Discussion of article "Neural networks made easy (Part 2): Network training and testing" - page 3

 
This is just my personal opinion.
I have studied Neural Networks in Metatrade4 for years, I have done 3 layer networks with n neurons in each layer.
As inputs I have used indicators, in multitimeframe, etc, etc
And the result is always the same. Terrible.
Excellent when they are trained, almost perfect results.
But when they have to put into practice what they have learned, they are a complete disappointment.
 
Zhiqiang Zhu:
12.68% hit rate? Does that have any real meaning? I'm getting about 50% probability on coin tosses too. With such a low probability, I don't know what the purpose of this thing is.
The neural network in transactional finance is not at all what it is used for, and what it learns is to look for laws and seek universal formulas, and the logic itself is absurd. The real transactional financial artificial intelligence is not to make judgements, but based on the five-dimensional four-dimensional chaotic system for the downgrading of the fight.
 
Zhiqiang Zhu:
12.68% hit rate? Does that have any real meaning? I'm getting about 50% probability on coin tosses too. With such a low probability, I don't know what the purpose of this thing is?
Trading algorithms are divided into three levels, the first level of technical indicators (rubbish), the second level of big data strategy (rubbish), the third level of logical necessity (invincible). My algorithm is the third level, security 100% five trading days plus a zero.
 
Shi Chao Ma:
Transactional finance in the neural network pressure is not it so used to learn what it is to find the law to seek the universal formula, the logic itself is ridiculous. The real transactional financial artificial intelligence is not to make judgements, but based on the five-dimensional four-dimensional chaotic system of downgrading strikes.
I would like to learn from my seniors about the downgrading of five dimensions to four dimensions.
 
Shi Chao Ma:
Trading algorithms are divided into three levels, the first level of technical indicators (rubbish), the second level of big data strategy (rubbish), the third level of logical necessity (invincible). My algorithm is the third level, security 100% five trading days plus a zero.
I also believe that trading logic is the most important and soulful part of EA, so please advise me more!
 

Hi Dmitriy, I have read your article and I have gone through the code and I see that you only use one data set for training, but you do not use another data set for validation and thus avoid over-optimization. With such a large neural network, with several hidden layers and so many neurons per layer, the network will surely memorize all the data, but it will not be able to predict, once the training is finished.

Greetings and thanks for your article and the code


Gerardo

 


For the Fractal.mq5 file, I receive the following error while debugging:


2022.01.13 08:30:54.502 Fractal_1 (BTCUSD,M1) CSeries::CheckLoadHistory: requested too much data (100801)

2022.01.13 08:30:54.502 Fractal_1 (BTCUSD,M1) failed to get 100801 bars for BTCUSD,PERIOD_M1



Any suggestions on how to edit the code to not request greater than the max number of bars of data (10,000)?

 
Here are the files I am working with.
Files:
Fractal.mq5  36 kb
NeuroNet.mqh  40 kb
 
Josh #:


For the Fractal.mq5 file, I receive the following error while debugging:


2022.01.13 08:30:54.502 Fractal_1 (BTCUSD,M1) CSeries::CheckLoadHistory: requested too much data (100801)

2022.01.13 08:30:54.502 Fractal_1 (BTCUSD,M1) failed to get 100801 bars for BTCUSD,PERIOD_M1



Any suggestions on how to edit the code to not request greater than the max number of bars of data (10,000)?

You are using M1 timeframe. And 10 000 min is only 7 days. It too small to train NN.

 
Dmitriy Gizlyk #:

You are using M1 timeframe. And 10 000 min is only 7 days. It too small to train NN.

Thanks for the reply, Dmitriy!! Should have started with 1H like you do at the end of the article.

Really appreciate your Neural Network Made Easy series! I hope to master these concepts in MQL myself (although it is easier to pass data to R using MTR lol).

These are high potential algorithms and you have done a masterful job with the library and series, thank you!!