MVO constraints

Can someone summarize the MVO constraints and how each one is mitigated?

-Not got for Multi period setting

-use for AO only

-Very sensitive to change in input-particularly-return expectation

-cannot for changes in some other variables…

Solution: Use MCS, Resampling, Black- Liteman ( should solve most of the problem).

Results in concentrated portfolios with allocations in a few asset classes and not in other - i.e. doesn’t result in logical diversification. Can use MCS to get averages around the inputs and can add constraints. Adding constraints (liquid positions, relative positions, position ranges, etc) is useful but you don’t want to use so many constraints that you end up just creating the portfolio that you want. But can add some in addition to budget and non-negativity constraint.

  • Ignores liabilities, skewness and kurtosis

  • very sensitive to model inputs (expected return, standard deviation, covariance etc.)

  • static, one-period model

  • leads to concentrated portfolios that require frequent rebalancing

  • tells us whether an asset allocation is diversified across asset classes, but does not identify common sources of risk

How to understand the one-period nature for MVO and multi-period setting for MCS? For example if the investment horizon is 5 year but no significant external CF, then we say it’s one period ?

MVO doesn’t account for costs associated with taxes and rebalancing. These can be incorporated into MCS.

Thanks guys. Another question, I have is - In the Schweser notes they speak of the Garbage-In-Garbage-Out issue with MVO. However, I cannot seem to find it anywhere on the CFAI notes. Can someone clarify this?

It relates to the estimations around your inputs specifically the E®