Skip to content
All library documents

Finding a Fully Invested Portfolio Uncorrelated with an Efficient Portfolio

Article Quant Q&A · Author: Wolfy

Summary

The document presents a portfolio geometry question: given a fully invested portfolio on the efficient frontier other than the global minimum variance portfolio (GMVP), can another fully invested frontier portfolio have zero covariance with it? The accepted answer gives a geometric construction. Draw a line from the given portfolio through the GMVP and extend it to the vertical axis; then draw a horizontal line from that intercept to the efficient frontier. The intersection identifies the proposed zero-covariance portfolio.

The explanation relies on the efficient frontier’s geometry and the stated property that the GMVP has the same nonzero covariance with every portfolio on the frontier. It says the construction fails when the given portfolio is the GMVP, because a line through those two coincident points cannot be drawn and the desired case is not solvable. The document points to a separate proof but does not reproduce it, and the questioner’s algebra is not validated by the accepted answer. Readers should consult the cited derivation for the assumptions and proof details.

Key ideas

  • A fully invested efficient portfolio other than the GMVP can be paired with a frontier portfolio constructed to have zero covariance with it.
  • The construction extends a line through the given portfolio and GMVP to the vertical axis, then projects horizontally to the efficient frontier.
  • The procedure does not apply when the given portfolio is the GMVP.
  • The answer cites a separate derivation but does not provide its proof.

Tags

Full text
# Show that there exists a fully invested portfolio such that the covariance between their returns is zero


# Show that there exists a fully invested portfolio such that the covariance between their returns is zero












Background Information:

I came across this question in chapter 2 of Active portfolio Management by Grinold and Kahn. It pertains to the efficient frontier which is displayed below:

Question:

> If $T$ is fully invested and efficient and $T\neq C$, prove there exists a fully invested portfoli $T^{*}$ such that $Cov(r_T, r_{T^{*}}) = 0$

We have that $T$ is fully invested on the efficient frontier so $h_{T}^{T}e = 1$ since we want to find a $T^{*}$ that is also fully invested and efficient but $T^{*}\neq C$ it seems to me by the picture above that the only way we can have $Cov(r_T, r_{T^{*}}) = 0$ then $T^{*} = Q$?

Additional thoughts:

Since $T$ is efficient and fully invested it must be a linear combination of $h_C$ and $h_Q$ thus $$h_T = c_1 h_C + c_2 h_Q$$

Then since $h_T^{T}e = 1$,

$$\left(c_1h_C + c_2 h_Q\right)^{T}e = 1 \ \Rightarrow \ c_1 + c_2 = 1$$

Like wise since $T^{*}$ is fully invested so $h_{T^*}^{T}e = 1$. Now, $r_T = h_{T}^{T} r = \left(c_1h_C + c_2 h_Q\right)^{T}r$ and $r_{T^{*}} = h_{T^{*}}^{T}r$ So, \begin{align*} Cov(r_T,r_{T^{*}}) &= Cov((c_1h_{C}^{T} + c_2 h_{Q}^{T})r,h_{T^{*}}^{T}r)\\ &= c_1 c_2 Cov(h_C^{T}r + h_Q^{T}r, h_{T^{*}}^{T}r)\\ &= c_1 c_2\left(E[(h_C^T r + h_Q^{T})h_{T^{*}}^{T}r] - E[h_C^{T}r + h_{Q}^{T}r]E[h_{T^{*}}^{T}r]\right)\\ &= c_1 c_2 \left(E[h_{C}^{T}r h_{T^{*}}r] + E[h_{Q}^{T}r h_{T^{*}}^T r] - E[h_C^{T}r]E[h_{T^{*}}^T r] - E[h_{Q}^{T}r] E[h_{T^{*}}^T r]\right)\\ &= 0 \end{align*}

I am not sure if this is correct although. I don't really know exactly how to approach this problem, any suggestions are greatly appreciated.

## Answer by Alex C (score 2, accepted)

https://quant.stackexchange.com/a/31433

There is a procedure for finding $T^*$ starting from the portfolio $T$ on the efficient frontier, such that $cov(T^*,T)=0$:

From the point $T$ draw a line thorough the point $C$ (which represents the global minimum variance portfolio or GMVP) until it intersects the Y axis at a point $R_z$. Draw a horizontal line from this point until it intersect the parabola at a point we will call $T^*$. This is the desired portfolio having zero covariance with $T$.

Clearly if T and C coincides the procedure fails: the line TC is not defined. We know that the GMVP has the same non-zero covariance with every portfolio on the parabola, so this case is not solvable.

The detailed proof can be found for example on page 6,7,8 of this paper

https://eclass.aueb.gr/modules/document/file.php/INF258/ConstMalliarisPortfolioChapter.pdf

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.