Bond Relative Value — Euro SSA & Repo Desk

End-to-end fixed income relative value framework for a Euro SSA repo desk: market-implied repo funding analysis, Z-spread and ASW modelling, and a synthetic Bund CTD specialness model — with a live Streamlit dashboard.

Live Dashboard

Launch the interactive dashboard

No installation required, it runs directly in your browser !

---

Overview

This project models two complementary views of the euro repo market, intentionally kept separate to reflect how a rates/repo desk actually thinks about fixed income relative value:

Framework	Notebook	Purpose
Market-implied repo	repo_funding_model	SSA carry, break-even funding, Z-spread & ASW vs Bund curve
Specialness-driven repo	specialness_model	CTD identification, OU specialness model, basis trade P&L

The two frameworks answer different questions:

• "Is this SSA bond cheap enough to carry vs its funding cost?" → Framework 1
• "Why is the Bund CTD trading special in repo, and how much carry does it generate?" → Framework 2

---

Repository Structure

bond-relative-value/
├── notebooks/
│   ├── repo_funding_model.ipynb     # Framework 1 — SSA funding & RV
│   └── specialness_model.ipynb      # Framework 2 — CTD specialness
├── dashboard/
│   └── repo_dashboard.py            # Streamlit interactive monitor
├── data/                            # Generated charts & outputs
├── requirements.txt
└── README.md

---

Methodology

Framework 1 — Market-Implied Repo (repo_funding_model.ipynb)

Objective: Quantify the net carry and relative value of euro SSA bonds (KfW, EIB) versus Bund benchmarks (Bobl 5Y, Bund 10Y), explicitly accounting for repo funding costs.

Bond universe (prices sourced from Deutsche Börse XFRA & issuer websites as of 21 Apr 2026):

Bond	ISIN	Coupon	Maturity	Repo spread vs GC
KfW 5Y	XS3344416287	2.875%	Jun-31	−2 bps
KfW 10Y	XS3326554261	4.680%	Mar-36	−3 bps
EIB 5Y	EU000A4EPCA0	2.625%	Jun-31	−3 bps
EIB 10Y	EU000A4EM8H8	3.000%	Jan-36	−4 bps
Bobl 5Y	DE000BU25067	2.500%	Apr-31	−8 bps
Bund 10Y	DE000BU2Z064	2.900%	Feb-36	−12 bps

Modelling choices:

① GC repo proxy $$r_{GC}(t) = \text{ESTER}(t) - 10\text{bps}$$ Calibrated on ICMA European Repo Market Survey (Dec 2025, published Mar 2026). ESTER sourced live from ECB API (EST/B.EU000A2X2A25.WT).

② Net carry (ACT/365 coupon · ACT/360 repo) $$\text{Net Carry} = \frac{C \cdot \Delta t}{365} - r_{repo} \cdot \frac{P \cdot \Delta t}{360}$$

③ Break-even repo rate — maximum funding cost before carry turns negative: $$r_{BE} = \frac{C}{P} \times \frac{360}{365}$$

④ Z-spread — constant spread over the ECB AAA Bund curve (Svensson model, 9 tenors, cubic spline interpolation) computed via Brent's method: $$P_{market} = \sum_i CF_i \cdot e^{-(r_{spot}(t_i) + z) \cdot t_i}$$

⑤ Asset Swap Spread (ASW) — spread over the EUR swap curve, proxied as ECB AAA curve − 25 bps (consistent with the structural negative swap spread documented post-2015 ECB QE): $$ASW = \frac{(C - r_{swap}) \cdot A + (1 - P/100)}{A}$$

ASW requires live Bloomberg EUSA quotes for production use. The proxy introduces a ~20 bps systematic bias — Z-spread is the reliable signal in this implementation.

---

Framework 2 — Specialness-Driven Repo (specialness_model.ipynb)

Objective: Model the repo specialness dynamics of the Bund CTD (Cheapest-To-Deliver) for the FGBLM6 futures contract (delivery June 2026), and compute the carry P&L of a specialness trade.

Part I — CTD Identification

Using official Eurex conversion factors (sourced from eurexchange.com/ex-en/data/clearing-files) and real cash prices from Deutsche Börse (21 Apr 2026), the CTD is identified as the bond with the lowest net basis:

$$\text{Net Basis}_i = (P_i - F \times CF_i) - (\text{coupon cash} - P_i^{dirty} \times r_{GC} \times t)$$

CTD identified: DBR 2.5% Feb-35 (DE000BU2Z049), stable under ±1pt futures shock and ±50 bps GC shock (sensitivity grid: 99 scenarios, 100% CTD stability).

Part II — Ornstein-Uhlenbeck Specialness Model

Since live special repo rates are not publicly available, specialness is modelled synthetically via a two-component framework:

Deterministic mean term structure — calibrated on Bund repo empirical ranges (ICMA ERCC 2022, IMF WP/18/258, ECB WP 2065): $$\mu(\tau) = \gamma + \alpha \cdot e^{-\beta\tau}$$

with $\alpha = 120$ bps (peak component), $\beta = 0.04 \text{ day}^{-1}$ (acceleration speed), $\gamma = 2$ bps (far-from-expiry floor).

Stochastic dynamics (Euler-Maruyama discretisation): $$dS_t = \kappa(\mu(\tau_t) - S_t)dt + \sigma dW_t, \quad S_t \geq 0$$

with $\kappa = 0.15 \text{ day}^{-1}$ (mean reversion), $\sigma = 3 \text{ bps}/\sqrt{\text{day}}$ (daily volatility).

Carry P&L formula — daily carry on notional $N$ at cash price $P$: $$\text{Daily PnL} = S_t \times \frac{N \times P/100}{10{,}000} \times \frac{1}{360}$$

Part III — Sensitivity & Stress Tests

One-at-a-time OU parameter sensitivity (±20% on α, β, κ, σ) and a CTD switch stress test across 4 macro/idiosyncratic scenarios.

---

Key Results

Framework 1 — Carry Analysis (30d horizon, €10M)

Bond	Net carry (bps)	Break-even repo	Safety margin
KfW 10Y	+23.5	4.62%	+281 bps
EIB 10Y	+10.0	3.00%	+122 bps
Bund 10Y	+9.7	2.89%	+117 bps
KfW 5Y	+8.5	2.87%	+103 bps

KfW 10Y dominates on carry due to its 4.68% coupon issued at the 2024 rate peak — a structural outlier. For risk-adjusted RV, EIB 10Y (+10 bps carry, +122 bps safety margin, Z-spread +26 bps) offers the most balanced profile across all three metrics.

Framework 2 — Specialness Carry Trade

Metric	Value
CTD (FGBLM6)	DBR 2.5% Feb-35 (DE000BU2Z049)
Current specialness (50d to expiry)	~18 bps
Projected at delivery	~100 bps (mean), 93–108 bps (P10/P90)
Trade: Entry 50d → Exit 10d, €10M	Mean P&L €3,853
P10 / P90	€3,455 / €4,270
Most sensitive parameter	α (±20% → ±€724 P&L impact)

---

Interactive Dashboard

Four tabs covering both frameworks:

• SSA Monitor — Live Z-spreads vs ECB AAA curve, bond universe table
• Repo Funding — Net carry decomposition, break-even repo, stress test
• CTD Specialness — OU model visualisation, three-level repo bridge (GC / SSA / CTD implied repo), carry P&L fan chart
• RV Matrix — Composite ranking across Z-spread, carry, safety margin

pip install -r requirements.txt
streamlit run dashboard/repo_dashboard.py

---

Getting Started

git clone https://github.com/mb69-code/bond-relative-value
cd bond-relative-value
pip install -r requirements.txt
jupyter notebook notebooks/repo_funding_model.ipynb

---

References

• Duffie, D. (1996). Special repo rates. Journal of Finance, 51(2).
• Buraschi, A. & Menini, D. (2002). Liquidity risk and specialness. Journal of Financial Economics, 64(2), 243–284.
• Arrata, W. et al. (2020). The scarcity effect of QE on repo rates. IMF Working Paper WP/18/258.
• Corradin, S. & Maddaloni, A. (2020). The importance of being special. ECB Working Paper No. 2065.
• Hill, A. (2022). r is not a constant. ICMA ERCC.
• ICMA (2026). European Repo Market Survey No. 50. March 2026.
• ECB Statistical Data Warehouse — data-api.ecb.europa.eu
• Eurex — Bund Future contract specifications & conversion factors