yieldrep is a Python research project for studying whether latent representations
of sovereign yield curves contain information beyond classical term-structure
features.
The project uses public fixed-income data to compare learned curve representations against baselines such as PCA, Nelson-Siegel factors, slope and curvature measures, and carry/roll-down style features across forecasting, relative-value, volatility, and curve-state tasks.
This project is developed as a learning and research effort with AI assistance for code generation, refactoring, and documentation. Design decisions, review, testing, and project direction are handled by the author.