Fixed Income Toolkit

This is an ongoing quant research project for fixed-income analytics, yield-curve modeling, and bond portfolio simulation/optimization.
It provides tools to fetch ECB yield data, fit Nelson-Siegel (NS) and Svensson (NSS) curves, estimate time-varying factors with an EKF+RTS smoother, run VAR forecasts, simulate multi-factor scenario returns, and construct CVaR-optimized bond portfolios.

---

Attention!

This project is a research-grade demo, not production. Limitations include:

• ECB SDMX endpoints and field names can change — the fetcher may need maintenance.
• Curve fitting and EKF hyperparameters are heuristic and dataset-dependent.
• Bond pricing uses simplified day-count assumptions and continuous compounding.
• Scenario generation is model-based; it does not replace market-calibrated risk models.
• CVaR optimizer depends on scenario quality.

---

Project Objectives

• Provide a way for parametric yield modeling (NS / NSS).
• Fetch and preprocess historical Euro area yield curves from ECB.
• Fit per-date parametric curves and produce time series of factors.
• Estimate dynamic factors (level, slope, curvature, and tau) using EKF + RTS smoothing.
• Forecast factor evolution using VAR and produce stochastic yield scenarios.
• Simulate bond returns under multi-factor NS shocks (and optional credit spread shocks).
• Construct portfolios with CVaR minimization, optionally with duration constraints.
• Provide a rolling backtest wrapper to prototype strategy performance.

---

Model Overview

Nelson-Siegel (NS) & Svensson (NSS)

• NS models yields with 3 factors plus a decay parameter (tau).
• NSS adds a fourth factor and second decay parameter to capture more curve shapes.
• Both models allow reconstructing yield curves from factor states.

EKF + RTS Smoother

• Extended Kalman Filter (EKF) estimates dynamic factor states from historical yields.
• RTS smoother refines factor estimates using forward-backward smoothing.
• Supports both NS (4-state) and NSS (6-state) models.

VAR Forecasting & Stochastic Scenarios

• Fits a VAR model to smoothed factor states.
• Simulates forward factor scenarios and reconstructs yield curves for each scenario.
• Produces scenario clouds and mean forecasts for portfolio simulations.

Bond Analytics & Scenario Returns

• Price fixed-rate bonds using discounting of coupons under parametric yield curves.
• Compute Macaulay duration, modified duration, and convexity.
• Generate scenario-based bond returns using factor shocks.
• Optimize portfolio weights via CVaR minimization with optional constraints.

Backtesting

• Rolling backtest framework using historical yields.
• Rebalances portfolios at configurable intervals.
• Supports transaction costs and evaluates realized strategy performance.