The market is a controlled dynamic system. - page 157

 
airbas:

1) But the difference between heater and pan temperature is small compared to the full temperature range of the process, does it make sense to consider it?

2) Imho, if you consider the primary parameter - the amount of heat going into and out of the pan, the imbalances will be much better seen there.

3) In the market, volume is probably something similar - let's say if there is a cluster of limit orders in the path of rising price, then until all those orders are gobbled up, price will not move a single point further. Only the volume is not in ticks, and not in contracts, but in money (price*number of contracts).

4) I do not talk any more, I look what will happen next.)


1) You must also pay attention to the temperature difference gradient.

2) Ultimately we are not interested in the pot, but in movements where the imbalances are orders of magnitude smaller.

3) I won't say anything about volumes -- I don't use them at all.

4) Well why "shut up" -- I'll always try to answer reasonable questions. And you should not only look, but also do it - that's the only way to get an understanding.

.

.

At the end of the day, of the whole range we are interested in its working section

 

e.g. m15

 
yosuf:
Why can't we limit ourselves to analysis of the output signal, without reference to the input signal? Especially since, we will never know the nature and nature of its origin.


We make assumptions about the nature and nature of the input signal in order to be able to analyse the system in a linear approximation. For example, as already mentioned, in the case of the cooker we know that the controlling influence is a non-linear function, so the output signal - the temperature of the water in the pot - is also a non-linear function of time. But if we make an assumption about the character of the input impact based on general considerations (for example, we theoretically know that electric stoves switch modes in steps, i.e. the input is a piecewise constant function), then we can reduce the problem to the first linear approximation and use appropriate linear methods, which you obviously have not studied yet, and if you do not, you will still ask silly questions.

In the case of the market, which is by definition an open system, i.e. a system subject to external influences, the linear methods of "output only" analysis are out of the question, simply because the output signal depends exactly on an unknown, non-linear input. Therefore in such problem it is necessary to make two assumptions: about structure of the system as linear in the first approximation element and about character of the input action. And if the first question is rather complicated (although even here we can start with relatively simple 2-3-4 order models), the second is somewhat clearer: the most important controlling signals for the market are news and currency interventions, and both are in fact shifts of the equilibrium price level to a new value. In other words, here too, to begin with we are entitled to treat the input (control) signal as a piecewise constant function. Like a tile switch, only levels and switching times are chosen randomly (from our, observational, point of view).

In summary we have: 1. a system of a given order with a set of unknown parameters (coefficients of the difference or differential equation), 2. a control signal of a given kind with a set of unknown parameters (switching times and levels), 3. Known output signal. After that we find the unknown parameters by any known method, hundreds of them.

 

In the case of the market, which by definition is an open system, i.e. a system subject to external influences, no linear "output only" analysis methods are possible, simply because the output depends precisely on an unknown, non-linear input.

Not at all obvious, mind you.

 
Oleg, a request to you - a suggestion: make an analysis only of the output signal by the linear method. On the example of potatoes on a slab :)
 
alsu:

We make assumptions about the nature and nature of the input signal in order to be able to analyse the system in a linear approximation. For example, as already mentioned, in the case of the cooker we know that the controlling influence is a non-linear function, so the output signal - the temperature of the water in the pot - is also a non-linear function of time. But if we make an assumption about the character of the input impact based on general considerations (for example, we theoretically know that electric stoves switch modes in steps, i.e. the input is a piecewise constant function), then we can reduce the problem to the first linear approximation and use appropriate linear methods, which you obviously have not studied yet, and if you do not, you will still ask silly questions.

In the case of the market, which is by definition an open system, i.e. a system subject to external influences, the linear methods of "output only" analysis are out of the question, simply because the output signal depends exactly on an unknown non-linear input. Therefore in such problem it is necessary to make two assumptions: about structure of the system as linear in the first approximation element and about character of the input action. And if the first question is rather complicated (although even here we can start with relatively simple 2-3-4 order models), the second is somewhat clearer: the most important controlling signals for the market are news and currency interventions, and both are in fact shifts of the equilibrium price level to a new value. In other words, here too, to begin with we are entitled to treat the input (control) signal as a piecewise constant function. Like a tile switch, only levels and switching times are chosen randomly (from our, observer's point of view).

In summary we have: 1. a system of a given order with a set of unknown parameters (coefficients of the difference or differential equation), 2. a control signal of a given kind with a set of unknown parameters (switching times and levels), 3. Known output signal. After that we find the unknown parameters by any method we know, hundreds of them.

Of course, I'm not as smart as you, that's why I ask stupid questions, as you rightly pointed out. But, look at how you can get at market regularities by studying only the output signal, i.e. the price https://www.mql5.com/ru/forum/133986/page114, although to understand them you have to sink to our, stupid, level, and that's not easy. The perfect, computer-accurate, coincidence of actual and calculated price values is achieved by analyzing the properties of the non-linear function. Have you ever seen anything like it?
 
tara:
Oleg, I would like to ask you a suggestion: to analyse only the output signal by the linear method. On the example of potatoes on the tile :)


As I understand, you mean reconstruction of input signal by known output signal.

And then make a comparison of the reconstructed input signal with the known actual input signal.

Will do. But can no one else do this conversion? I would love to see feedback from the audience.

 

And then let's do this:

First, I will make more counts so that there is more "history", and so that to this "history" it will be possible to apply the MAH for more visibility of approach divergence.

Secondly, forgetting water and potatoes, I will make several switchable limit levels. Previously there was one level 100 on water. Now I'll make it three or five or seven.

Then you can introduce the noise component, which inevitably introduces distortion.

.

And then I will make the system itself control its further development, i.e. I will add to it a supersystem of goal setting, a superstructure of a higher hierarchical level. And there may be a large number of such hierarchical levels. And each level has its own objectives. To an outside observer, it may look like a random or chaotic movement, very much like the market.

.

maybe... at my leisure...

 
yosuf:
Of course, I'm not as smart as you, which is why I ask stupid questions, as you rightly point out. But, look how you can derive market patterns by studying only the output signal, i.e. price https://www.mql5.com/ru/forum/133986/page114, although to understand them you have to sink to our, stupid, level, and that's not easy. The perfect, computer-accurate, coincidence of actual and calculated price values is achieved by analyzing the properties of the non-linear function. Have you ever seen anything like it?


Yusuf, please do not be offended with him -- alsu got a little excited. However, his general description of the problem is correct. But in the heat of the discussion, so to speak, he got a little excited...

As for the search for market patterns, there are, after all, many paths. And, of course, one way does not preclude the possibility of moving in other ways.

 
Why look for inputs etc.)) buy on the rise near room temperature level, sell at 100. That's the whole system))