Stochastic resonance - page 3
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
What do you mean by "steady states"?
Of course, we could divide the tasks. But then we have to look for an answer - Who benefits from it? But that sounds like a childish question. Although I could be wrong.
Let's hear what the participants have to say... Intuitively, it seems to me that this is a promising area and worth pursuing.
Of course, we could divide the tasks. But then we have to look for an answer - Who benefits from it? But it looks like a childish question. Although I may be wrong.
Let's hear what the participants have to say... Intuitively, I think it's a promising area and worth pursuing.
No one's arguing that. It's just that the task is not up to one person. Everyone has a part to play. But it may turn out that no one sees the "elephant".
If I understood the article correctly, then we must look for a constant source of influence. But it may turn out that there is no such thing. Or there are many of them, which is the same thing. So how do you go about it?
I have a strong feeling that I need to look for both, and all together is very frustrating. I'll start my journey with noise, especially as I've wanted to deal with it for a long time.
Wait, Sergey ! Not so at once. I need to think a bit more. :-))
From my point of view it's not quite like that. Resonance (especially between signal and noise) as such does not exist here. As far as I understand the physical meaning of the phenomenon: given certain parameters of noise, it brings the system into a state of specific excitation, in which a small regular signal is sufficient to initiate a more or less stable process. That is, noise, which also carries a certain amount of energy, excites the system to an over-threshold state. But because it is stochastic, a directed process cannot occur - there is no direction. At this point an insignificant steady signal is sufficient to initiate and continue the process. Thus, the main source of excitation energy is noise and the direction of the process is determined by the signal. It follows (if true) that of all noise-system interaction parameters, only one is important - the noise energy dissipation factor. If it is small, then the noise energy will be stored by the system and it will migrate towards the excited state.
In the market there is a quite adequate measure of "noisy" excitation - volatility. When the market is excited but does not know where to go, it goes up and down with the decent range staying in the flat. So it is not so difficult to determine the state when stochastic dissonance can occur. Moreover, playing at breakdown of support and resistance levels is an exploitation of this phenomenon and it wasn't born yesterday.
The situation with the stable signal is much worse. The Forex market is overloaded with signals. They, imho, are all incoming information - news, economic, political and other data... However, regularity, let alone periodicity, will be very difficult to find. Even regularly issued data has a completely opposite effect on the market, because, besides regularity, it is important the information - whether it is positive for the currency or negative. Personally, I see only one possibility for a regular and relatively unidirectional signal. When a new economic cycle starts, the news related to it have more or less homogeneous nature and influence to the market. An example is the cycle of interest rates going up or down. Or a cycle of economic ups and downs.
I don't think that all this can be used in forex. Not because the theory is wrong - on the contrary. It's because this theory gives nothing new to practice besides a better understanding of processes.
It's hard to say, Vinin. And the regular exposure itself can change abruptly and in leaps and bounds.
P.S. Again I got the urge to refer to my own results of 10 years ago, when I tried to build a quasi-newtonian model of the behavior of a stock (with a stochastic bias).
Everything is just as it should be - generalized Newton's second law (and there is quasimomentum, and even mass - stochastic, which, of course, varies) with nonlinearity. But it was a parametric oscillator (the one that takes its energy from parameter changes).
I thought that everything is forgotten... but no, there is something in it, though it is still a long way from practice... It's all nonsense so far, even crazy in some places...
It's hard to say, Vinin. And the regular exposure itself can change dramatically and in leaps and bounds.
There is also the problem of the director, although each of us can claim this role. And task staging is something many of us have been trained to do. But being able to see the "elephant" is a little different.
But to put it simply. The problem will not be covered by one person. When solving the problem, everyone will only see his part and will not see the whole.
The example about the "elephant" was not given for nothing. How to solve a problem, the problem lies with a "conductor" who can see beyond a particular solution. This is another "talent".
And what does 'steady states' mean in your understanding?
Hi Yuri, good to see you!
All true, though I didn't quite get what's wrong with my approach. Choosing "stochastic resonance" as the basis for model building (and for myself as an extension of the model), assuming it is true is something of a null hypothesis. In order to reasonably reject or further develop it, research needs to be done.
Assuming that under certain conditions (or rather, noise parameters) a directional signal evolution can occur (you, in general, write that signals in forex are over the top) my first step I see quite simple: gather statistics on noise parameters, compare with "local trends" (a simple LR to start with) and slip the results to Data Mining. If dependencies are found, I'll continue to develop the model, otherwise I'll abandon the idea, it's not that important (I created my model). After all, everyone gets rid of noise, filters it, kills it mercilessly, while according to this model it must be treated with care and attention, it is an important component of the system as a whole.
Yury, you must remember my attitude to volatility, I wish I could add a few nasty expressions about it. As a phenomenon it is the best, but volatility as any estimated figure is a false indicator that has nothing to do with reality and in no case it should not be included in the model. But that's not a reason to continue arguing, it's just my IMHO. The point is not to take elements of technical analysis for a model, but to base it, for example, on the theory of digital signal processing, theory of random processes etc. - But it's kind of conceptual.
It is true, but I have managed to accurately enough (in the sense for automated trading) find local trends as a linear regression and I am very optimistic about it. Hopefully soon I'll convert code from MathCAD to MT and see how my astrolabe works to the fullest extent. At least the tests in MathCAD are good.
If it is not possible to create a complete model, then provided its adequacy, one can create a really reasonable indicator of possible transition from one flat level to another and it is quite possible to supply it with an estimate of transition probability. To say "no" unambiguously now - in my opinion, based on the nasty volatility, is not quite right, one must be objective about it.
to VininYes, given the "different", one is at least a year away, if of course the results at the beginning of the research are encouraging. And with elephants it's always difficult, but you can just start with the whole and eat it in pieces.
As project manager, I propose to nominateMathemat the tick. No self-withdrawal. :о)
Yep, there's already a coalition of participants who believe a movement (trend?) is a steady state. I would like to hear some justification, Rosh. That movement without justification to the market phase is an internal state of the market is understandable.
Personally, I believe that there are no steady states in the market. There are either quasi-stable (i.e. unstable but seemingly stable) or transitions between them (disasters). And the market itself is constantly on the verge of a nervous breakdown. And serious nervous depressions (1987, say) are normal.
Well, yes, I agree. And this instability in the light of the concept of stochastic rehonance emerges precisely from the noise of the flat itself, which keeps the market in a state of constant readiness to collapse.