Discussion of article "Econometric Approach to Analysis of Charts" - page 9
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
Yes? And I think you asked faa1947 such questions that I think you are not aware of the issues.
For example, statistical distribution is a variation characteristic. Stationarity is temporal...
that's your pearl:
N.S. Kremer, Probability Theory and Mathematical Statistics, page 286. Verbatim.
The entire population of objects (observations) to be studied is called the general population.
Further:
The concept of a general population is in a certain sense analogous to the concept of a random variable (probability distribution law, probability space), since it is completely conditioned by a certain set of conditions.
Same thing in other words. I admit that I may understand something differently than you do. I don't know what a variation characteristic is. Now it's your turn.
I think we can. They should not affect the statistical parameters of the sample in any way (in particular, the distribution parameters). That's why they are outliers.
And I've seen data that the disadvantage of non-linear models, is the need for significant sampling..... about 1000 pieces.
Numerous examples in Matlab and other packages contain within a hundred. I don't understand if it's just an example or if there's something behind it. I will not expand this topic. Still, it is necessary to be consistent.
Clarify the meaning of the term "rack" please.
There was a text in English. - my translation. One rack includes all CBs falling into one interval. I give an example of 3600 candlesticks with different breakdown. In STATISTICS is the concept of width. By x is the value of quotes. From 1.2-1.3 occurred more than 700 times out of 3600 candlesticks.
The higher the number of racks, the worse with normality. Retrieved from STATISTICS
There are no universal methods for removing outliers...
That's why the sample size must be large.
No opinion on sample size. Take M1 for a year and H1 for the same year. Different number of candles. Which is better? Trends on M1 are different from treads on H1, but we are going to detrend..... It's not clear at all.
About trends. I haven't researched it. I'll keep it in mind.
It seems to me that in many ways the dog is buried in trends. The presence of trends distorts statistics. If they were poorly detrended, the distortion remains. What is a trend? Could it be a regression? If so, it is possible to get high quality detrending through non-linear regressions. But on what number of candles? Just unanswered questions.
I'm going to take some time out.... and I'll try to give you my thoughts later... but in general, I agree with the proposed procedural list....
To this list I would like to add one more wish: to use packages like Matlab or, in extreme case, STATISTICS for calculations. I attach a decisive importance to this, because (1) we will exclude different interpretations of terms, (2) we will limit the range of problems, (3) we will get results that can be compared without getting into the subtleties of calculations.
Bravo again! You think before you use it, and you raise the right questions. About the number of "racks" you can look at this thread of mine. There is something there. There is a link to a good book (there is something about emissions too).
http://www.nsu.ru/phpBB/viewtopic.php?t=22051
Kendall and Stewart's "Theory of Distributions" covers some of the issues in more detail.
N.S. Kremer, Probability Theory and Mathematical Statistics, p. 286. Verbatim.
Next:
Same thing in other words. I admit that I may understand something differently than you do. I don't know what a variation characteristic is. Now it's your turn.
On what basis is the above pearl said? (They are not deleted, by the way, because of this, although they affect the parameters).The problem of terminology is extremely unpleasant. I would like to remind you of teaching at university, when the teacher says "only according to my lectures" or "according to the textbook of such and such author" and gives an exact bibliographic reference. And this is not because Kremer has it wrong. It's just that he has "ammunition for a different system." If you look at and select decent MQL4 topics, all of them were eventually bogged down by attempts to harmonise terminology, and attempts to give calculations always failed. The point of the topic disappeared. So once again my suggestion is one package that has a section called Econometrics. The best candidate is Matlab, although there are specialised packages such as Eviews.
The problem of terminology is extremely unpleasant. I would like to remind you of university teaching, when the teacher says "only according to my lectures" or "according to the textbook of such-and-such author" and gives the exact bibliographic reference. And this is not because Kremer has it wrong. It's just that he has "ammunition for a different system." If you look at and select decent MQL4 topics, all of them were eventually bogged down by attempts to harmonise terminology, and attempts to give calculations always failed. The point of the topic disappeared. So once again my suggestion is one package that has a section called Econometrics. The best candidate is Matlab, although there are specialised packages such as Eviews.
http://www.nsu.ru/phpBB/viewtopic.php?t=22051
Kendal and Stewart's "Theory of Distributions"covers some of the points in more detail.
Thanks for the links. Suddenly there is clarity. It is worth remembering why we are making a garden? We need: market reversal, market continuation, and, ideally, to distinguish reversal from correction (flat). In this case, we need to start from the number of candlesticks, which will not be more than 100 (to the question of 1000 in denkir). Candlesticks in a quote are dependent and, as it seems to me, the sample size should be taken according to the ACF - where it has faded, this is the sample size.
... because you have politely made it clear that you wish to move forward within the framework that you think is right at the moment.
faa1947:
There was an English text. - my translation. One rack includes all CBs falling into the same interval. I give an example on 3600 candlesticks with different spacing. In STATISTICS is the concept of width. By x is the value of quotes. From 1.2-1.3 occurred more than 700 times out of 3600 candlesticks.
I get it, I use the concept of "class" or "interval" for this.
faa1947, in your figure I see that the distribution is not unimodal. This is another problem.
Then the number of classes (racks) is also calculated by some formula, rules... the most famous ones are:
Sturges' formula, Freedman-Diaconis rule, Scott's rule, Square-root choice, etc.