of which gives its own forecast according to the current situation (its current phase), and then we compare those three forecasts. As soon as they become close enough - this is a signal for a rollover on a trade TF. <br / translate="no">
Mathemat In the first two paragraphs of your post, you said a clever thing, but in the last one
You're a little bit mistaken. The fact is that this approach is generally correct,
only from a slightly different angle.
You automatically condemn yourself to a number of difficulties, with the selection of periods.
I'll tell you right away, so as not to waste either your time or yours. The result is frankly bearable. But if you go a little on the other side of the issue and automatically solve the problem with neural networks, the question is the amount of data needed.
You need to build waves from the bottom up and not from the top down. And build on the structure of the waves. Thus, passing from a small wave to a larger wave.
We start with the basic ZigZag, which is considered the initial vector, and it is used to build waves of the first, second, third, and so on orders.
The result will be something similar.

And it is another story how to send all this stuff to the inputs of the nets.
 

I still don't understand why everyone wants to zig-zag the entrance. It makes more sense that the zigzag is an exit, something to be trained for

 
Prival:

I still don't understand why everyone wants to zig-zag the entrance. It makes more sense that the zigzag is an exit, something to be trained for


More on that point, please. :)
 

The markings are 50 and 100 pips. Where is the take and where is the stop, let everyone decide for themselves. :)

 

All this talk about Zigzag makes one think that the trading systems market will be a sure thing. (And of course, without Zigzag.)

 
SK. писал (а):

All this talk about Zigzag makes one think that the trading systems market will be a sure thing. (And of course, without Zigzag.)

There will always be a Zigzag. It's just that sometimes you can see it and sometimes you can't. :)
123456789...11
New comment
Reason: