Market phenomena - page 27
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
"Inertia is the phenomenon of a body maintaining its speed of motion (both in magnitude and direction) when no forces are acting on the body" :)
Why are you actually surprised by the presence of a global long-term trend?
What does inertia have to do with it?
Why not suppose that such a matrix is not a cause but a consequence of these two processes? If they are present, of course.
Perhaps, this is what I want to check.
(2) A command is given to make an accurate forecast. In progress:
- identification of the "on" current structure
Colleagues, I will be leaving the forum for an extended period of time.
Good luck.
Vladimir, what do you mean by adding up the system? I too am wondering how many steps it takes to fold it - only 4.
What does inertia have to do with it?
Well, didn't you draw a graph of uniform rectilinear motion? Then see the definition.
... each process has its own hierarchy ...
... But if the Markovianness is not respected, then things will be much more complicated, you'll have to reinvent them :o(
Well, finally, at least someone has revealed the secret, so to speak, the phenomenon of the market.
I can add more: in a series of black candles new bars are opened by white ones for some unclear reason, but they are closed by black ones, and vice versa for white candles.
I don't understand how a candlestick is formed at the opening.
Yes, cumulative BP (for this example) . Again (used my post from another thread and modified it slightly):
Market model
After much searching, adopted this "control systems with random structure" thing as a working version of the market model. In my opinion (though not mathematics) - this model adequately describes the quoting process with all its subtleties.
Its essence is very simple. There is a finite number of structures that describe the transformation of input into output. Each such structure implies some model according to which the transformation takes place. The observed process is formed by a transition (switch) between the structures. All this is shown in the picture below:
Each model has a set of parameters, which can also change at each switching. So, I assumed that there are two main processes, each process has its own hierarchy, each element sitting at a node in the hierarchy has its own structure.
Process interactions
These two processes compete with each other according to the transition matrix (presumably), i.e. there is an "external" (conventionally of course) to the market some system that switches the generation of quotes between these processes. Later, I will show in more details, with reference to
Adaptation to practice.
Everything is great - but it's impossible to exactly identify such a system. Therefore, I introduce "combined model": A=W(1)MODEL1(parameters)+ W(2)MODEL2(parameters)+....+ W(n)MODELn(parameters). Where W(n) are some weights of participation of these models in prediction. It may be possible to explicitly partition the processes due to the invented transformation. But that's for later.
What am I working with?
I don't work directly with quotes - it's an extremely complicated process. I introduce all sorts of tricky transformations, but what I have said also applies to them. The complexity is not going anywhere - it is inherited. You cannot simplify the process. And if you do simplify it, you can lose the process itself. (i.e. even a little more complicated than I described, but I have shown the phenomenon and some more interesting observations)
Analysis of time series evolution
Basic stage. At this stage, I identify all possible structures by some criteria. I estimate statistics of transitions between these structures. I determine a transition frequency matrix for the structures. In the future, I am thinking to use the so-called impulse neural networks (or wave networks). It is a very promising direction.
Algorithm
(1) By making some assumptions about behavior, a probabilistic estimation of the future state of the system at the given moment on the planning horizon is performed. The neural network crawls through the resulting probability assessment matrix p=f(time,cotir) of the initial state and in turn makes an assumption about the presence of an entry/exit point. It can very accurately tell whether or not there will be an entry/exit on the planning horizon. All that remains is to find it.
(2) A command is given to build an accurate forecast. It is performed:
- identification of the "on" current structure
- assessment of the choice of the most probable future structures
- Identification of parameters of future models
(3) A simulation is run
(4) Next, a neural network estimates the coefficients of the combined model.
No randomness has already been found, proof of this is the extensive research by Alexey (Mathemat). I confirm them, everything is correct. But if the Markovianness is not respected, everything will be more complicated and I will have to reinvent it :o(
Picking up a topic, maybe off-topic... Tried to distinguish frequencies in artificial random numbers - by fits in and out of RMS range - then read Farnsworth's post
realised that there is a "sieve", the mystery of which remains, and what I was doing gives neither alpha nor omega.
It's all about the "sieve". What is it and how is it? There are more questions than answers...