Changelog¶

Version 1.6.0 - 2026-01¶

Python 3.14 Support: Added official support and testing for Python 3.14.
Risso Datasets: Updated return calculation methodology in garpar.datasets.risso for improved accuracy and consistency.
Dependencies: Updated all project dependencies to their latest stable versions.
Tox Configuration: Enhanced test environment configuration with:
- Descriptive documentation for each test environment
- Environment categorization using labels (static/dynamic/docs)
- Enabled isolated builds for better dependency isolation
- Enabled notebook execution in documentation build process
- Updated workflow references and configuration
Code Quality: Fixed docstring style to use imperative mood, achieving pydocstyle D401 compliance.

First stable relase
Implemented MVOptimizer with new mean-variance optimization models.
Added support for the Markowitz model (Markowitz class) for portfolio optimization.
Introduced UtilitiesAccessor with tracking error and quadratic utility calculations.
Implemented HDF5 storage capabilities in garpar_io.py for saving and loading stock sets.
Added StocksSet.to_dataframe() method for easier data extraction.
Introduced new pruning methods (weights_prune and delisted_prune) to optimize datasets.
Refactored StocksSet to improve weight and entropy handling.
Improved metadata management with GARPAR_METADATA_KEY.
Optimized covariance and correlation calculations for better performance.
Resolved issues with missing data in covariance matrix calculations.
Fixed incorrect behavior in to_hdf5() when handling metadata attributes.
Addressed rounding errors in weight calculations for portfolio optimization.
Deprecated older portfolio optimization methods in favor of new mean-variance models.
Removed redundant utility functions now covered by UtilitiesAccessor.

For a full list of changes, visit: Garpar Repository.

Pre relase don’t use.