Bond Trading: Relative Value Beyond the Curve
Fixed Income Trading and Risk Management by Alexander Düring (ISBN 9781119756354)
Previous: Curve Trading | Next: Principal Component Analysis
Chapter 31 was curve shape. Chapter 32 is bond-specific: when you care which German 5Y is cheap versus another, not whether the whole curve steepens.
Yield alone is a weak RV tool (Chapter 16 covered why). Use curve spreads matched to how you would hedge.
Which spread for which bond?
| Bond type | Natural spread |
|---|---|
| Government | Z-spread vs gov spline |
| SSA | Z-spread vs composite spline (G-spreads cross-issuer) |
| High-grade corp / covered / senior bank | I-spread or asset swap |
| Wider credit | Swap curve + CDS model (beyond this book) |
The spread that matches your hedge is the first cut. Deeper value needs substitutes, supply outlook, and investor demand. A KFW bond 2bp cheap to a KFW spline is only interesting relative to other German agencies, not as a standalone “buy.”
Splines ignore repo specials. Consistent fix would calibrate to forwards, but repo beyond overnight is opaque.
Spread widener and tightener
Expect bond to underperform curve? Widener: short bond, receive curve (asset swap or matched swap). Tightener is the reverse.
Naming follows positioning: short bond = widen, even if the spread is negative and becomes less negative.
Spline signals do not trade directly. Express via bond spread or butterfly against bonds with opposite signals. Butterflies cut duration mismatch between two bonds.
Index managers: benchmark = short the curve. Underweight = widener, overweight = tightener.
Basis trade
Short bond, long futures (or opposite). Assumes future tracks your curve and CTD maturity is close.
Hedge: N_B × PVBP_B = −N_F × PVBP_F. DV01-neutral packages often trade tighter than outright bonds.
If bond and futures CTD maturities differ, use two futures weighted by distance to the bond (linear steepening assumption). Futures need not bracket the bond; signs on notionals can differ.
Bond spread vs curve steepener
Same math as a steepener, different motive: relative valuation of two bonds, not a curve shape call.
Curve-hedged bond spread
Two bonds plus two futures to offset curve risk (e.g. 3Y-4Y bond spread hedged with 2Y and 5Y futures). Three equations for three unknowns plus aggregate risk sizing.
Residual risk remains if futures reference another market (Dutch bonds hedged with German futures).
Keep it simple
Swap futures and money market strips may replace some bilateral swap legs over time. Düring’s parting advice: over-engineered multi-leg trades need constant adjustment and muddy attribution.
Portfolio manager vs trader framing
Index-aware investors think in active weights versus benchmark. A widener is underweight; a tightener is overweight. That language connects Chapter 32 to Chapter 35: the benchmark is a synthetic short of the valuation curve, and RV trades are tilts around it.
For pure RV desks, the same trade might be expressed as a bond spread without referencing the index at all. The economics align when the spline approximates the index curve; they diverge when the issuer’s spline is thin (few SSA lines) or when composite curves blend issuers with different repo specials.
Repo and specials (again)
Düring repeats the spline limitation: no repo in standard curve fits. A bond rich on spline but special in repo can be fair or cheap in total return space. Ignoring specials is a choice for simplicity, not truth. When specials blow out (year-end, auction squeeze, regulatory window), RV signals based on Z-spread alone invert.
Chapter 32 is the bridge between spline screens and tickets. Curve chapter gave you the geometry; this one reminds you that bonds are not fungible and the hedge you pick defines the spread you should stare at.