Hearst index - page 38
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
I have attached the text. Unfortunately the formulas and theoretical part didn't turn out.
But the FARIMA fitting process is specifically described
The ACF on the charts is strange. It doesn't look like a market one at all.
The ACF on the charts is strange. It does not look like the market one at all.
It seems fine. I've seen plenty of these for the original cotier, not its variations.
For me, the question is different.
The issue is the accuracy of the model.
The absolutely accurate model is the quotient itself.
You can take a very imprecise model in the form of a straight line. And then a curve, and then something else to consider ...... Where to stop. The more accurate the model, the more it is bound to the data used to build it (fitting). Hence. Do we need a model like FARIMA? Maybe it's redundant? Smoothing, ARMA for the residual, and maybe (not necessarily) ARCH for the residual?
You can identify this very long memory in any quotes. But ACF is not appropriate here.
Peters gives an interesting definition of long memory. Read it. There is a lot of interesting information in his books on this subject. According to him such processes cannot be measured by trivial ACF. ACF works on a scale of 5-6 lags and that's it. If H is expressed as a moving particle with spread equal to square root of distance, then we get a special case of normal distribution StdDev = Sqrt(T) = T^(0.5). So, if the scattering of the particle is slightly above or below 0.5 it is possible in one and only one case: the particle must remember its past state and hence such process will possess a memory. I.e. H is not a characteristic of external action, but depends on previous state of the process. And if the escape trajectory is preserved, it means that it depends on previous values and the memory period can be calculated. And it often happens that on all scales of calculations the slope angle does not change, and at the same time, it is not equal to 0.5. In this case, the process is said to be a true Hearst process with infinite memory. Except the ACF won't show any such thing.
I can't agree with the highlighted one. There are two alternatives about what is the carrier of information about the past: either it is the particle that remembers its past state, or it is the environment that remembers the particle's state. Given that the external environment is usually a much more complex object, and the particle can be a material point altogether, I would rather agree with the second option.
I can't agree with the highlighted one. There are two alternatives about what is the carrier of information about the past: either it is the particle that remembers its past state, or it is the environment that remembers the particle's state. Given that the external environment is usually a much more complex object, and the particle may be a material point altogether, I would rather agree with the second option.
If we consider straying of electron in a crystal lattice of silicon it will be expressed by the law with H=0.5. But if we add to the crystal n or p admixture we obtain quite another character: the so called anomalous diffusion for which H will essentially differ from 0.5. Obviously, in this case H is exactly a characteristic of the medium. By the way, the impurity atoms are distributed through semiconductor volume statistically fractally, so such wandering is also called wandering on fractal.
What effect would that have in our case? I mean, what difference does it make to us whether the memory is stored in a material point or in some external environment?
The difference is probably in the approach to modelling: in order to build a theoretical model, it is nice to have some fundamental explanation of what is going on, so as not to point fingers in the sky. If we believe that some factors are a result of external environment, we will look for them there (based on our life experience), and having found them, we will be able to offer a more adequate model of influence. If the factor is internal, then we will use some considerations about the internal structure of the system. In other words, internal and external forces are described by different equations, and it would be nice to know which one we are dealing with.
Specifically.
We have autoregression. Strictly kotir. Nothing externally. This is an internal factor?
We have a regression, according to which our quote is modeled on the basis of other quotes, e.g. EURUSD= GBPUSD+...... But these are so to say homogeneous variables. Is it an external factor?
Now we add the time of day to the regression and model activity depending on the time of day. There may be a lot of such "external" variables. And this is completely external?
I don't see a place for particle theory and external environment.
It is there, but who is going to do it here?
No, there is no such theory.
You have to go from a verbal description of the model.
And these are economic processes in a large variable number and variable interrelationships. Cotier is the realization of this process. It is not a Brownian process in which a molecule moves and collides, i.e. it is an independent object with its own properties.
Specifically.
We have autoregression. Strictly kotir. Nothing externally. This is an internal factor?
We have a regression, according to which our quote is modeled on the basis of other quotes, e.g. EURUSD= GBPUSD+...... But these are so to say homogeneous variables. Is it an external factor?
Now we add the time of day to the regression and model activity depending on the time of day. There may be a lot of such "external" variables. And this is completely external?
I don't see a place for particle theory and external environment.
Regression can be built on anything, and this method is called the rule of thumb. The question is whether we can say in advance that out of many possible regression models, this one will better describe the behaviour of the quotient for some reasons. Describe these reasons mathematically. Write a difference equation, calculate the regression coefficients analytically - so that it is clear which ones represent the influence of external factors, which ones characterize the internal properties of the system, and which ones combine internal and external factors.
Try, for example, to construct a difference equation of one of the simplest systems - an oscillating circuit. In regression terms this will be an ARMA model and its coefficients will be a combination of parameters of the circuit itself and the input signal:
Y(k) = 2*a*cos(w0)*Y(k+1) - Y(k+2) + X(k) - a*sin(w0)*X(k+1)
Here X is the unknown external influence, Y is the observed response, a is the damping parameter, w0 is the natural frequency of oscillation