From theory to practice - page 502
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
But even the matstat has problems with the basis for the series of Forex rates, these series do not have the property of statistical stability (relative frequencies tending to the probabilities), and therefore the laws of large numbers are not satisfied. What are the criteria?
Well, first of all, there are some methods for dealing with sampling from differently distributed values in the matstat. Second, the assumption of invariance of the distribution of increments was made by the TS, not by the matstat. It would have been correct to check this assumption with standard methods.
the assumption that the distribution of increments is unchanged was made by the TS and not by the matstat. It would have been correct to check this assumption by standard means.
It has long been rejected, Alexey... Alas, if it had been - this thread would not have grown to such size.
But, for example, Orlov openly spits on this non-stationarity, accepting it as it is. He simply advises to exit the trade faster. And to enter using standard statistical methods of BP analysis with sufficiently large sample sizes.
Yes, that's right - get out of the deal quicker, without bringing matters to a tragic denouement and continuing your career near the station toilet...
How much faster? How much faster should the time window for deciding to get out of a deal be than the time window for entering a deal?
Do I know that?!!
It's long since been rejected, Alexei... Alas, if it had been, this thread would not have grown to this size.
Good that it is rejected, but bad that it is not done by the results of some criterion (like Kolmogorov-Smirnov). It would have been useful both for you and for us, the readers of the thread.
Same thing the week before last:
Man, why the hell didn't I listen to Wizard?
That's what a project manager means - he says do it!
Increase the window? To a week? We're almost out of deals as it is...
Ugh... I'm getting frustrated. Time to wash my hands of....
These are charts of binomial distribution as a function of p. Something tells me they are similar to your charts, Alexander, except the last one (it's almost a normal distribution).
Material taken fromhttp://hr-portal.ru/statistica/gl3/gl3.php
Alexander, what are the statistical characteristics of the floating sample distributions? Kurtosis, asymmetry, etc., can you show me a couple, the ones that are very scattered?
I can't yet. I'm away - I'll be there in a week.
Novaja:
Посмотрела, Ваша система построена на разнопериодных производных или на производных от производных?
Interesting turns of phrase, though... ;)
I will put it simply: it is a tracking system. Its task is to detect local and global movements (for the current TF). Decision-making block is a superstructure.
szy
Now I have figured out how to make the result even more simple and clear. The next example will be with these changes.
As promised, I am making it more vivid.
That's more clear. What do you think?
It shows 1st, 2nd, 3rd derivatives - v,w,r - velocity, acceleration, jerk (filtered differences of appropriate order).
you could sign them in detail and make the lines in different colours. What do you think?
zy
Why did you delete your posts?
So you have the MA on large timeframes as a job for a discrete PID controller, or what? You get a big mismatch, you open a trade?
So you have the MA on large timeframes as a job for a discrete PID controller, or what? You get a big misalignment, you open a trade?
1) It's not an MA.
2) It is not a PID controller.
3) The system functions on any TF.
4) For efficient operation it is necessary to take into account the state of the older/current/smaller TF.