Portfolio Optimization: Constraints, Risk Tradeoffs, and Faster Solving
Summary
This research summary reviews portfolio optimization choices, including transaction-cost penalties, turnover and tracking-error limits, risk aversion, position bounds, style-factor exposure, and limits on the number of holdings. It explains that stricter style constraints can narrow the attainable risk-return range, while risk aversion or tracking-error settings can be adjusted to target different portfolio profiles. It also describes using a factorized covariance structure to reduce variance-calculation complexity from quadratic in the number of stocks to proportional to stock count times factor count.
The document reports solver timings for market-wide index-enhancement examples and notes that cardinality limits turn the problem into a mixed Boolean optimization that can be slow with many stocks. It proposes a two-stage approach that narrows the feasible region before branch-and-bound. The evidence is limited to the stated computational examples; no investment-performance results are presented. It flags model failure and extreme market conditions as risks, and notes that constraints can conflict.
Key ideas
- Portfolio objectives and constraints include transaction costs, turnover, risk, tracking error, position bounds, style exposure, and holdings count.
- Risk aversion and tracking-error settings can change the portfolio’s risk-return profile.
- Strict style exposure limits can reduce the range of feasible risk and return outcomes.
- A factorized covariance structure can lower the computational cost of portfolio variance calculations.
- A two-stage method can reduce branch-and-bound iterations for optimization problems with holdings-count limits.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.