Towards a Portfolio Governance Monitoring Framework for Boards
Inspired by MAN AHL’s “Hidden Dependencies” article by Steven Desmyter: http://bit.ly/4wrZ0Iy
Because their DEAR framework has not yet been published, I started exploring how a more general multi-objective portfolio preference framework might work in practice. (any errors or differences in interpretation are mine )
More than 70 years after Harry Markowitz introduced Mean-Variance Optimisation, it is becoming increasingly clear that real-world portfolio construction is not a single-objective problem.
Boards and investment committees rarely care only about maximising expected return or minimising volatility. They also care about:
• Downside protection
• Drawdowns
• Diversification quality
• Regime sensitivity
• Stress / tail behaviour
• Path dependency
and desirable upside characteristics such as positive asymmetry and skew.
In practice, portfolio construction is fundamentally a balancing act between what we want (return, diversification, convexity/asymmetry) and what we are willing to pay to reduce what we dislike (tails, stress, drawdowns and concentration).
This led me to develop a preliminary Portfolio Preference Score (PPS / “DEAR-like”) framework which attempts to decompose portfolios into:
✅ Expected Return
✅ Diversification Benefit
✅ Positive Asymmetry / Convexity
❌ Downside Tail Penalty
❌ Stress Penalty
❌ CDaR (Conditional Drawdown at Risk) Penalty
while also incorporating regime-aware behaviour and path-dependent risk using the LP formulation of Stan Uryasev 's Conditional Drawdown at Risk (CDaR).
The goal is not to replace existing optimisation methods, but to move toward a more transparent governance framework where boards can explicitly see:
• the trade-offs between protection and upside,
• the economic cost of reducing drawdowns,
• how portfolios behave under different market regimes,
• and whether the “cost of protection” is justified by the resulting portfolio quality.
One interesting observation is that portfolio attractiveness can change materially even when portfolio weights remain almost unchanged - simply because the underlying regime probabilities change.
There are still several unresolved issues, and this is very much a work in progress.
The current framework does not yet perfectly separate “likes” and “dislikes” into completely orthogonal components, and some interaction effects remain difficult to interpret cleanly.
Version 2 will likely move toward a directly user-weighted preference utility framework, allowing boards and allocators to explicitly express the relative importance of return, diversification, asymmetry, stress protection and drawdown control within a single coherent optimisation structure - without having to ever use the word “lambda”.
Early days - but potentially an interesting step toward more explainable, governance-friendly portfolio construction.
For more information, contact Peter Urbani : peter.urbani@knowrisk.co.nz
You can download the presentation here:
Comments