Suppose there are only two risky assets and we want to optimize our portfolio. Constraints are that we have a minimum return $\overline{r}$ and we can only invest $w_1 + w_2 = 1$.

Is it possible that in this setting the constraint $w_1 \times r_1 + (1-w_1) \times r_2 = \overline{r}$ always solves the problem or am I doing something wrong here?

I tried to set it up with Lagrangian: The constraint with $\lambda$ always provides me directly with the solution.

But how is that? I mean it seems strange that the solution is completely independent of the variance and covariance.

  • $\begingroup$ No. It is not possible, unless r1 and r2 are perfectly correlated. You have to show the minimization problem that you want to solve and the Lagrangian. $\endgroup$
    – Andre
    Apr 26 at 4:14

1 Answer 1


With $r_1$ and $r_2$ constants and your constraint the return equation reduces to an affine function (a line in the plane) which indeed has only one solution for a given level.

  • 1
    $\begingroup$ I disagree. I think his problem is he is going from a statement about a minimum return, which I would represent using an inequality constraint, to something with an equality constraint. The equality constraint is not the same as a minimum return for the portfolio, it is that the portfolio has that specific return. $\endgroup$
    – John
    Apr 19 at 17:17
  • $\begingroup$ That could very well be. Maybe something got lost in the translation to math. $\endgroup$
    – Bob Jansen
    Apr 19 at 20:00
  • $\begingroup$ "Suppose there are only two risky assets and we want to optimize our portfolio", what is the criterion that is being optimized. With only the constraint it is not clear... $\endgroup$
    – nbbo2
    Apr 20 at 16:21
  • $\begingroup$ r1 and r2 are not constant. $\endgroup$
    – Andre
    Apr 26 at 4:12

Your Answer

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

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