A quick and free library for MT4, much to the delight of neuralnetworkers - page 56
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
You've been fiddling with this expert for a long time, if I understand correctly (about that particular version) - it's built on RSI and stop-loss level. The user (you) only opts (selects) the value of the stop-loss (based on history - as conceived by the author) - the very idea of the strategy is unprofitable (marginally profitable), the fish are not there...:-)))
I'm not a fisherman, but I've been told by the old-timers:
-- There are twelve experienced fishermen standing on the shore ten metres away.
As a result, one fishes while the rest nervously smoke and ask: "What are you fishing with?"
;-)
There can be many reasons for this phenomenon:
1. so-called "overtraining".
2. An "inadequate" teacher.
3. Fixed feet.
4. Inadequate number of neurons.
5. Overabundance of neurons.
6....
7...
We could go on and on.
Experiment. Note mistakes (your own).
Hi all !
I've read about the net and the proposed Expert Advisor and realized I do not know where to start training, and I need to figure out how to do the first step!
Please send me some info or a step-by-step Guide to get started in a nutshell!
Hi all !
Actually I would like to understand how to do the first step first
Enable spell checking in your browser
That's one,
Two, read a book: A. Ezhov, S. Shumsky "Neurocomputing and its applications in economics and business" can be found everywhere. Nothing special, but it will do for the first steps.
Understood.
Troubled with these EAs for 2 weeks. Can't train them.
None of these EAs give any positive result on any timeframe or on any pair during a year of optimization (2008-2009, 2009-2010, 2010-2011).
I have nothing to choose from. Did exactly as described in https://www.mql5.com/ru/code/9386.
By the way, when trying to test with default parameters the tester writes to the log:
16:35:42 FANN-EA_Reshetov_right_SLTP inputs: StopLoss=180; TakeProfit=180; porog=0.0001; porogDigits=4; Lots=1;
16:35:42 2009.01.02 10:00 FANN-EA_Reshetov_right_SLTP EURUSD,M5: dll calls are not allowed; 'Fann2MQL.dll'-'f2M_create_from_file'
16:35:42 2009.01.02 10:00 FANN-EA_Reshetov_right_SLTP EURUSD,M5: expert stopped
16:35:42 2009.12.31 18:59 FANN-EA_Reshetov_right_SLTP EURUSD,M5: dll calls are not allowed; 'Fann2MQL.dll'-'f2M_save'
16:35:42 2009.12.31 18:59 FANN-EA_Reshetov_right_SLTP EURUSD,M5: expert stopped
How do I allow the import of dll in the tester? Or the problem is that the called functions are not in the library?
It's a very interesting thing, this FANN, and the results are sometimes very surprising.
I've read articles and dug sources and decided to attach this device to my "dummy" Expert Advisor (I had a lot of ugly inputs on the flat, so I decided to try to get rid of them).
At the input I implemented 8 signal sources (with history for 30 bars - as a result 240 neurons per input) to avoid false positives, set market entry threshold for the committee... At the beginning everything is as it should be, it learns, opens correctly, if it has opened incorrectly - "gets smacked" and keeps trying without errors, but here is a problem - this clever creature is gradually moving away from the threshold for the committee, crosses it and stops opening trade! Moreover, the value does not fall further - it stays within several hundredths of a deal opening. I tried to decrease the threshold, the result is the same, variations with the number of neurons and their asymmetry in hidden layers don't help... Now I'm in a stupor, trying to wean her off her laziness. But I'm going to break her or she's going to break me :)
People, either I'm dumb or the skis are wrong. I keep getting this problem - 'path' - parameter conversion not allowed fann-ea_4.mq4
71 line during initialization
int ann_load (string path) {
int ann = -1;
ann = f2M_create_from_file (path); //here)
if (ann != -1) {
debug (1, "ANN: '" + path + "' loaded successfully with handler " + ann);
}
if (ann == -1) {
/* Create ANN */
ann = f2M_create_standard (4, AnnInputs, AnnInputs, AnnInputs / 2 + 1, 1);
f2M_set_act_function_hidden (ann, FANN_SIGMOID_SYMMETRIC_STEPWISE);
f2M_set_act_function_output (ann, FANN_SIGMOID_SYMMETRIC_STEPWISE);
f2M_randomize_weights (ann, -1.0, 1.0);
debug (1, "ANN: '" + path + "' created successfully with handler " + ann);
}
if (ann == -1) {
debug (0, "ERROR INITIALIZING NETWORK!)
}
return (ann);
}
People, either I'm dumb or the skis are wrong. I keep getting this problem - 'path' - parameter conversion not allowed fann-ea_4.mq4
71 line during initialization
int ann_load (string path) {
int ann = -1;
ann = f2M_create_from_file (path); //here)
if (ann != -1) {
debug (1, "ANN: '" + path + "' loaded successfully with handler " + ann);
}
if (ann == -1) {
/* Create ANN */
ann = f2M_create_standard (4, AnnInputs, AnnInputs, AnnInputs / 2 + 1, 1);
f2M_set_act_function_hidden (ann, FANN_SIGMOID_SYMMETRIC_STEPWISE);
f2M_set_act_function_output (ann, FANN_SIGMOID_SYMMETRIC_STEPWISE);
f2M_randomize_weights (ann, -1.0, 1.0);
debug (1, "ANN: '" + path + "' created successfully with handler " + ann);
}
if (ann == -1) {
debug (0, "ERROR INITIALIZING NETWORK!)
}
return (ann);
}
Change path to p by the example below
void