How to build a NN-EA in MT4 - page 5

To add comments, please log in or register
ruparaw
117
ruparaw  

NeuroMTC Training results

Here we go...i got all with different result. I run EURUSD

Period 5 Minutes (M5) 2007.11.08 13:15 - 2007.12.07 20:55...hope it will good enough feed to make it clever.

1) Backtesting by without visual mode.

2) Backtesting by visual mode.

3) After run backtesting by visual mode then paste new Vasya;64;64;7 to experts/files from..../tester/files, then rerun again backtesting by without visual mode.

well i don't know what was happened...i will forward test tomorrow using this new Vasya;64;64;7, will see what going to be happened.

Sergey Golubev
Moderator
114247
Sergey Golubev  

When you are backtesting (in any mode) so EA is using Vasya from this folder ..../tester/files.

When you are trading so EA will use Vasya from this folder experts/files.

So, make sure that your Ea will use trained Vasya.

When you are training Vasya (=Willy) so you are training him by backtesting in visual mode together with attached Etalon indicator to visual chart.

If you want to trade so you must copy trained Willy from ..../tester/files to experts/files because EA will use Willy from this folder experts/files during the trading.

It's all that I understand from author's post.

Anyway I think some coder should look at this EA to say exactly what to do as i am not a coder: I am making suggestion according to my experience so ...

ruparaw
117
ruparaw  
newdigital:
When you are backtesting (in any mode) so EA is using Vasya from this folder ..../tester/files.

When you are trading so EA will use Vasya from this folder experts/files.

So, make sure that your Ea will use trained Vasya.

When you are training Vasya (=Willy) so you are training him by backtesting in visual mode together with attached Etalon indicator to visual chart.

If you want to trade so you must copy trained Willy from ..../tester/files to experts/files because EA will use Willy from this folder experts/files during the trading.

It's all that I understand from author's post.

Anyway I think some coder should look at this EA to say exactly what to do as i am not a coder: I am making suggestion according to my experience so ...

well i think you are right NG, the third run i am tested is means nothing. Only during forward test EA will use my latest Vasya in experts/files.

ruparaw
117
ruparaw  
newdigital:
As to this post.

Place NS2-32.dll to C:\Windows\system32\

As to Koh_2007.def file so I am not sure: may be c:\Koh_2007.def, or create folders and place: C:\NeuroShell 2\EXAMPLES\Test\Koh_2007.def

I have no idea, sorry.
bolt:
yes stick in your c drive ie c:\Koh_2007.def as the EA refers to this location but as far as i can tell this is a wieghts file and assumes you have already have access to neuroshell where you perform your own optimization first then when you have a great forwards and backtest generated in neuroshell you copy the wieghts file and paste it into the koh-def file which then tells the EA what to do as i dont think it can do it itself. Im trying the neuroMTC ea but having trouble to get the tester file to update. I tried deleting the contents etc but it still wont write new wights to it so im stuck.

Back to KOHONENN v15, I manage to backtest as well. My doubt is I don't see any updated file either at c:\Koh_2007.def or C:\Windows\system32\NS2-32.dll occured after testing...trying to search at all my C drive also not find any updated one...we still have to figureout how to get the trained def to be replace.....or it might also no need to be train...well i also don't know just thought in blind ...will also forward test this ea tomorrow.

Here is part of the KOHONENN v15 mq4 code i am helping our programmers here to translate from Russian language:

// This for the analysis of the code of the return of the function

int ret;

// Here is assigned way and name of file to neyroseti, trained into Neyroshelle2

string defpath="C:\Koh_2007.def";

// Further is used small craftiness. Since in MQL4 there is no address arithmetic and it is cannot

// to peredovat' variables according to the reference into the external functions. That in order to avoid this limitation

// we transfer massif, by size into 1 element. After the fulfillment of function OpenNet we obtain:

int netnumber[1]; // Indicator to the network

int Nin[1]; // The size of massif is input

int Nout[1]; // Size of the massif of the outputs

double inarray[], outarray[]; // Let us here declare massifs it is input and outputs respectively.

Symbol EURUSDm (Euro vs US Dollar)

Period 1 Hour (H1) 2007.11.01 00:00 - 2007.12.07 20:00

Sergey Golubev
Moderator
114247
Sergey Golubev  

This KOHONENN v15 is having different logic with training.

tiger_wong did good post on the previous page:

..Have a function to collect its own tick data ...

This EA must have additional tool: script to collect the data for NS2-32.dll ...

I will post it when i will read those page in this russian link. Because this KOHONENN v15 was created by modules: module to collect the data for NS2-32.dll , module for training ...

I will read this russian webpage and will the post something which I understand from that.

ruparaw
117
ruparaw  
newdigital:
This KOHONENN v15 is having different logic with training.

tiger_wong did good post on the previous page:

This EA must have additional tool: script to collect the data for NS2-32.dll ...

I will post it when i will read those page in this russian link. Because this KOHONENN v15 was created by modules: module to collect the data for NS2-32.dll , module for training ...

I will read this russian webpage and will the post something which I understand from that.

appreciate that NG...awaiting your updated post

teldon
374
teldon  

i found these indicator and expert somewhere,

Can't test it.

Can somebody look at the codes and help testing please.

TQS

Files:
ruparaw
117
ruparaw  
teldon:
i found these indicator and expert somewhere,

Can't test it.

Can somebody look at the codes and help testing please.

TQS

I saw this code from end part of Neural Networks and Stock / Forex Trading. It is a kind ea refer to forex_nn indicator to decide the decision. The weight is exported from the NN created and teached using Cortex software pasted into forex_nn indicator.

I also don't know why the ea won't run (execute order). The logic is simple it will place buy if the indicator just above 0.21 from previos bar and place sell if the indicator just below 0.9 from previos bar ...

tiger_wong
150
tiger_wong  

hi ruparaw,

did you already put the EA at Eur/Usd with TF 1H ?

ruparaw
117
ruparaw  
tiger_wong:
hi ruparaw, did you already put the EA at Eur/Usd with TF 1H ?

yes i did it last night back & forward test at Eur/Usd with TF 1H . Nothing triggered but no error.

To add comments, please log in or register