A formal definition of a master-slave - is there one? - page 4

 
faa1947:

No correlation has a place in the series - it is a characteristic of a sample of two series.

Just as there is no specific place for your favourite cointegration and XP-filter - they are the same characteristics of samples as a whole.

The point of what is proposed is not about "place correlation" at all, but about estimating its change, and that is a value tied to that particular point in time.

 
There are combat algorithms that determine master-slave. These are based on Kalman filtering. These are subtle structures for analysing the reflected signal, but it is possible and it works. Because the 'nature' of master and slave movement is different (slightly different) slightly different. This difference is incorporated as a behavioural model into the Kalman filter
 
alsu:

Just as there is no specific place for your favourite cointegration and XP-filter - these are the same characteristics of samples in general.

The point suggested is not the "place of correlation" at all, but the estimation of its change, and this is already a value tied to that particular point in time.

It's not about location, it's about the inapplicability of correlation on the foreground. Let's not be petty.
 
Prival:
There are combative algorithms that determine master-slave. These are based on Kalman filtering. These are subtle structures for analysing the reflected signal, but they are possible and work. Because the 'nature' of master and slave movement is different (slightly different) slightly different. This difference is incorporated as a behavioural model into the Kalman filter

The Kalman filter is a separate and extremely interesting topic.

What is being discussed here is the Granger causality test

I think he got a Nobel for it. The test is very often included in statistical packages. Accessible, very easy to use.

 
In fact, the same thing is being discussed here.
 
None of the instruments in the financial markets will be correlated enough to trade on them as described here. But in order for there to be a pattern of different instruments to make money, they do not need to be correlated. Moreover, the correlation can be 0 at all, but the patterns will exist.
 
LeoV:
None of the instruments in the financial markets will be correlated enough to trade on them as described here. But in order for there to be a pattern of different instruments to make money, they do not need to be correlated. Moreover, the correlation can be 0 at all, but the patterns will be there.


golden words...

+100500

 
faa1947:
It's not about location, it's about the inapplicability of correlation on fore. Let's not be petty.

It sounds very artistic , but it makes no sense. For it is not true.

If it doesn't work for you, don't make a profound theory out of it.

You just have to be careful to separate: correlation apart / causation apart.

 

LeoV:

None of the instruments in the financial markets will be correlated sufficiently to trade on them as described here.

_________________________________________

There is correlation. But why are we getting into it.

Let's talk about group movement

 
MetaDriver:

It sounds very artistic , but it makes no sense. For it is not true.

If it doesn't work for you, don't make a profound theory out of it.

You just have to be careful to separate correlation separately / causation separately.

Preferably a proof and if possible not in-depth and with references.