It's impossible to make money on Forox!!! - page 39

 
Shniperson >> :
And pine why all this mathematical arguments? The foundation can always interfere, it almost always does, and it interferes very severely.

A foundation, only a shadow of a shadow.

 

.

Mathemat wrote (a) >>

Could you be a little more specific here, Oleg?

.

.

Among the subclasses of adaptive automatic control systems are extreme control systems (ECS).

Extreme regulators were proposed in the early twenties and were theoretically justified in the forties. These regulators were designed to keep a certain parameter of functioning of a real object with natural extreme dependence of the indicated parameter on the input quantities of the object at an extreme level.
The extreme regulator and the object of extreme regulation constitute an ERS. Characteristic for BER are a priori unknown, usually relatively slow transformations (drift) of object characteristics. Therefore, BER from the very beginning developed as search systems in which the lack of a priori information was compensated by current information obtained in the form of reactions of the object to artificially introduced search (trial, test) influences. For example, the response of an oscillating circuit with resonance response can be measured at two or more frequencies simultaneously.
In the above-mentioned understanding of SER, it is assumed that the extreme output of an object is available for direct measurement. BMS also includes systems in which the extreme value is not measured directly, but is calculated from the measurement of some set of output quantities of the object.
SER includes a device for forming an extremum indicator (target function Q), a search organization device and control bodies. The search organization device includes elements of logical action. Depending on change of Q(t), it generates command signals, received by controlling elements, necessary for system approach to extremum of Q indicator.
The system operates as follows. Search (trial) influences are applied to the inputs of the object and the response of the object to them, manifested as a change in Q(t), is evaluated. Then, those influences u(t), which approximate to extremum Q, are determined. Then, the signals at the object's input are changed in the desired direction, i.e. operational impacts are applied. Further, we attach further search stimuli to the object inputs and determine those of them that bring Q nearer to an extremum. Then work actions are applied to the object, and so on. After passing the value U ek corresponding to extremum of Q exponent, it reverses the object input and starts oscillatory movements of the system around the point of extremum. Sometimes the search and working influences are produced at the same time (i.e. are combined). In some cases, random effects (fluctuations) of artificial or natural origin may be used as search signals.

A further generalization of the SER concept is possible, when instead of the target function Q we consider a functional calculated, in particular, on the predicted motion of an object. With this generalisation, BER becomes indistinguishable from search optimal control systems in general.

.

.

This is a very broad topic, with many pitfalls...

As for the proposed joo synthetics, I see the following possibility.

1. An object model (i.e. its transfer function) is set. A reasonable number of such models can be set.

(2) A sample signal with quite definite characteristics is used.

3. A superposition mixture G1=<P+S> and G2=<P-S> (not necessarily additive) of input stream P and trial signal S is formed.

4. Two (or more) copies of the model are fed G1 to one and G2 to the other in parallel.

5. The outputs of the models are fed to a phase discriminator.

6. Depending on the mismatch at the output of the phase discriminator, a correction is made to the sample signal.

7. Return to step 2.

.

It should be mentioned that there can be many variants of construction here.

And I will also mention a very limiting feature of the indicators in MT4: there are 8 indicator buffers. It's very inconvenient and sometimes you have to build the whole cascade of linked indicators to get the result.

 
Reshetov писал(а) >>

P...ck all you want, but dx is not dispersion or RMS at all, it is the distance (displacement) from one point to another as a function of time along any of the chosen axes.

see experimental data:

Brownian motion "through the eyes" of a digital microscope


I quote for the especially gifted:

"So, if in 1 min a Brownian particle moves on average by 10 µm, then in 9 min it should on average move by -10 = 30 µm, in 25 min by -10 = 50 µm, etc."

keep dumb, nerd :)

Even your lamer example for schoolchildren says that it measures: the standard deviation of the particle over time. You take all points that the particle visited in time t and for all of them you find the RMS, not for a single point: "the deviation of the distance of the current point from the starting point as a function of time".

Tracks of a Brownian particle are plotted on the screen of the monitor with the mouse and the distances between nodes of the polyline are calculated automatically.

Don't you understand the difference between predicting the position of a particle and the average placement of its whole trajectory?

You can use 3 sigma rule to estimate, for instance, the range from which a particle will not leave in a certain time. Take a coin, heads=+1, tails=-1 and cumulative sum. In 100 tosses the particle has more than 99% probability of not getting beyond +-30. I.e. not one point in its trajectory, not a single one. In 400 throws it will not go beyond +-60. It helps to specify confidence interval in which the particle will be at time t with required probability, but the most probable value of the particle after any time is 0, if we predict at the beginning before straying. One can also calculate the probability that the particle will move beyond a certain boundary at least once in time t, but not where (or at what distance) it will end up after time t.

So no formula predicts the distance of the current point from the starting point as a function of time. You are imagining this due to your complete lack of knowledge of the subject ;)

 
Mathemat >> :

Starting with Einstein and Wiener, the highbrow know very well what Brownian motion is. This does not help them predict it. The specifics of the Wiener process is that it is a random process, not a deterministic function.


What a Brownian process is, they know. But what a market is, they don't. That is the difference. Besides, you can make a fortune on a Brownian movement, but with demolition. And the highbrow know this too.

 
Avals >> :

keep dumb, nerd :)

Even your lamer example for schoolchildren says it measures: the standard deviation of a particle over time.

Boy, learn the math.


It says that it calculates the standard deviation from the initial coordinate - the distance, the distance. The RMS (the square root of the variance, which you insist on with donkey-like persistence) is the standard deviation from the arithmetic mean.


Avals >> :


You can, for example, use the 3 sigma rule to estimate the range from which a particle will not exit in a given time. Take a coin, heads=+1, tails=-1 and construct a cumulative sum. In 100 tosses the particle has more than 99% probability of not getting beyond +-30. I.e. not one point in its trajectory, not a single one. In 400 throws it will not go beyond +-60. It helps to specify confidence interval in which the particle will be at time t with required probability, but the most probable value of particle after any time is 0, if we predict at the beginning before straying. It is also possible to calculate the probability that the particle will go beyond a certain boundary at least once in time t, but not where (or at what distance) it will end up after time t.

Why do I need the botanical rule of three sigmas or cumulative sum, when all this can be calculated quite acceptably by Moivre formula (see Moivre-Laplace theorem) or more exactly by binomial distribution (more general case by geometrical distribution)?


Why the fuck cut out the tonsils through the anus, when the entire mathematical apparatus has long since been established and described in probability theory books?


Besides, to measure boundaries beyond which a point will not go is not quite true for wandering of a particle according to Bernoulli scheme, because even at symmetric wandering, a particle will behave asymmetrically in time, according to the law of arcsinus. I.e. it will spend most of time on one side relative to initial coordinate (coordinate axis).


In fact, I repeat again for special gifted persons, that Brownian motion has no relation to trading, because it is strictly a physical process where such characteristics, as, for example, dynamic viscosity of medium, particle radius and diffusion coefficient are taken into account. None of this is present in trading. Not mentioning the fact that the shift of price occurs relative to only one coordinate axis, i.e. time may be shifted only to the right and strictly proportional to time, while in Brownian motion, a particle moves relative to all coordinates available to it. In Brownian motion a particle interacts not only with the medium in which it is situated, but also with other particles. In contrast to the price, a Brownian motion particle has no spread and no gaps.


In general, to discuss Brownian motion in relation to trading is a clear manifestation of nerdism.

 
Reshetov писал(а) >>

Boy, learn your math.

It says it calculates the RMS deviation from the initial coordinate - distance, distance. And the RMS (the square root of the variance, which you insist on with donkey-like persistence) is the standard deviation from the arithmetic mean.

Nerd, I have explained to you several times where you are lying, and you still do not understand. Both in relation to Brownian motion and any other mathematical model of which SB is a model. It does say that it "calculates the standard deviation from the initial coordinate - distance, distance". But it does not predict "the deviation of the distance of the current point from the starting point as a function of time". Predict how far away a Brownian particle will be in an hour, or two from the start of observation? :)

I'm not going to tell you how to study the matter, I see it's useless ;)

Reshetov wrote (a) >>

In general it's obvious nerdishness to discuss Brownian motion as applied to trading.


I don't know why in the hell you started to talk about it here, especially without grasping the elementary things

 
Avals >> :

>> Rest!

 
You should not be getting into physics with your abstract approaches. The subject is of course accessible, if you have the right approach and the right experience. Which you do not have. You see, in physics, as well as in programming, everything is without fools. If you do it wrong, it will not work. Unlike mathematics :)
 

Hello! I would like to contribute my views on the theory of probability and chaotic motion. First I would like to analyse the above to Avals.

"No formula therefore predicts the distance of the current point from the starting point as a function of time. You are imagining this due to your complete lack of knowledge of the subject ;)" "You don't understand the difference between predicting the position of a particle and the average placement of its entire trajectory?"
Don't take so verbatim the physical movement of molecules and the movement of the market price. We can only compare chaotic movement i.e. change of direction.

Let's keep it simple. We open a trade for a change of direction of a particle. We do not know where it will go next, like in the market. But we can invent a theory (system) according to which we will trade. For example - the particle moves chaotically and constantly changes direction like the price. We may suppose from this that the particle cannot move in one direction like the price. Therefore the longer the particle moves in one direction, the more likely it is that the particle will turn in the same direction as the price. How much distance is another matter, but in terms of reversal it is quite predictable in chaotic motion. The distance of the path is almost impossible to predict, we can only limit in advance by the system, by averages in the history of the movement.

Any movement in the market is a random movement for us mere mortals who do not work for the US national bank and have no information about upcoming currency movements. We have scant information and insight into how a particular currency pair will behave.

That is why it is better for us to consider the market as a random process and watch the market as a random movement. Moreover, there are clues on the market, which together with the chaotic movement theory gives more results than any system based on indicator readings. The clues - for example the price comes back to its peak, draws a spike the price slows down, the accumulation happens and the reversal is provided at 75% with a stop above the peak and a profit after a precipitous fall. Nothing can be predicted accurately. And in the market all the more so.

But how do you turn the chaos to your advantage? How to behave? One can train oneself on the movement of molecules or study the construction of the universe or find the origins of the market in the story of the early Templars. Who rules the world and the market? One thing is certain, we see the market, we see the movement and we see the losses.

Do you ever ask yourself a question, why when we open a deal, we analyze, think, draw indicators, wait for time, waste our nerves and eyesight and press the BUY button, but at the same time our brokerage company opens the opposite deal, wasting seconds and winning in the end?
I
know the answer, don't you?

 

Come on...

Don't stop, buddy.

We're waiting for you in that thread and you're here.

You promised us three indicators, remember?