PyPortfolio/PyPortfolioOpt
PyPortfolioOpt
PyPortfolioOpt is a Python library for financial portfolio optimization that implements mean-variance optimization, Black-Litterman models, and Hierarchical Risk Parity methods. It provides a complete toolkit for constructing risk-adjusted asset portfolios by combining expected return estimation, covariance modeling, constraint handling, and discrete allocation into a single optimization framework.
The library distinguishes itself through its integration of multiple optimization approaches within a unified interface. It includes a Black-Litterman Bayesian framework that blends market equilibrium returns with investor views, a Hierarchical Risk Parity method that allocates capital based on asset correlation clustering, and a mean-variance optimization engine that solves the efficient frontier with user-defined constraints. The constraint-aware solver enforces position bounds, market neutrality, and long-short limits during weight optimization, while the discrete share allocation algorithm converts continuous portfolio weights into integer share counts for a given investment amount.
The toolkit covers the full portfolio management workflow, from estimating expected returns using historical averages, exponential weighting, or CAPM, to modeling portfolio risk through sample, shrinkage, or robust covariance matrix estimation. It then applies these inputs to compute optimal asset allocations under specified constraints, producing investable portfolios ready for implementation.
Features
- Multi-Method Portfolio Optimizers - Provides the core portfolio optimization library implementing mean-variance, Black-Litterman, and HRP methods.
- Constraint Enforcement - Enforces position bounds, market neutrality, and long-short limits during weight optimization.
- Optimization Methods - Implements mean-variance optimization, Black-Litterman models, and Hierarchical Risk Parity for asset allocation.
- Portfolio Weight - Enforces position bounds, market neutrality, and long-short limits during portfolio weight optimization.
- Portfolio Optimization - Computes optimal asset allocations using mean-variance, Black-Litterman, and Hierarchical Risk Parity methods.
- Efficient Frontier Solvers - Maximizes risk-adjusted returns by solving the efficient frontier with user-defined constraints.
- Mean-Variance Optimization Solvers - Computes optimal asset weights by solving quadratic programming problems for the efficient frontier.
- Asset Return Estimation - Calculates forward-looking expected returns using historical mean, exponential weighting, or CAPM.
- Financial Portfolio Return Estimators - Implements the Black-Litterman model, a Bayesian framework for blending market returns with investor views.
- Portfolio Return Prior Models - Implements the Black-Litterman model for combining market equilibrium returns with investor views.
- Portfolio Constraint Enforcement - Restricts asset allocations with bounds, market neutrality, or long-short limits during optimization.
- Portfolio Constraint Engines - Restricts asset allocations with bounds, market neutrality, or long-short limits during optimization.
- Covariance Matrix Calculators - Estimates covariance matrices from asset returns using sample, shrinkage, and robust statistical methods.
- Hierarchical Risk Parity - Builds portfolios by recursively clustering assets based on correlation distances and allocating risk equally.
- Portfolio Risk Metrics - Estimates covariance matrices and quantifies portfolio risk through sample, shrinkage, or robust methods.
- Integer Share Allocation Algorithms - Converts continuous portfolio weights into integer share counts for investable portfolios.
- Portfolio Share Allocation Algorithms - Converts continuous portfolio weights into integer share counts for investable portfolios.
Star history
How this analysis was created: This summary and feature list were written by an AI model that read the project's README and public documentation pages. Each feature links to the documentation it came from; stars, license and language come straight from the GitHub API. The model does not read the source code, and the analysis is refreshed when the project is re-analysed. Learn more on our About page.
