Quant Memo
Core

Duration-Neutral vs Cash-Neutral Curve Trades

Betting that a yield curve steepens or flattens sounds like a single trade, but sizing the two legs by equal dollar notional (cash-neutral) versus equal dollar-value-of-a-basis-point (duration-neutral) produces two very different exposures to parallel rate moves.

Prerequisites: DV01 and PV01, Bond Duration and Convexity

A trader who thinks the curve will steepen buys a short-maturity bond and sells a long-maturity bond — or is it the other way around? Before that question even matters, a bigger one does: how much of each? Buy $100 million of both legs and you have a cash-neutral trade that is still secretly a large bet on the overall level of rates, because a 10-year bond moves far more per basis point than a 2-year bond. Size the legs so their basis-point sensitivities offset and you have a duration-neutral trade, isolating the curve shape and nothing else.

Cash-neutral means equal dollar notional on both legs. Duration-neutral means equal dollar-value-of-a-basis-point (DV01) on both legs. Only the duration-neutral version is a pure bet on curve shape; the cash-neutral version is still substantially exposed to the level of rates.

Why DV01, not notional, is the right ruler

A bond's DV01 is the dollar change in its price for a one-basis-point move in yield. A 2-year note has a small DV01 because there is little time for a rate change to compound into price; a 10-year note has a DV01 several times larger for the same face amount, because more distant, larger cash flows are more sensitive to the discount rate. Weighting a curve trade by notional treats a dollar of 2-year risk as equivalent to a dollar of 10-year risk, which it is not — weighting by DV01 makes a basis point of one leg's move worth the same as a basis point of the other's, which is the only sense in which the trade is "neutral" to a parallel shift.

Worked example: sizing a 2s10s steepener

Say the 2-year note has a DV01 of $1,900 per $100 million face, and the 10-year note has a DV01 of $8,700 per $100 million face. A trader wants to put on a duration-neutral steepener: long $100 million of the 2-year, short enough 10-year to offset it.

10y face (millions)=100×1,9008,700=21.8\text{10y face (millions)} = 100 \times \frac{1{,}900}{8{,}700} = 21.8

so the short leg is about $21.8 million face of the 10-year. Check: the 2-year leg's DV01 is $1,900 (on $100m face); the 10-year leg's DV01 on $21.8m face is 8,700×(21.8/100)=1,8978{,}700 \times (21.8/100) = 1{,}897, i.e. about $1,897 — matching the 2-year leg to within rounding. A 10-basis-point parallel rise in both yields moves the long 2-year leg by about $19,000 of loss and the short 10-year leg by about $18,970 of gain — they roughly cancel. If instead the 2-year yield rises 5 basis points while the 10-year falls 5 basis points (the curve steepens), the position gains on both legs, because that is precisely the shape it was built to capture.

Worked example: what cash-neutral actually is

Now size the same idea cash-neutral instead: $100 million long 2-year, $100 million short 10-year. The 10-year leg's DV01 is $8,700 versus the 2-year leg's $1,900 — a net short DV01 of $6,800. A parallel 10-basis-point rally (yields fall across the board) would lose roughly 6,800×10=68,0006{,}800 \times 10 = 68{,}000, i.e. about $68,000, on the net short-duration position, regardless of whether the curve steepens or flattens at all. The "curve trade" is actually a large, disguised bet that rates rise, dwarfing whatever the curve-shape view was worth.

cash-neutral ($100m / $100m) 2y DV01 $1,900 10y DV01 $8,700 net $6,800 short duration

duration-neutral (100m/100m / 21.8m) 2y DV01 1,900</text><rectx="390"y="40"width="40"height="31"class="frisk"/><textx="410"y="86"textanchor="middle"class="muted">10yDV01</text><textx="410"y="98"textanchor="middle">1,900</text> <rect x="390" y="40" width="40" height="31" class="f-risk" /> <text x="410" y="86" text-anchor="middle" class="muted">10y DV01</text> <text x="410" y="98" text-anchor="middle">1,897 net ≈ $0, pure curve bet

Same trade idea, two sizings. Only the DV01-matched version isolates curve shape from the level of rates.

Duration-neutral at one instant does not stay duration-neutral. DV01s change as yields move and as bonds age toward maturity, so a curve position needs periodic rebalancing to stay level-neutral, and desks track this drift explicitly rather than assuming a trade set up neutral today remains neutral next week.

Beyond parallel: curve trades still carry curve risk

Duration-neutral only cancels a parallel shift — it does nothing to protect against the curve twisting in some other shape, which is exactly the risk the trade is designed to express. It also does not cancel convexity: a 10-year note's price is more convex than a 2-year note's, so a very large parallel move (not just a small one) leaves a residual gain or loss even on a DV01-matched book, because DV01 itself changes as yields move. Desks running curve books track this second-order effect and periodically rebalance the notional ratio, not just at inception but as the trade ages and yields drift.

Where it shows up

Curve steepeners and flatteners — 2s10s, 5s30s and similar — are the standard macro rates expression of a view on the shape of the curve (recession fears flattening it, growth and inflation re-steepening it), and they are essentially always run duration-neutral for exactly the reason above: a directional rates view belongs in an outright long or short bond position, not smuggled into a curve trade by accident.

Key terms

  • DV01 (dollar value of a basis point) — dollar price change of a position for a one-basis-point yield move.
  • Cash-neutral — equal dollar notional on both legs of a trade.
  • Duration-neutral — equal DV01 on both legs; isolates exposure to curve shape from the level of rates.
  • Steepener / flattener — a trade that profits if the spread between two maturities widens (steepener) or narrows (flattener).

Related concepts

Practice in interviews

Further reading

  • Tuckman & Serrat, Fixed Income Securities (ch. 6)
  • Fabozzi, Bond Markets, Analysis, and Strategies (ch. 6)
ShareTwitterLinkedIn