Machine learning for robots - page 8

New comment
 
Ivan Negreshniy:

There must be more signals, that's for sure. But it is important for the input data to be adequate to the task to be solved.

I purely intuitively, based on your words about the maximum size of a pattern and seeing the pattern indicator, entered 6 stochastic values and 150 price bars in the training sequence.

Perhaps you, knowing the signs that define your level pattern, can suggest other features and / or formulas to calculate them, at the moment the EA uses these:

I am very unfamiliar with metatrader, at the level of a housewife((

If you want, I can describe the pattern and we will think about it together.

 
mytarmailS:

I am very unfamiliar with metatrader, at the level of a housewife((

If you want I can describe the pattern and we can think together

The problem comes down to selecting the minimum required set of parameters that define the pattern and formalizing it so it does not depend on the instrument, timeframe, broker, etc., and the solution may result in the fact that the neural network is no longer needed :)
 
Ivan Negreshniy:
and as a result of the decision it may turn out that a neural network is no longer needed:)

shhhh, that's no way to write on this forum! I've tried it a couple of times.... the answer is "you don't understand anything about NS" and "your view of NS is very outdated"...

 
Ivan Negreshniy:
The problem comes down to choosing the minimum required set of parameters defining the pattern and formalizing it so it does not depend on the instrument, timeframe, broker, etc. And the solution may result in the fact that a neural network is no longer needed :)

Good afternoon! Excuse me... I am not a programmer... Unfortunately, God has not given me the talent and ability to do this difficult and admirable thing...

I have a question for you: Can I generate an EA that sends signals with arrows and opens/closes positions based on these signals, based on an indicator like this:

In the bottom window, dots of different colours show directions of fractals from different timeframes: red - top fractal, blue - bottom fractal, yellow - double (uncertain) fractal?

 
Igor Makanu:

shhhh, that's no way to write on this forum! I've tried it a couple of times.... The answer was "you don't understand anything about NS" and "your view of NS is very outdated"...

Maybe you don't understand anything about NS, )) but NS and other IOs really are sometimes necessary, sometimes not necessary. Sort of a truism.

We have enough outdated views in general, since overcomplicated algorithms in the end do not give us much extra. What is "overcomplicated" is decided in each specific case. And there is evidence of this from the field of mathematical modelling of complex systems.

 
A123:

Good afternoon! Excuse me... I am not a programmer... Unfortunately, God has not given me the talent and ability to do this difficult and admirable thing...

I have a question for you: Can I generate an EA that sends signals with arrows and opens/closes positions based on these signals, based on an indicator like this:

In the bottom window, dots of different colours show directions of fractals from different timeframes: red - top fractal, blue - bottom fractal, yellow - double (uncertain) fractal?

You can write a template file with signal arrows and I will generate an EA for you, trained on a given number of price bars, specific symbols and timeframes.

In order to be included in the training sample, indicators must also be attached to the template and you need to specify how many of their values should be used in the pattern.

For automatic arrows, I can add your logic/indicator as a filter in the above script - makeSignals.

And for example, purely intuitively, I tried to generate the layout of your EA.

I took signals from EURUSD H1. Since there is no your indicator, I took predictors from standard fractals on several timeframes summing them up with the following formulas: 

#define  CALC_X0(n) ((iFractals(NULL,PERIOD_M1,MODE_UPPER,n)==iHigh(NULL,PERIOD_M1,n)?1:0)+(iFractals(NULL,PERIOD_M1,MODE_LOWER,n)==iLow(NULL,PERIOD_M1,n)?-1:0))
#define  CALC_X1(n) ((iFractals(NULL,PERIOD_M5,MODE_UPPER,n)==iHigh(NULL,PERIOD_M5,n)?1:0)+(iFractals(NULL,PERIOD_M5,MODE_LOWER,n)==iLow(NULL,PERIOD_M5,n)?-1:0))
#define  CALC_X2(n) ((iFractals(NULL,PERIOD_M15,MODE_UPPER,n)==iHigh(NULL,PERIOD_M15,n)?1:0)+(iFractals(NULL,PERIOD_M15,MODE_LOWER,n)==iLow(NULL,PERIOD_M15,n)?-1:0))
#define  CALC_X3(n) ((iFractals(NULL,PERIOD_M30,MODE_UPPER,n)==iHigh(NULL,PERIOD_M30,n)?1:0)+(iFractals(NULL,PERIOD_M30,MODE_LOWER,n)==iLow(NULL,PERIOD_M30,n)?-1:0))
#define  CALC_X4(n) ((iFractals(NULL,PERIOD_H1,MODE_UPPER,n)==iHigh(NULL,PERIOD_H1,n)?1:0)+(iFractals(NULL,PERIOD_H1,MODE_LOWER,n)==iLow(NULL,PERIOD_H1,n)?-1:0))
#define  CALC_X5(n) ((iFractals(NULL,PERIOD_H4,MODE_UPPER,n)==iHigh(NULL,PERIOD_H4,n)?1:0)+(iFractals(NULL,PERIOD_H4,MODE_LOWER,n)==iLow(NULL,PERIOD_H4,n)?-1:0))
#define  CALC_X6(n) ((iFractals(NULL,PERIOD_D1,MODE_UPPER,n)==iHigh(NULL,PERIOD_D1,n)?1:0)+(iFractals(NULL,PERIOD_D1,MODE_LOWER,n)==iLow(NULL,PERIOD_D1,n)?-1:0))
#define  CALC_BAR(x0,x1,x2,x3,x4,x5,x6) (x0+x1*2+x2*4+x3*8+x4*16+x5*32+x6*64)

When tested the Expert Advisor shows a small number of deals, because it is trained on large movements, but their profitability is already some confirmation that your "fractal" idea is working.


Files:
FRACTAL_RF.mq4  459 kb
 
Ivan Negreshniy:

You can write down a template file with signal arrows and I will generate you an EA trained on a given number of price bars, specific symbols, timeframes.

In order to be included in the training sample, indicators must also be attached to the template and you must specify how many of their values should be used in the pattern.

For automatic arrows, I can add your logic/indicator as a filter in the above script - makeSignals.

And for example, purely intuitively, I tried to generate the layout of your EA.

I took signals from EURUSD H1. Since there is no your indicator, I took predictors from standard fractals on several timeframes summing them up with the following formulas:

When tested the Expert Advisor shows a small number of deals, because it is trained on large movements, but their profitability is already some confirmation that your "fractal" idea is working.


fun)))

What about your robot from the first page?

 
mytarmailS:

fun)))

What's your robot from the front page?

I don't know, it doesn't make sense, it's based on pure prices from MetaQuotes-Demo training and is tightly bound to the broker, because even on hourly timeframes the difference in quotes is significant.

By the way, maybe somebody has done this activity and can suggest a formula to unify OHLC that would use the same result for quotes from different brokers and provide the maximum informative value of the candlestick?

 
Ivan Negreshniy:

I don't know, it doesn't make sense, there are pure prices from MetaQuotes-Demo training and tight binding to the broker, because even on hourly timeframes the difference in quotations is significant.

By the way, maybe somebody has done this activity and can suggest a formula to unify OHLC that would use the same result for quotes from different brokers and provide the maximum informative value of the candlestick?

I think there is no such thing with forex

 

Backing up the conversation, bros....

In the early days of becoming and acquiring popularity among the masses there was one of the fundamental rules comparable to the rule of input rubbish output and it sounds something like this "If a task can be solved without the help of neural networks, it should be solved", i.e. the abbreviated meaning of the phrase: when a task has no direct or explicit solution, only in that case it is reasonable to use NS. That is, NS is a last resort when solving problems of current or future uncertainty in complex areas, with an implicit solution, etc. But if the problem can be solved so.... without NS, then it should be solved that way.... without NS. Then the result of solution will always be stable, while NS implies some freedom in solving.... like I want to do this today, and tomorrow I will want to do this.... As an example.

Unfortunately, maybe that's the reason I am so dumb and don't know much about IO, during all my career I have read only 2-3 books at the very beginning of my way, but no matter how many times I returned to IO literature, it was always boring, because it often contained things I already knew and I couldn't get anything new from it. Therefore, I have an interesting task which I will devote a separate topic to... So... everyone else can do it,but I can't????

123456789101112
New comment