QuantFinance — Fixed Income Analytics

A practitioner-level fixed income analytics framework for bank treasury and ALM contexts. Covers the full stack from yield curve foundations to carry decomposition, OCI risk, and Monte Carlo capital simulation.

---

Notebook Series

The notebooks are designed as a progressive series. Each layer assumes the one before it.

Layer 0 — Yield Curves: Theory and Practice Yield_Curve_Theory_and_Practice.ipynb

Yield curve construction from live US Treasury data, three interpolation methods (cubic spline, Nelson-Siegel, NSS), spot/forward/discount factor decomposition, bond pricing off the curve, duration and convexity, scenario analysis, and trading strategies (roll-down, bullet vs barbell). Uses the YieldCurve, Bond, and interpolation classes from the FixedIncome package.

Layer 1 — Carry, OCI, and Monte Carlo CET1 fixed_income_treasury_analytics.ipynb

Carry decomposition across a three-bond UK gilt portfolio, a deterministic OCI scenario tool (IFRS 9 / CET1 impact of a +100 bps parallel shift), and a Monte Carlo framework producing a full CET1 distribution with tail statistics. An interactive Streamlit version of the OCI tool is in Apps/oci_tool.py.

Layer 2 — Portfolio Construction and DV01 Hedging quant_prep_treasury.ipynb

Portfolio representation and DV01 computation for a synthetic UK bank treasury book (gilts + ESHLA loans). Builds to a CVXPY optimisation with regulatory constraints (Basel III RWA, LCR) and an internal DV01 limit, showing which constraints bind and why.

Layer 3 — Convexity, ALM Constraints, and Regime Detection fixed_income_convexity_alm.ipynb

Convexity as a second-order correction to DV01-based P&L; the NII vs OCI trade-off as the binding ALM constraint in a banking book; and a Hidden Markov Model applied to interest rate regime detection.

---

Interactive App

OCI Scenario Tool (Apps/oci_tool.py) — Streamlit companion to Layer 1. Applies a parallel rate shock to the AFS gilt portfolio and computes the mark-to-market OCI loss and CET1 impact interactively.

streamlit run Apps/oci_tool.py

---

Structure

QuantFinance/
├── Apps/
│   └── oci_tool.py                          # Streamlit OCI scenario tool
├── FixedIncome/
│   ├── bonds.py                             # Bond pricing, DV01, duration, convexity
│   ├── yield_curve.py                       # YieldCurve class
│   ├── interpolation.py                     # Cubic spline, Nelson-Siegel, NSS
│   ├── data_fetchers.py                     # US Treasury data fetchers
│   └── README.md                            # Package API reference
├── Notebooks/
│   ├── Yield_Curve_Theory_and_Practice.ipynb
│   ├── fixed_income_treasury_analytics.ipynb
│   ├── quant_prep_treasury.ipynb
│   ├── fixed_income_convexity_alm.ipynb
│   └── README.md                            # Notebook descriptions
├── requirements.txt
└── README.md

---

Dependencies

numpy  pandas  scipy  matplotlib  cvxpy  hmmlearn  streamlit

pip install -r requirements.txt

---

DV01 Conventions

Layer 2 computes portfolio-level DV01 in £m per basis point using notional × modified_duration × 0.0001. Layers 1–3 use bump-and-reprice via compute_dv01, returning £ per £100 face per basis point. Both are standard approaches; the difference reflects portfolio-level vs. instrument-level analysis.