How to minimise index correlation - page 4
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
Maybe I just don't understand something. but I am surprised that you (and not only Sabluk also wrote some time ago) are focusing on the decomposition method, which is applicable to different timeframes and different instruments. and which allows you to get the anticipation regardless of multi-analysis of currencies (even on the same instrument).
But if the anticipation takes place, then it will probably show up in the indices of currencies that form it (or rather in change rates of these indices, anticipation by rates will show anticipation by the movement of the instrument itself).
As the result, if you do not know the method, then you can go the other way, to build indexes so that there is an instrument prevalence (composite synthetic runs earlier), but it cannot be calculated in a couple of formulas. it is necessary to build one database, transform it into another, pull out the third, etc. because these are not linear regularities and cannot be described by regression
In fact, how can the question be posed about "reducing the correlation of indices"?
The indices are what they are. And their correlation is as it will be.
We should probably ask about the basis (orthogonal functions).
And that would seem to be PCA with "factors" and "loads"- or SSA.
Except that in PCA the loads behave like rabid beasts and jump unpredictably.
The correct currency index has already been calculated and even detailed in the thread you cited. The topic can be closed. But if you don't do the obvious, no one on the forum will help you.
The issue is that the index calculation is not about you, you don't have to close the thread.
Actually, how can one put the question about "reduce the correlation of indices" ?
The indices are as they are. And their correlation is what it turns out to be.
Probably, one should ask about the basis (orthogonal functions).
And it seems to turn out to be PCA with "factors" and "loads"- or SSA.
Except that in PCA the loads behave like rabid beasts and jump unpredictably.
Yes, your clarification is probably closest to what I wanted to say.
By the way, all (mostly) topics about index correlation or something like that are quickly silenced, not exceeding even the number of fingers on the hands.
By the way, this should also show up in the correlation analysis (anticipation) only speed and acceleration will be replaced here by correlations and correlation between correlations...
And why does the greater number of pairs in the calculation according to that formula indicate the independence of the indices? you also once set the orthogonality condition. so i think that orthogonality is not obtained by simply adding the number of instruments to the calculation according to that standard formula.
A large number of currencies is an orthogonality condition, I could not think of another one.
Suppose the euro has grown by 10%, the other currencies have not changed. It means that the increments of EURUSD,EURGBP,EURJPY,EURCHF, etc. are equal to 0.1, the increments of all others are equal to 0.
Let's calculate the indexes:
for 2 currencies(for EURUSD) EUR=MathPow(1.1,2.0)=1.049 USD=MathPow(1.0/1.1,2.0)=0.953
by 3 currencies(EURUSD,GBPUSD,EURGBP) EUR=MathPow(1.1*1.1,3.0)=1.067 USD=MathPow(1.0/(1.1*1.0),3.0)=0.969
Then count for yourself, ideally it should be EUR=1.1 USD=1.
It seems obvious that orthogonality increases as the number of currencies increases.
A large number of currencies is an orthogonality requirement, I could not think of another one.
Suppose the euro has risen by 10%, the other currencies have not changed. So the increments of EURUSD,EURGBP,EURJPY,EURCHF, etc. are 0.1, the increments of all others are 0.
Let's calculate the indexes:
for 2 currencies(for EURUSD) EUR=MathPow(1.1,2.0)=1.049 USD=MathPow(1.0/1.1,2.0)=0.953
by 3 currencies(EURUSD,GBPUSD,EURGBP) EUR=MathPow(1.1*1.1,3.0)=1.067 USD=MathPow(1.0/(1.1*1.0),3.0)=0.969
Then count for yourself, ideally it should be EUR=1.1 USD=1.
It seems obvious that orthogonality increases as the number of currencies increases.
and what is the benefit of such orthogonality (exactly in this form). and in fact I want to move away from the word orthogonality, now they will give an anathema ato.
If you do not understand what you mean, you start to display orthogonality and calculate indexes according to the generally accepted formula, then you write that you have your own method of index calculation, which is, of course, a secret.
There's nothing more to explain...
At the point where the vertical line is, all currencies are stupidly equated to one USD=EUR=GBP=JPY=CHF=CAD=AUD=NZD=1.0 (i.e. 0.0% change). On each bar we calculate the percentage change of each currency and draw a point. I cannot show you how it is calculated. We calculate it basing on the orthogonality requirement for the currencies.
I personally see anticipation in phase changes.
You have not started the topic, you do not have to close it. If you have made a mistake here, you can express yourself in other brilliant topics, such as locks, but do not read this one.
What you wrote at the beginning of the thread is simply unnecessary if the indices are calculated correctly. And you certainly shouldn't use ma. Because you will get a delay. A currency index can be used to find overbought oversold indices.
Yes, or maybe the opposite, currency indexes calculated using your, as you put it, long described and correct formula, lose their meaning if you do not produce in the future what you wrote in the beginning.
What is your understanding of what is right and what is wrong . in relation to what is right and in relation to what is wrong.