Stochastic resonance - page 5
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
Yep, there's already a coalition of participants who believe that 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 appears precisely from the noise of the flat itself, which keeps the market in a state of constant readiness to collapse.
Alas, I can't possibly articulate it. I read Peters (again) about market fractality and I agree with him that the normal state of any stable system is non-equilibrium. Here the property of self-similarity of the fractal agrees with the presence of the investor on the horizon of any duration, and nonlinearity and asymmetry in the decision-making, and many other things.
but how would one calculate the Fractality of a series? You promised to post an algorithm... :)
The paper proposes a dynamic model including a noise component, which allows generating quasi-chaotic time series simulating a phenomenon called "stirring layer", i.e. the "chaotic behaviour - bistable mode (jumps between two significantly different states) - selection of one stable state" scenario . This scenario is typical of many processes in economics, medicine, etc. A method for the analysis of generated series, based on the study of some statistical characteristics, is also proposed. It is shown that the analysis (with preliminary creation of a training set) allows to define "the moment of truth", i.e. that moment in time, at which it is possible, with the set probability, to predict, which stationary state the given system will choose. http://ellphi.lebedev.ru/12/pdf19.pdf
Good luck.
but how would you calculate the Fractality of a series? You promised to post an algorithm. ... :)
Haven't accepted it yet, although the algorithm has once again been confirmed for myself. Recently I read about variation index, very interestingly written, especially when you consider that the authors of the algorithm claim that it requires much less data to calculate this index than to calculate Hirst. And Hurst index, fractal dimensionality and variation index are closely related.
INVESTIGATION OF STOCHASTIC RESONANCE EFFECT IN A BISTABLE SYSTEM
V.N.Ganin, A.A.Dubkov Nizhny Novgorod State University
In this paper a new approximate method for investigation of stochastic resonance effect in a bistable system with piecewise linear potential is discussed.
We consider the motion of a Brownian particle in a potential field, something analogous to Candida's second link ? http://forex.kbpauk.ru/download.php?Number=16275
Good luck.
II Congress of Biophysicists of Russia
http://www.biophys.msu.ru/conferences/99_bpii/10_OBZOR/10_Otchet.htm
well there yada yada ( biophysicists, no offence) and here already formulas appear. ...
The phenomenon of stochastic resonance, observed in various systems, with threshold activation, under the simultaneous influence of noise and a coherent, usually periodic force, has recently attracted much attention. The possible role of stochastic resonance in biosystems was first pointed out by V.Y. Makeev. Under certain conditions, increased intensity of external noise leads to more orderly behaviour of the system.
Stochastic resonance is a cooperative effect in non-linear systems in which noise energy, distributed over a broad spectrum, is pumped into output energy at the signal frequency. In this case the amplitude of the system response is described by a resonance-type function in which the argument is the noise level.
Its mechanism is approximately as follows: in presence of noise a particle makes transitions from one state to another; characteristic time of such transitions is determined by Kramer parameter. In the presence of deterministic modulation the barrier height begins to depend on time and the ratio of transition probabilities in phase and in antiphase with forcing becomesW+/W-=exp(-2QD), where Q is the barrier height and D is the noise intensity.
As the noise intensity increases the Cramers time decreases. - The less is the volatility, the more probable is transition from trend to flat ?
If the forcing force changes slowly enough it is possible to reach a regime when Cramers time becomes the order of a period of this characteristic change time but theW+/W- ratio is still quite high. Then the transitions in the system are sufficiently reliably modulated by the signal and we deal with stochastic resonance. At larger D the Cramers time becomes too small compared to the characteristic modulation time, and W+/W- ~ 1, and stochastic resonance is not realised.
Stochastic resonance makes it possible to amplify signals with an amplitude much smaller than the noise intensity at the expense of the noise. First of all this possibility is interesting in connection with the kT-problem, the essence of which boils down to the question: "can an effect with a characteristic energy that is less than the average thermal background energy (kT) have any biological significance". In particular, scepticism about the possibility of exposure of living tissues to weak electromagnetic waves is based on such arguments. We consider a simple model of a membrane channel, the switching between conduction levels in which can be modulated by a weak external signal.
....especially I liked passage about transition times, i.e. there appears possibility to READ the duration of trend ? although what is Kramers parameter, as it turns out, it is an average time needed for escape of Brownian particle from potential well. I have not even mastered the first work, the second link Candida, and here again ... Wikipedia is crying from my queries and I'm crying from my stupidity, well, let's go read on.
Good luck.
Question
I read all literature on stochastic resonance and became even more convinced of the correctness of my approach. One of important conditions for existence of stochastic resonance is existence of two "stable" states. Accepting the model: trend as a transition from one level of flatness to another, we obtain that the stable states are two levels of flatness. We can tell more or less confidently about one level, but the second one remains a big mystery. Perhaps I do not understand something completely, or do not understand it at all, but it seems to me that the search for the potential function of the model for our case is absurd. To know this function means to know practically everything about the system or else to find the "life formula". To find the second potential energy minimum based on signal parameters, noise and one level of potential energy minimum also seems to be difficult, rather impossible.
I still maintain that all that can be done practically is to find the optimal noise characteristics at which one can confidently speak of a future directional spike with a certain probability, but one can only determine a new level by empirical dependencies derived from typed statistics.
I believe we need to move from collecting statistics on noise, trends and flat levels to looking for patterns, taking into account the specifics of stochastic resonance. Intuition still says there should be such patterns. I wrote about it earlier "grasn 12. 10. 2007 14:08". However, after reconsidering it I realized that indeed what I said was very similar to volatility, but what I meant was noise parameters in terms of digital signal processing, which is something else entirely. Earlier when collecting statistics I forgot about noise, while noise is a very important component of the system and must not be ignored.
I'm getting a little carried away here, but here's the question - how do I calculate the noise intensity? I searched my books and internet, I couldn't find anything. There is a Relative Intensity Noise (RIN) parameter - But this is calculated for lasers and similar systems.
Trend or Flat
What is a steady state, trend or flat? In my humble opinion, this is just terminology and some agreement of views among the participants. The authorities teach us that the market mostly sits in a flat and spends very little time in a trend. After my own, trivial experiments I've come to another conclusion: local (and there are no others) flats and trends exist in approximately the same proportion. I am going to show you, as an occasion for reflections, the first segment of EURUSD (hours, (H+L)/2) that has come to hand. The algorithm for gathering statistics is simple, I "step with the times" in the tested interval, fix the length of the initial time series, look into the future at every interval and determine the duration of a sideways channel (flat) and a linear regression channel, of course, using the parameters of the same initial sample. This is what I got for the window of 600 samples:
The x-axis represents the samples of the analyzed range, while the y-axis represents the channel length reduced to the window size of the time series (i.e. lifetime - 2 means that the channel with the initial parameters lived, two more initial lengths, 2*600). If we take the whole history, and go through the window lengths, we still get approximately the same picture (almost like in the figure). The average duration of the "flat" channels is slightly longer than the linear regression channels, but none of this is "significantly" what the authorities are writing about. Of course, the argument is circumstantial, but it has led me to some thoughts.
Potential pits
There is an article, see 'Mapping support and resistance levels'. There are references to previous publications. And there are Fibs, you just need to find them. With Swaney's approach you won't find Fibs.
I am reminded of a fragment of dialogue from the film Hitchhiker's Guide to the Galaxy, a robot with a manic depressive personality and the main character: the worker, touched by the protagonist's distress tries to help, "you want me to calculate your chances of survival, .... but you won't like it...". The same chances of using these "potential pits". Paradoxically, the price doesn't care which of these pits is "more potential". Once you "measure" these pits, you'll never guess which one the price likes better. The point is that the resulting curvature has nothing to do with the potential function.
I must be missing something in the first paper, because the result seems rather trivial to me - as the noise decreases, cases of exceeding the transition threshold become rarer and finally stop altogether - the system then retains the state in which it finds itself at that moment. Why do the authors call this a prediction? Again naming the first phase as a stochastic resonance suggests a simple idea - the authors simply do not know that this term is already used for a completely different phenomenon. That is to say, I think that the review in the introduction and the list of references are of value in this paper. As for other two references: they have not changed my opinion - stochastic resonance is quite a narrow term, the key point (facilitating calculations) is cyclicity of the signal, the market will not indulge it. I should note, however, that the dynamic part is an integral part of the models. Therefore, I still think that I should start with it :)
P.S. This was from the literature.
For literature, it is better to read this: http://eprint.ufn.ru/article.jsp;jsessionid=aaa81x5hHOgj8Y?particle=1784
(from wikipendia link)
все, что возможно сделать практически – это найти оптимальные характеристики шума, при котором с определенной вероятностью можно уверенно говорить о будущем направленном скачке, а вот определить новый уровень можно только по эмпирическим зависимостям, выведенным на основе набранной статистики.
It is possible to speak confidently about a future directional jump at almost any point in time :), the question is its direction. And it is not a bad idea to have at least a rough idea of the time. The nearest new levels (from above and below), imho, are rather well defined by traditional means of TA, having guessed the direction it is possible to specify further, after all there is a trailing.
I think a considerable part of potential criticism can be explained by grasn's bad mood. About labour-intensive - who has it easy nowadays? :) By the way I've once wrote that I've frozen work with potential model just for this reason :)
For literature, it is better to read this: http://eprint.ufn.ru/article.jsp;jsessionid=aaa81x5hHOgj8Y?particle=1784
(from wikipendia link)
I've looked at Wikipedia, and quite deliberately don't want to delve into "SR" - as I don't see its usefulness to this case yet.