How do you practically assess the contribution of a "specific" input to the NS? - page 14

 
Graff:

How about if we feed the percentage of price change (open-close) for all instruments from the market overview to the input of the NS, and the Zig-Zag as the expected value. Thus, after training, based on price movements of all instruments the NS would predict the ZigZag for the current instrument. what do you think?

It all depends on the architecture of the network. If the network is without a teacher, how will you (by feeding the client) explain the network to predict the Z-Zig-Zag. If it is a network with a teacher, you must feed the teacher during work too. What will you feed. And so Svetlana is right, will draw.... ZZ and NS are not compatible..... Forget this idea. What you are trying to implement now has already been researched, not only by me, about 5 years ago..... No success.....