correlation for the uninitiated

A

aPismoClam

let's say the correlation of B to A is 0.4, according to Bloomberg.

is either following statement a correct interpretation of this infomation:

"B will move in the same direction of A 4 times out of ten;"

"B will move in the same direction of A with a magnitude of n*A 4 times out of ten;"

or is there a better way to describe it?

Thanks.
 
Here is an example that has exactly ten data points. The correlation coefficient is 0.408. In the first seven rows, A and B are equal. In the final three rows, A and B are not equal.

These might represent the daily price changes of Asset A and Asset B, or perhaps the trade profits of System A and System B, or possibly the daily weight gains and weight losses of Patient A and Patient B.

corrhk5.png
 
first you correlate based on returns not levels. you can then say that 40% of a move is reflected in the other instrument
 
Quote from aPismoClam:

let's say the correlation of B to A is 0.4, according to Bloomberg.

is either following statement a correct interpretation of this infomation:

"B will move in the same direction of A 4 times out of ten;"

"B will move in the same direction of A with a magnitude of n*A 4 times out of ten;"

or is there a better way to describe it?

Thanks.
More...

A correlation has an R-squared, which is the strength of the correlation (the closer to one the better), and then the dependent variables have a coefficient which measures the average change in the independent variable based upon a change in the dependent variable.

The statement that you made sounds to me like the r-squared or adjusted r-squared are what are being described, and so that would be interpreted as B and A moving together 40% of the time. Your second statement would be correct if you meant n* to be the coefficient, or as you stated, the magnitude of the change.
 
Quote from ChaosNSX:

lol I do not even know where to move this one.
More...
lol wtf is that supposed to mean?

I agree, there must be other threads that are worthy of your moderation.

cya.
 
Quote from Deebee230:

A correlation has an R-squared, which is the strength of the correlation (the closer to one the better), and then the dependent variables have a coefficient which measures the average change in the independent variable based upon a change in the dependent variable.

The statement that you made sounds to me like the r-squared or adjusted r-squared are what are being described, and so that would be interpreted as B and A moving together 40% of the time. Your second statement would be correct if you meant n* to be the coefficient, or as you stated, the magnitude of the change.
More...

bang -- a spot on answer. thank you.
 
You must log in or register to reply here.
Back
Top