Zero sample correlation does not necessarily mean there is no linear relationship - page 13

 
faa1947:

Wonderful formulas, and most importantly, correct. But where is the proof that they can be applied to forex BPs?

They can be applied to any price BP. See the Recycle description for the results of finding linear relationships.
 
hrenfx:

They can be applied to any price BPs. See the Recycle description for the results of finding linear relationships.


Nonsense. By the way, logarithmically.

 
hrenfx:

They can be applied to any price BPs. See the Recycle description for the results of finding linear relationships.

Generally references are to generally recognised textbooks, and you are referring to some person's private opinion. Recycle may be an interesting and useful idea, but the whole thread raises the question of the applicability of correlation, especially Pearson in Forex. The question of applicability is a basic question, don't you see that everything else is secondary.
 
faa1947:


Nonsense. By the way, logarithmic.

Did you really want to see a nice distribution after logarithms!

You are too lazy to delve into Recycle, which, among other things, solves the problem of cointegration.

 
faa1947:


Nonsense. By the way, logarithmic.

The passion for bags is commendable...

But what's the point of this muzzleloader?

Many of the issues raised in the thread are relevant.

And logarithm solves, as it were, some of the problems...

But Prival rightly notes the problems with computational accuracy at all stages of the calculations.

I urge to consider any currency pairs in the area of definition, which is usually different from the usual - from 0 to infinity...

There is no such thing on forex. 10% of the current price value is usually already infinity in one direction, and 0 in the other.

;)

 
faa1947:


A general question: can we look for a correlation between two BPs that are trending? In my opinion, you can't, you have to subtract the regular component which is the trend. Below is a histogram with the trend subtracted from the log

And personally for the mathematician. Histogram of the first difference. Almost normal. But what does the correlation between two differences of different BPs tell us?

 
faa1947:


A general question: can you look for a correlation between two BPs which are trending? In my opinion, you can not, you should subtract the regular component, which is a trend. Below is a histogram with the trend subtracted from the log

And personally for the mathematician. Histogram of the first difference. Almost normal. But what does the correlation between two differences of different BPs give us?

Is this a trend you have detected since the historical start?

The hryvnia or Nigerian currency is steadily depreciating... ;)

Or have you detected a trend in the Euro-dollar?

;)

 
faa1947:


General question: can we look for a correlation between two BPs which have a trend? In my opinion it is not, you should subtract the regular component, which is the trend. Below is a histogram with the trend subtracted from the log

What is a trend? QC is used for any BPs in order to find linear relationships.
 
hrenfx:

............

If the sample seems small, let's take something bigger from the correlation table:

Corr = 0.0000, #NGX0 - EURGBP, bars = 24943 (2010.05.28 21:25 - 2010.09.28 18:40), November 2010 Natural Gas Future - Euro vs British Pound

Corr = -0.0015, USDNOK - USDSGD, bars = 54961 (2010.01.01.01 00:00 - 2010.09.28 17:20), US Dollar vs Norwegian Krone - US Dollar vs Singapore Dollar

Wow, there is almost no linear relationship between the Norwegian krone and the Signpura dollar - nonsense!
Corr = -0.0008, GOLD - USDCAD, bars = 54898 (2010.01.01 00:00 - 2010.09.28 16:45), SPOT Gold Once vs US Dollar - US Dollar vs Canadian
Even funnier, there is almost no linear correlation between gold and the Canadian dollar - dick!

In fact, there is always a linear relationship between any two random variables on a finite sample.

.............

I will repeat my question to the topicstarter, which unfortunately went unanswered......

So, your main thesis is the necessary/obligatory presence of a pronounced relationship (i.e. correlation coefficient close to +1/-1) between any traded assets; not finding such a relationship has disappointed you in traditionally used statistical/mathematical methods and made you look for new approaches.....

In this thread, it has already been repeatedly pointed out by many forum members to the need to first have a logical relationship between the two series, and only after that, try to apply mathematical methods of analysis, but not vice versa!

So, I ask for an explanation of how, on a fundamental or technical level, there should be or should be justified the existence of a mandatory pronounced linear interdependence between any assets (you could use the example of the assets cited above from your post...).

If we come to the understanding that your original thesis, in fact, exists only to find an excuse to discuss your mathematical innovation - then I will consider the issue closed for me (except for a sporting interest to see who will get bored faster: the majority, changing your mind, or you, proving your theory..., and most importantly: how many pages it will take).

If you are able to justify your thesis, it will be very interesting.

 
Azerus:

I will repeat my question to the topic-starter, which, unfortunately, remained unanswered......

So, your main thesis is the necessary/obligatory presence of pronounced relationships (i.e. correlation coefficient close to +1/-1) between any traded assets; failure to find such a relationship has disappointed you in traditionally used statistical/mathematical methods and forced you to look for new approaches.....

In this thread, it has already been repeatedly pointed out by many forum participants to the need to first have a logical relationship between the two series, and only after that, try to apply mathematical methods of analysis, but not vice versa!

So, I ask you to explain how, on a fundamental or technical level, there should be or should be justified the existence of a mandatory pronounced linear interdependence between any assets (you can use the example of the assets given above from your post...).

If we come to the understanding that your original thesis, in fact, exists only to find an excuse to discuss your mathematical innovation - then I will consider the issue closed for me (except for a sporting interest to see who will get bored faster: the majority, changing your mind, or you, proving your theory..., and most importantly: how many pages it will take).

If you are able to justify your thesis, that would be very interesting.

But I must add something. There is a known relationship between the yen and the oil price. Oil up - yen down. Yen and dollar are in the same boat... :)

I.e. FA-level connections are easily justified.

;)

----

I don't expect linearity, even after logarithms and other dissections...

;)