Using artificial intelligence at MTS - page 4
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
I will digress from the topic a little bit, although it is also quite in line. While driving from work today, it occurred to my dumb brain that we should all reconsider our attitude to indicators. It is in the perspective of their use in neural networks, but not only in them. In short, I want to start from the idea that an indicator is not a small tool for screen decoration, but a tool that helps to trade. In my opinion, the best way to help this process is to estimate the probability of price movement up or down by a certain amount of points without any wisdom. Therefore, let us consider that the indicator is a number (or rather the function of price series) that changes from +1 to -1. The sign of this number shows the supposed direction of price movement - '+' up, '-' down, while the module - the probability of reaching a significant amount of points in this direction, for example 30 (it will be better to make it an obligatory indicator parameter). I.e. all indicators have a unified and uniform interface. What they have inside them is entirely up to the authors' conscience. I came up with it especially for the purpose of connecting indicators to neural networks. In this case they are extremely easy to connect. But I think the idea has its own value. I won't have to deal with a new indicator written using this standard, its curve is immediately understandable. Otherwise, it may often happen that you see an indicator on the Internet. There is no description of it. And even if there is a source code, it is difficult to say what the author had in mind and what to do with it... Alas, such popular things as various moobs and Bollinger lose their right to exist with such an approach. But no one promised it would be easy... The advantages of such a standard, it seems to me, many times outweigh its disadvantages.
Good point! Unfortunately, it has too many chances to remain an idea. Although in fact the question seems to be simple.
There is a certain indicator with its own range of values. There is only one input parameter - the N value in points of price change. We need to build a new indicator the area of definition of which will be the area of the old indicator and the area of values - range (-1,1). And the meaning of these values - the probability of the price change by N points in the corresponding direction, as described by eugenk1.
The theoretical solution of this problem can be implemented only for each indicator individually and is hardly possible. Therefore, this question probably shouldn't even be discussed. But the phenomenological solution (again, for each indicator separately) may be implemented. To do this we just need (just do it :-) to analyze statistics of this indicator on the history. What is missing - the correct, from the point of view of mathematical statistics, formulation of the problem. Unfortunately, I am not strong in this field.
Eugene, maybe you could formulate a universal procedure for assessing this probability for each value of some indicator ? Or some particular one?
Who's not lazy;-) Optimise the optimisation period in my Expert Advisor from the Championship. My Expert Advisor itself occasionally shows profit on M15, H1. I have not got time to experiment with it.
If it is not a secret - how much time has passed between the official announcement of the Championship and the end of registration?
Three months
Three months.
Yeah, I read it, chuckled at some of the posts claiming an unconditional win :).
Official announcement - July 19, end of registration - September 25. 2 months and a week. In principle, with a working idea, it could be possible (if it does not require thousands of lines to implement).
I will digress from the topic a little bit, although it is also quite in line. While driving from work today, it occurred to my dumb brain that we should all reconsider our attitude to indicators. It is in the perspective of their use in neural networks, but not only in them. In short, I want to start from the idea that an indicator is not a small tool for screen decoration, but a tool that helps to trade. In my opinion, the best way to help this process is to estimate the probability of price movement up or down by a certain amount of points without any wisdom. Therefore, let us consider that the indicator is a number (or rather the function of price series) that changes from +1 to -1. The sign of this number shows the supposed direction of price movement - '+' up, '-' down, while the module - the probability of reaching a significant amount of points in this direction, for example 30 (it will be better to make it an obligatory indicator parameter). I.e. all indicators have a unified and uniform interface. What they have inside them is entirely up to the authors' conscience. I came up with it especially for the purpose of connecting indicators to neural networks. In this case they are extremely easy to connect. But I think the idea has its own value. I won't have to deal with a new indicator written using this standard, its curve is immediately understandable. Otherwise, it may often happen that you see an indicator on the Internet. There is no description of it. And even if there is a source code, it is difficult to say what the author had in mind and what to do with it... Alas, such popular things as various moobs and Bollinger lose their right to exist with such an approach. But no one promised it would be easy... The advantages of such a standard, it seems to me, many times outweigh its disadvantages.
Three months.
Yeah, I read it, chuckled at some of the posts claiming an unconditional win :).
Official announcement - July 19, end of registration - September 25. 2 months and a week. In principle, with a working idea, it would be possible (if it does not require thousands of lines to implement).
Someone will win for sure).
I'm going to digress from the topic a little bit, although it's also quite in line. While driving from work today, it occurred to my dumb brain that we should all reconsider our attitude to indicators, especially in terms of their application in neural networks, but not only in them. In short, I want to start from the idea that an indicator is not a small tool for screen decoration, but a tool that helps to trade. In my opinion, the best way to help this process is to estimate the probability of price movement up or down by a certain amount of points without any wisdom. Therefore, let us consider that the indicator is a number (or rather the function of price series) that changes from +1 to -1. The sign of this number shows the supposed direction of price movement - '+' up, '-' down, while the module - the probability of reaching a significant amount of points in this direction, for example 30 (it will be better to make it an obligatory indicator parameter). I.e. all indicators have a unified and uniform interface. What they have inside them is entirely up to the authors' conscience. I came up with it especially for the purpose of connecting indicators to neural networks. In this case they are extremely easy to connect. But I think the idea has its own value. I won't have to deal with a new indicator written using this standard, its curve is immediately understandable. Otherwise, it may often happen that you see an indicator on the Internet. There is no description of it. And even if there is a source code, it is difficult to say what the author had in mind and what to do with it... Alas, such popular things as various moobs and Bollinger lose their right to exist with such an approach. But no one promised it would be easy... The advantages of such a standard, it seems to me, many times outweigh its disadvantages.
Good point! Unfortunately, it has too many chances to remain an idea. Although in fact the question seems to be simple.
There is an indicator with its own range of values. There is only one input parameter - the N value in points of price change. We need to build a new indicator the area of definition of which will be the area of the old indicator and the area of values - range (-1,1). And the meaning of these values - the probability of the price change by N points in the corresponding direction, as described by eugenk1.
The theoretical solution of this problem can be implemented only for each indicator individually and is hardly possible. Therefore, this question probably shouldn't even be discussed. But the phenomenological solution (again, for each indicator separately) may be implemented. To do this we just need (just do it :-) to analyze statistics of this indicator on the history. What is missing - the correct, from the point of view of mathematical statistics, formulation of the problem. Unfortunately, I am not strong in this field.
Eugene, maybe you could formulate a universal procedure for assessing this probability for each value of some indicator ? Or of some particular indicator?
My question about perceptrona was wrongly formulated. I'll try to reword it: why do you use a linear combination of AC (plane) instead of the conditions if(a1<x1 && a2>x2 && a3<x3 && a4<x4) that describe a polyhedron.
Although, for you - dumnya all the same it is not clear, but describe the principle of the linear filter:
In a linear filter, the equation of the plane is given as a linear equation (a function, as in a perceptron). For example, for a three-dimensional space with X, Y and Z, this would be an equation of the form:
simplified:
f(X, Y, Z) = 0;
- f(X1, Y1, Z1) = 0; If the function is equal to zero, the point with coordinates (X1, Y1, Z1) belongs to the plane or, as they say, lies in the plane.
- f(X2, Y2, Z2) > 0; If the value of the function is greater than zero, the point with coordines (X1, Y1, Z1) does not belong to the plane, but lies on one side of the plane.
- f(X3, Y3, Z3) < 0; If the value of the function is less than zero, the point with the coordinates (X1, Y1, Z1) does not belong to the plane or, but is on the other side of the plane.
The last two inequalities are the principle of a linear filter (not a smoothing one, as Integer bubbled). And there is no more primitive way to determine geometric position of points relative to sides of the plane, except substituting their coordinates in the linear equation of this plane and comparing the result to 0. In this way the linear filter separates flies from cutlets. And weight coefficients in perceptron, as I've already repeated more than once, are not some incomprehensible figures, but constants, using which the linear equation of the plane separating one objects defined by point coordinates of their features values, from other similar objects is given. Since a plane in any 2-x or more dimensional space has only two sides (in two-dimensional space not a plane but a line), consequently, objects can be divided into two classes only by such a filter.You can also read all this on page 172 of "Handbook of Mathematics for Secondary School" (c) by A. G. Tsypkin.
But you probably didn't go to that very school, but picked up cigarette butts at bus stops during lessons? Now you pretend to be a hoser and try to pretend to be an "expert" on line filters. When in fact you're a lamer and a two-timer. And all the same sooner or later it would be found out, because being illiterate you will make a mistake on some leap. And once they figure out your bullshit nature, don't expect any leniency from the people around you. Your opinion will be ignored.
Folks, a little off-topic, but also quite in line. On my way home from work today I got a thought in my stupid brain that it would be good for all of us to reconsider our attitude to indicators. Exactly in the perspective of their using in neural networks, but not only in them. In short, I want to start from the idea that an indicator is not a small tool for screen decoration, but a tool that helps to trade. What is the best way to help this process? I think the best way is to estimate the probability of price movement up or down by a certain amount of points. Therefore, let's consider an indicator to be a number (or rather a function of the price series) varying from +1 to -1. The sign of this number indicates the supposed direction of price movement - '+' upwards, '-' downwards. And the module - the probability of reaching a significant amount of points in this direction, for example 30 (it will be better to make it an obligatory indicator parameter). I.e. all indicators have a unified and uniform interface. What they have inside them is entirely up to the authors' conscience. I came up with it especially for the purpose of connecting indicators to neural networks. In this case they are extremely easy to connect. But I think the idea has its own value. I won't have to deal with a new indicator written using this standard, its curve is immediately clear. Otherwise, it may often happen that you see an indicator on the Internet. There is no description of it. And even if there is a source code, it is difficult to say what the author had in mind and what to do with it... Alas, such popular things as various moobs and Bollinger lose their right to exist with such an approach. But no one promised it would be easy... It seems to me the pros of this standard outweigh the cons many times over.
Good point! Unfortunately, it has too many chances to remain so. Although in fact the question seems to be simple.
There is a certain indicator with its own range of values. There is only one input parameter - the N value in points of price change. We need to build a new indicator the area of definition of which will be the area of the old indicator and the area of values - range (-1,1). And the meaning of these values - the probability of the price change by N points in the corresponding direction, as described by eugenk1.
The theoretical solution of this problem can be implemented only for each indicator individually and is hardly possible. So, probably, this question should not even be discussed. But the phenomenological solution (again, for each indicator separately) may be implemented. To do this we just need (just do it :-) to analyze statistics of this indicator on the history. What is missing - the correct, from the point of view of mathematical statistics, formulation of the problem. Unfortunately, I am not strong in this field.
Eugene, maybe you could formulate a universal procedure for assessing this probability for each value of some indicator ? Or some particular one?
There is Bayes' theorem, of course, but only for independent events. It means that an indicator that will calculate the probabilities according to this theorem must take the information from independent sources. But all technical indicators produce values, dependent on the quotes, therefore, they are not suitable for the Bayesian approach.
Reshetov, you have to keep it simple. Naturally, we are only talking about estimating probability by frequency. That's what statistics is for - otherwise a theorist would be a completely useless abstraction. The variance, for example, we know how to estimate from a limited sample. And we know how to estimate the chances that the sample estimate differs from the population estimate by no more than a given value...