So I'd like some help w/ this question.

Given 3 assets with means, variances, and correlation:

enter image description here

Two portfolios are created (A and B), each with the three assets above with weights ($w_n$) as follows:

Portfolio A: $w_1=0.2$, $w_2=0$, $w_3=0.8$

Portfolio B: $w_1=0.4$, $w_2=0.1$, $w_3=0.5$.

The assets' correlation are


I would like to know the correlation of the two portfolios?

My attempt: So I've calculated the two portfolios' expected values and variances as follows: $E(A)=0.084, Var(A)=0.0024576$

$E(B)=0.092, Var(B)=0.006193$

And If I use $$\rho_{AB}=\frac{Cov(A,B)}{\sigma_A \sigma_B}=\frac{E(AB)-E(A)E(B)}{\sigma_A \sigma_B}$$ how would I calculate $E(AB)-E(A)E(B)$? Is $E(AB)$ the Expected Value of the assets' products, or is it the Expected Value of their weighted products? Thanks


3 Answers 3


You may be over-thinking it. It is a straightforward calculation using matrices, as easy as turning the crank of a sausage-making machine.

The standard deviation matrix is

               |0.16 0    0   |
S = Diag(s) =  |0    0.15 0   |
               |0    0    0.04|

The correlation matrix is

     |1.0  0.5  0.2| 
R =  |0.5  1.0  0.0|
     |0.2  0.0  1.0|

Therefore the covariance matrix is

                |0.02560  0.0120  0.00128| 
 C= S * R * S = |0.01200  0.0225  0      |
                |0.00128  0       0.00160|

The portfolio weights are

       |0.2|                   |0.4|
 wa =  |0  |     and     wb =  |0.1|
       |0.8|                   |0.5|

Therefore the covariance of portfolio A with portfolio B is

Cov(A,B) =wa^T * C * wb = 0.003466

and the covariance of A with itself, also known as the variance of A is

Var(A) = wa^T * C * wa = 0.002458

And similarly the variance of B is found as

Var(B) = wb^T * C * wb = 0.006193

Finally we can compute the correlation between A and B according to the definition

$\rho(A,B)=\frac{Cov(A,B)}{\sqrt{Var(A) Var(B)}}$ giving

rho(A,B) = 0.888326

You can solve for the covariance of the two portfolios and since you have E(A) and E(B) you can back into the E(AB)

  • $\begingroup$ What do you mean "back into E(AB)"? And how would you calculate the covariance? $\endgroup$
    – tsp216
    Commented Jun 9, 2019 at 16:33

Asset product = Weighted sum of the product of the three assets.

You can write the following Since:

$A = \omega^A_1.A1+\omega^A_2.A2 +\omega^A_3.A3 $


$B = \omega^B_1.A1+\omega^B_2.A2 +\omega^B_3.A3 $

You can develop $E(A.B) $ from there as a linear combination of the mean and variance couple for the three assets...

Therefore, the formula for $cov(A,B) $ is straight forward ...

  • $\begingroup$ How would you calculate E(AB) though? Do you multiply the portfolios corresponding assets, and then add these products up? $\endgroup$
    – tsp216
    Commented Jun 9, 2019 at 17:03
  • $\begingroup$ Yes and that's why you need the correlations between the different assets because in order to calculate $E(A.B)$ you'll need to calculate $E(A_i.A_j)$ , $i$ and $j$ rangig from 1 to 3. From the formula you've written in your question you can see there is a way to get $E(A_i.A_j)$ from $\rho_i$, $\sigma_i$ and $E(A_i), E(A_j)$... $\endgroup$
    – Xman
    Commented Jun 9, 2019 at 20:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.