How Portfolio Constraints Change the Mean-Variance Efficient Frontier

Summary

The document asks whether limiting portfolio weights to long-only bounds or allowing bounded short positions can cause a sudden change in the slope of an efficient frontier as risk aversion varies. It frames the effect as a consequence of restricting the portfolios available to the optimizer: as the preferred allocation changes, one or more weight limits may become binding, changing the shape of the optimal risk-return trade-off.

The post gives an intuitive hypothesis but no derivation, cited source, numerical example, or empirical evidence confirming it. It therefore serves as a focused question about constrained mean-variance optimization rather than a complete explanation. The precise frontier shape depends on the optimization setup, including the covariance and expected-return estimates, other portfolio constraints, and how risk aversion is varied; the document does not specify these details or establish when a visible gradient change must occur.

Key ideas

• Weight bounds restrict the feasible set of portfolios in mean-variance optimization.
• A change in frontier slope may occur when portfolio weights reach or leave their bounds.
• The post presents this as an intuitive question and does not provide a proof or example.
• The observed frontier depends on the model inputs and the full set of constraints.

Tags

# Portfolio constraint effects on efficient frontier


# Portfolio constraint effects on efficient frontier












I'm trying to just replicate the mean-variance optimization problem but with variable risk-aversions.

I added constraints that limit the constituent weights to [0,1] and [-1,1] for long-only and shorting allowed situations, respectively.

The result was an efficient frontier with an abrupt change in gradient at some risk-aversion level. So, intuitively this makes sense. We're restricting the available portfolios to us. But, I'm having trouble confirming it from any credible source. The best I found was a slide deck from a UW professor, but the notes are vague and hard to read.

Could someone confirm if this effect is expected in the presence of constraints?

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.