Portfolio Optimization Master

What Problem It Solves

When designing asset allocations, return targets, sample covariance, investor views, risk contributions, and clustering structure are often handled as separate scripts. portfolio-optimizer-zx puts these steps into one Python workflow: it pulls A-share daily data from akshare, estimates covariance, and produces comparable weight solutions from different optimizers. It is better suited for quant learning and research prototyping than live trading.

How It Works

The tool covers four common strategies:

• Markowitz: computes maximum Sharpe, minimum variance, and the efficient frontier, with weight bounds and shorting constraints.
• Black-Litterman: derives equilibrium returns from market weights, blends absolute or relative views, and estimates view uncertainty automatically.
• Risk parity: targets similar risk contributions from each asset, then reports marginal risk contribution and risk share.
• HRP: allocates weights through hierarchical clustering and recursive bisection, avoiding direct covariance matrix inversion.

It then summarizes annualized return, volatility, Sharpe ratio, maximum drawdown, Calmar, Sortino, information ratio, VaR, and CVaR, and can generate an HTML report with the efficient frontier, weight pie chart, risk contribution, and correlation heatmap.

Boundaries

The skill currently depends on A-share interfaces and daily data, so it is not intended for high-frequency trading, options/futures portfolios, or real-time execution. akshare failures, short samples, unstable covariance estimates, and biased investor views can all affect results. Treat the output as research material, not investment advice.