Page 1 of 1
Any idea about how to find the formula for 3 items?
Posted: Fri Aug 19, 2011 11:51 am
by oem7110
Based on the formula on image, I would like to know the formula to calculate 3 items.
Does anyone have any suggestions?
Thanks in advance for any suggestions
Please see following image on formula for calculating 2 items.
http://i1093.photobucket.com/albums/i43 ... lation.jpg
w = ratio
p = probability
q = correlation between 2 items
Posted: Fri Aug 19, 2011 12:32 pm
by sluggo
Treat it is a related problem:
Find the variance of a weighted sum of 3 random variables. Assume you know the variances of each random variable, the covariances of each pair, and the values of the weights. Then when you have the variance, you can calculate the standard deviation fairly straightforwardly.
As I recall, Papoulis presents it, including the more general N-dimensional case, step by step.
(ref)
Posted: Fri Aug 19, 2011 6:55 pm
by Eventhorizon
I am feeling mean ...
Let me Google that for you
Sorry!
Posted: Wed Aug 24, 2011 12:33 pm
by mojojojo
ha ha First time that I saw that.
I don't see q in that formula and p is correlation, not probability.
Posted: Wed Aug 24, 2011 1:17 pm
by sluggo
mojojojo wrote:I don't see q in that formula and p is correlation, not probability.
Code: Select all
Symbol What the OP called it What it stands for
========================================================================
Roman letter "w" w weight
Greek letter "rho" p correlation
Greek letter "sigma" q sqrt of variance
Posted: Thu Aug 25, 2011 2:37 pm
by mojojojo
sluggo wrote:mojojojo wrote:I don't see q in that formula and p is correlation, not probability.
Code: Select all
Symbol What the OP called it What it stands for
========================================================================
Roman letter "w" w weight
Greek letter "rho" p correlation
Greek letter "sigma" q sqrt of variance
I was forced to memorize and use this formula for a while now.
I think you might have your "What the OP called it" column mixed up. Didn't he/she call "rho" probability and "sigma" correlation?
To the OP, think of the forumal as two parts. The first part relates to individual securities and the second part relating to pairs.
First part = the weight times variance
second part = 2 times weight times weight etc
Basically ou need the first part for each security and the second part for each pair ... but as pointed out this is easily explained by a quick google search.