I've found a few sources for instance this one, that makes the claim that over a portfolio, simple returns are additive. I can see why it doesn't make a difference if we use net or gross returns as long as the weights add to one, but I can't see why we should get this additive behaviour for a quantity that has a division in it. Perhaps I am using a bad definition of weight?

If someone could share a derivation for a portfolio of two assets that would be sufficient.


We have a portfolio of two assets, with prices $P_1(0), P_2(0)$ at $t=1$. In the next period, their prices are $P_1(1), P_2(1)$.

The portfolio market value at $t=0$ is $V(0) = W_1 P_1(0) + W_2 P_2(0)$.

The portfolio weights $\omega_i$ (for $i =1,2$) are: $$ \omega_i = \frac{W_i P_i(0)}{W_1 P_1(0) + W_2 P_2(0)}. $$

(We can easily see that $\omega_1 + \omega_2 = 1$.)

The return for an asset $R_i$ is given by: $$ R_i = \frac{P_i(1)}{P_i(0)} - 1. $$

The market value of the portfolio $V(1)$ is equal to: $$ V(1) = W_1 P_1(0)(1+R_1) + W_2 P_2(0) (1+R_2). $$ Then, $$ V(1) = (W_1 P_1(0) + W_2P_2(0))(1 + \frac{W_1P_1(0)R_1}{W_1 P_1(0) + W_2P_2(0)} + \frac{W_2P_2(0)R_2}{W_1 P_1(0) + W_2P_2(0)}), $$ $$ V(1) = V(0) (1 + \omega_1 R_1 + \omega_2 R_2). $$ If we define the portfolio return $R$ as $\omega_1 R_1 + \omega_2 R_2$, then $$ V(1) = V(0) (1 + R). $$

This is easier to see if we normalise prices.


It is not true.

$$w_t=p_{1,t}q_{1,t}+p_{2,t}q_{2,t}$$ Without loss of generality, to ignore splits, $q_{x,t}\equiv{q_{x,t+\Delta{t}}}.$






Except for a couple of special cases, such as bankruptcy where the assets go to zero, this is not generally true. For example, if you made a 1000% return on a \$1 investment and a 1% return on a \$100,000,000 then clearly the returns are not additive.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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