Quant Portfolio Opt

Contract

• Input: problem description and inputs defined by the skill body.
• Output: Markdown artifact with completed process steps.
• Side effects: none.
• Dependencies: none.
• Stop condition: all process steps executed; artifact saved with required sections.
• Risk: low.
• Boundary: produces reasoning artifact only; no system changes.

Portfolio Optimisation

Build a portfolio from first principles — return model, risk model, objective, constraints — and solve for weights with explicit robustness checks.

When to use

• The user wants to construct an optimal portfolio (minimum variance, max Sharpe, risk-parity, etc.).
• Asset allocation decisions need a quantitative basis.
• A factor strategy or multi-asset portfolio needs weight optimisation.

Process

1. Define the problem

State:

• Universe — assets, time window, currency.
• Return model — historical mean, CAPM alpha, factor model; if historical, use shrinkage.
• Risk model — sample covariance, factor model covariance, shrinkage (Ledoit-Wolf), or realised vol.
• Objective — minimum variance, maximum Sharpe, risk-parity, maximum diversification, minimum drawdown.
• Constraints — long-only, box, sector, turnover, leverage, carbon / ESG screens.

Completion criterion: universe, return model, risk model, objective, and constraints all explicit.

2. Handle estimation error

Historical covariance and mean are noisy. Address:

• Shrinkage on covariance (Ledoit-Wolf or oracle approximating).
• Resampled efficiency or Bayesian posterior for mean.
• Black-Litterman to blend market-implied equilibrium with views.
• Dropout estimation — what fraction of estimated alpha survives out-of-sample?

Completion criterion: at least one technique to combat estimation error applied.

3. Solve

Run the optimisation:

• Quadratic solver (QP) for mean-variance.
• Non-linear for risk-parity (risk budgeting).
• MIP if cardinality constraints (max N assets).
• Report weights, expected return, expected vol, Sharpe ratio.

Completion criterion: weights computed; metrics (return, vol, Sharpe) reported.

4. Robustness checks

• Sensitividad — perturb each input by ±10%; which weights change most?
• ** turnover** — if rebalanced monthly, annual turnover and transaction cost drag.
• Regime — does the portfolio behave in 2008 / 2020 as it did in calm markets?
• Corner solutions — if weights are extreme (e.g. 90% in one asset), re-run with tighter constraints.

Completion criterion: each check completed; extreme or fragile weights flagged.

5. Deliver

Markdown artifact: universe, return/risk models, objective, constraints, weights (table), metrics, robustness checks, and the key sensitivity — which input moves the most.

Completion criterion: deliverable complete; extreme weights acknowledged.