Discussion of article "Practical application of correlations in trading" - page 2
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
ALEXANDER FEDOSOV: Суть любой торговли, так или иначе сводится к тому, что приходится прогнозировать дальнейшее развитие событий на рынке и потенциальная прибыль сильно зависит от успешности прогноза.
I stronglydisagree. There are trading systems that do not make any forecasts, but rather exploit the statistical properties of a particular financial series or a set of them.
Then where did this conclusion come from -"Take a trend section, mark it with numbers in order (monotonic function) and look for correlation between them"?
In Figure 1 we have only 1 row, then in the table of calculations, if you repeat them, you can see that we are counting the Pearson correlation between closing prices and the numerical row, which is just numbered candlesticks. This is something strange in general, and it does not look like autocorrelation and rank correlation as from China.
Further on 4 types of correlation are considered and it seems to me without understanding when and where to apply each of them.
Further on there is no such section as data preparation.
Further in the answers there is some sarcasm about gurus who can advise to calculate correlation from the first difference - it can all be done, the only question is when to take the first difference, when to take the second, when to take the initial data, and when to bring them to another dimension, etc. - this is all missing. As a result, the article is for beginners, and there is nothing for them.
I'm not attacking you personally, you seem to have a good head, but the article is unfortunate.
In Figure 1 we have only 1 row, then in the table of calculations, if you repeat them, you can see that we count the Pearson between the closing prices and the numerical series, which is just numbered candlesticks. This is something strange in general, and it does not look like autocorrelation and rank correlation as from china.
Then we consider as many as 4 types of correlation and it seems to me without understanding when and where to apply each of them.
About 10 years ago Jurick's indicators (if I remember the author's name correctly) were very popular. They cost money and were very secret. But then some clever people found out that one of them was based on Spearman's Correlation Coefficient. The others were based on digital filters. So a lot of complex things are based on "simple" maths applied even to bar numbers.
About 10 years ago, Jurick's indicators (if I remember the spelling of the author's name correctly) were very popular. They cost money and were very secret. But then some clever people found out that one of them was based on Spearman's Correlation Coefficient. The others were based on digital filters. So a lot of complex things are based on "simple" maths applied even to bar numbers.
I'm always in favour of simplicity too, but that doesn't mean you can cross cats and dogs.
..... someone will come running and say that such a hybrid is also sometimes found in nature :)
In Figure 1 we have only 1 row, then in the table of calculations, if you repeat them, you can see that we count the Pearson between the closing prices and the numerical series, which is just numbered candlesticks. This is something strange in general, and it does not look like autocorrelation and rank correlation as from China.
Further on 4 types of correlation are considered and it seems to me without understanding when and where to apply each of them.
Further on there is no such section as data preparation.
Further in the answers there is some sarcasm about gurus who can advise to calculate correlation from the first difference - it can all be done, the only question is when to take the first difference, when to take the second, when to take the initial data, and when to bring them to another dimension, etc. - this is all missing. As a result, the article is for beginners, and there is nothing for them.
I am not attacking you personally, you seem to have a good head, but the article is a failure.
Correlation between the price and the sloping line. It's a standard way of applying correlation, known for centuries. That's what correlation is for - to compare something with something. If the price goes up - positive correlation, if it goes down - negative correlation. As a result, it turns out to be a kind of oscillator.
The main thing is that there are functions for calculating several correlation methods, and those who need them will refine them to what they need. Theorising makes no sense.
Correlation between price and sloping line.....
Aha, it is the same as to consider correlation between left and right eyes....and sometimes it is broken.
And if to take a sloping line, then in the same coordinate system where the digital series. In general, as you wrote yourself, there is no point in theorising.
Aha, it is the same as to consider correlation between the left and right eyes....and sometimes it is broken.
And if to take a sloping line, then in the same coordinate system where the digital series. In general, as you wrote yourself, there is no sense in theorising.
Correlation for calculation does not require normalisation of data.
The correlation calculation does not require normalisation of the data.
It requires the data to be normally distributed.
it requires the data to be normally distributed.
It certainly doesn't require that.