[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Correlation Math Help



PureBytes Links

Trading Reference Links

Hi all

If one believes that the dependence structure of 2 variables is linear then
the Pearson coefficient is fine BUT....

A more general measure of concordance such as Kendall's Tau or Spearman's
Rho provides a more rigorous measure of dependence, especially when markets
are behaving wildly or away from what is normally expected.

Cheers
Tim

-----Original Message-----
From: Jim Johnson [mailto:jejohn@xxxxxxxxxxx]
Sent: Friday, August 23, 2002 5:48 PM
To: omega-list@xxxxxxxxxx; Ross S Bond
Subject: Re: Correlation Math Help


Hello Ross,

Coefficient r is the Pearson Product Moment Correlation coefficient.
Linear regression is the equation for the line of best fit between the
two variables.  The way I think of it correlation coefficient
(squared actually) tells you the amount of variance predicted in variable a
by variable b.

If you know r you can establish the regression equation.  The larger
the absolute value of r, the less the variation around that line of
best fit.

That said, I have no idea how one establishes the correlation between
systems.



Best regards,
 Jim Johnson                           mailto:jejohn@xxxxxxxxxxx

-- 
Friday, August 23, 2002, 7:56:03 PM, you wrote:

RSB> Hello OmegaList,

RSB>   I am trying to determine correlation between market systems - am I
RSB>   correct to use the linear correlation coefficient r (from page 10 of
RSB>   Vince's Mathematics of Money Management), or should I be looking at
RSB>   something like linear regression?

RSB>   Thanks in advance :)!