Random Flow Theory and FOREX - page 78
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
Is it possible to program without knowing the subject area you are programming for?
Is it possible to automate a company you have never seen and know nothing about?
Listen to you - why all this education that drags on for years - come and work straight away.
Of course, there is a lot to learn before using regression analysis. There's no doubt about that. But there are a fair number of people on the forum who have that initial training. Prival is among those people. He can take the next step in the form of mastering a specialist package. There are 700 people who have downloaded the R wrapper, which just to try it out as an indicator will not do. So they will not download the wrapper just for the sake of curiosity.
PS/
Bundles by bundles - I did the bundling for singular decomposition.
While the package was, yes.
And we bought the same engine that was in the package.
And the numbers we ended up with matched 1:1.
But to talk about packages in the forex context is ridiculous.
Packages are written by nerds and geeks.
And to use/study they need something else than the ability to just run numbers.
That's why the breakdown as a practitioner is a couple of notches higher than using "packages".
For example, if we are talking about a manual study, a person has to mark out
the area on the graph directly in the MT - lines.
And for the system to calculate everything with one button.
Because in order to do a study, you need to go through the history of, for example, a year.
And if every time you make an export to Csv and then manually leave
Every time you do export in Csv and manually leave a window for necessary dates and cut dates from data - it's too inconvenient.
And for automation, a user should set the testing step and the learning area
(optimisation - forward).
For multicurrency we should prepare data.
Exclusively by means of software, and on your own hand.
And no package will be of much help to you there.
There are 700 people who have downloaded the R-wrapper, which just to try it out as an indicator will not do. So just for the sake of curiosity they will not download the wrapper.
My posts are addressed to these people.
I don't know why they downloaded it and developed it, though I'm not disputing that the product is cool, I'm just saying that I guess I'm right, because if 700 people downloaded it without knowledge, maybe someone will start a discussion with you, but no, why not know(?), Although I would be interested to understand it. But alas, we are not countess)).
For example, if you're talking about a manual study, a person should highlight
the area on the graph directly in the MT lines.
And for the system to calculate everything with one button.
Because for the study you need to go through the history for example a year.
And if every time you make an export to Csv and then manually leave
Every time you do export in Csv and manually leave a window for necessary dates and cut dates from data - it's too inconvenient.
And for automation, a user should set the testing step and the learning zone
(optimisation - forward).
For multicurrency we should prepare data.
Exclusively by means of software, and on your own hand.
And no package will be of much help to you there.
The problems you listed are not known to me. The capabilities of R (EViews) are far beyond me.
Comparing a trading terminal with a statistics package is meaningless. It is like comparing wheels with an engine.
Besides, I want to emphasize that it is incorrect to evaluate what you are not familiar with.
The problems you listed are not known to me. The capabilities of R (EViews) are far beyond me.
Comparing a trading terminal with a statistics package is meaningless. It is like comparing wheels with an engine.
Besides, I would like to point out that it is not correct to evaluate something with which you are not familiar.
And you are. Okay. (chuckles)
I want a system that searches in increments of a week for the best method of processing
and the best processing parameters.
Then it will take a few instances of the parameters found and apply them to the forward data.
The results of the forward trades will go into a table.
The system will take the multi-currency data beforehand,
And it will align them with the dates so that the matrix is... correct.
So, is there a package for this kind of work?
What I'm writing here is torture.
And you know each other. Okay.
I need a system that searches in weekly increments for the best method of processing
and the best processing parameters.
Then it will take a few instances of the parameters found and apply them to the forward data.
The results of the forward trades will go into a table.
The system will take the multi-currency data beforehand,
And it will align them with the dates so that the matrix is... correct.
So, is there a package for this kind of work?
What I'm writing here is torture.
Above I have posted the composition of R, which refers to the time series. Naturally, there are means to deal with dates in a wide variety of ways.
But the point is not whether there are tools available to solve the problems you see at the moment. The point is that there are a lot of tools that you do not know exist, but which are used by your counterparties in the market. For example, bootstrapping, or Monte Carlo or something else, if we're talking about testing.
Before you test, you need to know WHAT you are testing. My article here shows that one should first find out if it makes sense to test at all. If the error of estimation of model parameters is up to 5%, it is reasonable to test, but what if it is 100%? And the question of estimating errors of TC parameters is not addressed at all. We ran it in the tester and then we enjoy the sacred cow called "forward test", while the error of model parameters is prohibitive - we simply don't know about it and don't want to know.
Besides, there are many other problems in any TA-based TS about which the developer just does not know. And then he is surprised: "it passed the test, passed the forward test, brought money, but then it emptied the deposit". And if it is not so, it only means that the author has walked the tightrope over the abyss blindfolded. Just lucky. The lucky number is known - 5%. Casino.
But it is not a question of whether the means are available to solve the problems you see at the moment. The point is that there are a lot of tools that you don't know exist, but that are used by your counterparties in the market. For example, bootstrapping, or Monte Carlo or something else if we are talking about testing.
and using the criteria he trusts.
The market doesn't seem to care about the "parameter error" you're talking about.
and it can go down either way.
And if your models are so good, do some forward testing on them,
show a Monday spreadsheet of how much your system has earned in a year.
I think there is no other criterion for most of us here.
And won't you have a lot of followers if you put the question like that?
... that he trusts.
It is advisable to use those criteria that have justification.
The market does not seem to care about the "parameter error" you speak of.and a leakage is possible in any case.
Not in any case, but for quotes that have kinks.
I am not agitating anyone. I am interested in like-minded people. Hangout. Like on MQL. Or on TA on this site. There are academic sites on econometrics, but they are not usually of practical orientation.
... that he trusts.
It is advisable to use those criteria that have justification.
The market does not seem to care about the "parameter error" you speak of.and a leakage is possible in any case.
Not in any case, but for quotes that have kinks.
I am not agitating anyone. I am interested in like-minded people. Hangout. Like on MQL. Or on TA on this site. There are academic sites on econometrics, but they are not usually of practical orientation.