FRM Part II · FRM Exam Part II · Portfolio Construction

A portfolio manager's mean-variance optimizer, run without any constraints, recommends a highly concentrated portfolio with large long and short positions in a few assets whose expected returns are only marginally different from one another. Which is the most appropriate explanation and response?

Unconstrained optimizers magnify small estimation errors in expected returns and covariances into extreme, concentrated long-short weights. Adding position limits or shrinking the inputs reduces this sensitivity, so it is the appropriate response, rather than implementing the weights as if they reflected true mispricings.

1. AThe optimizer is error-maximizing: small estimation errors in inputs are magnified into extreme weights, so adding position limits or shrinking inputs is appropriateCorrect
2. BThe optimizer is correctly identifying genuine mispricings, so the weights should be implemented in full
3. CThe result shows that covariance estimates are too stable, so they should be replaced by a single-day estimate
4. DThe result shows that the efficient frontier has no solution, so the manager should switch to equal weighting only when returns differ greatly