Zero sample correlation does not necessarily mean there is no linear relationship - page 6
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
You are being told about the proper preparation of price BPs for correlation assessments. And it doesn't matter which market the financial instrument belongs to. It is, indeed, fundamental.
Have you decided to lecture us?
Based on your knowledge of the specifics of the subject? Or by reading a book?
There is a built-in function in the matrix package. I don't care what it is, you can logarithm it or not. The output is ACF. I'll do what I promised, I'll give you all three methods of calculation. {...}
What's the point? Tell me the development methodology and I'll tell you what you'll get.
If the Mq4 indicator matched Mathcad, then what could be the point of argument?
The fact that the indicator showed the same thing is a clear diagnosis. "Health".
.
If you can, please write what you think about the calculation that hrenfx is talking about.
When two offset windows are taken and the line reg. and RMS are counted separately in them and the corr. on them.
The method may be naive, but for some reason I like it :-).
....
You have to understand that it is a global mistake to count the correlation for EURUSD and USDJPY without logarithm.
and may I ask you to calculate correlation for the same pairs "your way", but only if USDEUR & JPYUSD are used instead...
Please present the results to the public...
;)
-----------
for correlation there should be no difference between forward and reverse pairs.
That's how I test any TA attachments.
I wish you to do the same.
Based on knowledge of the specifics of the subject? Or from books?
It makes no difference where the knowledge comes from. What matters is their correctness (logic).
Here is an example of the realization of correlation, where Reshetov raises the logical question of "buckets and kilograms". And you just have to logarithm it.
You just have to logarithm it.
Who prevents you from simply...
F' = k1 + (F - k2) * k3 ?
So that G and F' end up being from [y1; y2] ?
.
Or just.
F' = (F - LinearRegression) / RMS
Who cares where the knowledge comes from. What matters is their correctness (logic).
Here is an example of a correlation implementation, where Reshetov raises the logical question of "buckets and kilograms". And you just have to logarithm it.
Well if you "restless" Reshetov decided to educate?
God help you...
;)
and may I ask you to calculate the correlation for the same pairs "your way", but only if USDEUR & JPYUSD are used instead...
No problem. The correlation between EURUSD and USDJPY equals the correlation between USDEUR and JPYUSD. Because, log(EURUSD) = -log(USDEUR), log(USDJPY) = -log(JPYUSD).
Do you really think that flipping a quote has any effect on the degree of correlation?
P.S. Look at Recycle, it does not use correlation to determine the relationship at all. Because the correlation is not between two BPs but any number. Especially telling is the example given there when a cross is added to the majors.
No problem. The correlation between EURUSD and USDJPY equals the correlation between USDEUR and JPYUSD. Because, log(EURUSD) = -log(USDEUR), log(USDJPY) = -log(JPYUSD).
Do you really think that flipping a quote has any effect on the degree of correlation?
You do the math first...
;)
I told you -
There should be no difference between forward and reverse pairs for the correlation relation.
That's how I check any TA tricks.
I wish you the same.You do the math first...
Who prevents you from simply...
F' = k1 + (F - k2) * k3 ?
So that G and F' end up being from [y1; y2] ?
.
Or just.
F' = (F - LinearRegression) / RMS
I don't get it.