Covariance on historical data

@ Dreary
st.dev X = 1.224744871
st.dev Y = 2.121320344
r = 0.866025404
thus cov x,y = 0.8866025404 * 1.224744871 * 2.121320344 = 2.25
this is if you use sample st.dev.
cheers
 
but it’s wrong…
you divided the squared deviations by n instead of n-1.
 
FisherSU Wrote:
——————————————————-
> @ Dreary
>
> st.dev X = 1.224744871
> st.dev Y = 2.121320344
> r = 0.866025404
>
> thus cov x,y = 0.8866025404 * 1.224744871 *
> 2.121320344 = 2.25
>
> this is if you use sample st.dev.
>
> cheers
—assuming you’re using the TI BA2 calc, I think you’re looking at the wrong numbers,
it gives you sigma (which is population), and S subscript X/Y, which is the sample std dev…
sigmax = 1.224
Sx = 1.414
sigmay = 2.121
Sy = 2.449
so if you use the sample stats, you have
cov x,y = 0.8866 * 1.414*2.449 = 3….
 
That sample variance where you take sum (x - x-bar)^2/n isn’t a population anything cause it’s got an x-bar in it. About the only thing going for it is it’s maximum likelihood if the x’s are normal.
 
Dreary u are right,
i was so sure sigma is sample variance, even now i could bet that i have used sigma for sample variance notation somewhere.
thanks.
 
You must log in or register to reply here.
Back
Top