Thoughts on the random - page 16
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 characteristics of noise do you need for trading purposes? There is no need to know the specific characteristics of noise. Well, maybe you want to define these characteristics? Then the question - why?
And notice that for some reason you don't raise the question, or even mention it, of signal characterisation. But it's the signal that counts. And noise is noise... ;)
Don't lie, what characteristics noise has is well known - the distribution density and correlation function at least. And they can even be very different for different noise. Take "thermal" and "impulse" noise, for example.
Let's take a simplified example, but it's very close to the real market situation. Let the market be an oscillating link with the parameters w (natural frequency) and a (damping). Both are not constant quantities, but random processes (let them be Gaussian) with correlation time Tw, Ta >> 1 (in other words, we have a loop with "floating parameters", i.e. a quasi-linear system). An additive mixture of two processes is fed to the loop input:
1. white noise n(t) - this will be such a "fraction", thermal noise
Generalized Poisson flow of pulses L(t) with intensity l also "floating": Tl >> 1. This will be a flow of fundamental data (news, for example) coming into the market.
At the output of the system we have a price - the result of processing two noises by the loop (absolutely random processes, no determinism, mind you. However, I am not saying that the output cannot be predicted in some areas, rather the opposite!)
The first question is what characteristics of the output random process is, and the second question is what should be considered a signal (note that deals are made not by the price of "signal" but by the price of "signal+noise").
This is a far-fetched model that has nothing to do with reality. It can, with some stretch, be called a "direct" problem - there is an attempt to fit the real input process to an a priori given basis. There is no justification for this choice of basis, nor can there be. But I consider the "inverse" to this problem.
Why, it is a very real model (only simplified) which describes the market as a physical system in a real external environment, not as an unknown ...arya ...nything. You would not deny that the market is a system with a transfer function, would you? Or that the market is influenced not only internally (fluctuations in the mood of traders), but also by external factors? I merely gave (again, the simplest and most natural) example of such a system. (If you want more complicated, you can construct a system consisting not of one but two oscillating units, then at certain ratios of their parameters the output would be a spherical Elliott theory in a vacuum, with waves 5-3, etc.).
And in general, the "inverse" problem is always inextricably connected with the "direct" one: it makes no sense to build an adaptive filter without having in mind the physical meaning of its characteristics. This follows at least from the fact that AF (in the broadest sense of the word) essentially simulates the behaviour of the system, trying to get an output similar to the real one, i.e. it is de facto a market model. If we have not made the model assumption, then we cannot build an adaptive device either.
Why, it is a very real model (only simplified) which describes the market as a physical system in a real external environment, not as an unknown ...arya ...nything. You would not deny that the market is a system with a transfer function, would you? Or that the market is influenced not only internally (fluctuations in the mood of traders), but also by external factors? I merely gave (again, the simplest and most natural) example of such a system. (If you want more complicated, you can construct a system consisting not of one but two oscillating units, then at certain ratios of their parameters the output will be Elliott spherical theory in a vacuum, with waves 5-3
, etc.In general, the "inverse" problem is always inextricably linked with the "forward" one: there is no sense to build an adaptive filter without having in mind the physical meaning of its performances
.This follows at least from the fact that AF (in the broadest sense of the word) essentially simulates the behaviour of the system, trying to get an output similar to the real one, i.e. it is de facto a market model. If we haven't made the model assumption, then we can't build an adaptive device either.
Do you have the results of this "just as real a model"of yours? Do you? Show me.
And I have the results of my model, which to you appears as "unknowable ...ah..."
And you are mistaken about the relationship between "forward" and "backward" tasks.
Do you have the results of this "just-as-feasible-model" of yours? Do you? Show me.
There is, I show you (pictured is a detector of those very pulse impacts, based on a model of ... er... with pulse shocks. It's pretty simple.)
And I have the results of my model, which to you appears as "unknowable ...ah ...nything"
And you're mistaken about the relationship between the "forward" and "backward" tasks.
taki is not a fact)) at least I don't see any difference in our positions. Although... the model I described above is quite writeable (and writable:) as a difference equation, i.e. practically perfect for building an adaptive filter like any quasi-linear model. The inverse problem in such a case is to screw up an adaptive filter, and then think about how to interpret its coefficients. Or not to think, and consider it a black hole and hope that it will work without failures. But I don't like it that way, I prefer to know the meaning of the money-making machine, and consequently where it may fail.
There is, I show you (the picture shows the same pulse detector based on a model of ... er... with pulse shocks. Simple enough)
And how do you use it? What do these impulses do? Are there positive results?
Practice as a criterion of truth
So how do I use it? What do these impulses do? Any positive results?
Well, how can I tell you, if it all already worked in real life, I would hardly discuss it here))) But it will work.
And the discussions are where I get fresh ideas plus verbalise my own thoughts. By the way, the last couple of days the thought has gone further)))
If it's about what the pulses do, then see my question to Alexei the mathematician above and his answer.
There is, I show you (pictured is a detector of those very pulse impacts, based on a model of ... er... with pulse shocks. Simple enough)
Oh, I have a similar one. I have my doubts.
doubts
I have a new version every 30 minutes now, so I'll make it up to you. And doubt is good)))