Guide

Bond roll-down return explained

Harbor Capital runs a core fixed-income sleeve benchmarked to the Bloomberg U.S. Aggregate. In 2024 the curve was steep: 2-year Treasuries yielded 4.6%, 7-year notes 4.1%, 10-year notes 3.9%. The team held a bullet of 7-year Treasuries and reported total return of 6.2% for the year — but the 10-year yield rose 40 basis points. Coupon income contributed 4.1%. The rest came from roll-down: as each bond aged from seven years to six, it slid down a downward-sloping price-yield curve and repriced at a lower yield, lifting price even without a parallel rate cut. Roll-down added 1.8 percentage points; curve shift (rates rising) subtracted 0.7. Understanding that decomposition is how PMs decide whether to “ride the curve” or flatten exposure before inversion.

Roll-down return is the price gain (or loss) from a bond migrating to a shorter maturity bucket on an unchanged yield curve. It is distinct from carry (coupon minus financing) and from curve shift (yields at each maturity moving up or down). On a normal, upward-sloping curve, roll-down is usually positive for long positions. On a flat or inverted curve it can vanish or flip negative. This guide covers the mechanics, horizon return decomposition, interaction with duration and convexity, the Harbor Capital refactor, a technique decision table versus barbell and immunization strategies, pitfalls, and a production checklist.

What roll-down return is

A bond's remaining maturity falls every day. If the term structure is static, the bond's yield should equal the yield of a bond with that shorter maturity today. On an upward-sloping curve, shorter maturities carry lower yields. Lower yield means higher price — that price change over the holding period is roll-down return.

Example: a 5-year note yields 4.0%. One year later it is a 4-year note. If the 4-year point on today's curve is 3.7%, the note's price drifts toward the 3.7% yield level even if the 5-year point is still 4.0%. You did not need rates to fall at the 5-year tenor; you benefited from aging into a cheaper (lower-yield) bucket.

Roll-down is sometimes called riding the yield curve when managers deliberately buy intermediate maturities on a steep curve, expecting roll-down plus carry to exceed the duration loss if parallel rates rise modestly.

Horizon return decomposition

Over a short horizon (one month to one year), total bond return approximates:

Total return ≈ Carry + Roll-down + Curve shift + Other (spread, FX)

Carry — coupon accrual minus cost of funding (repo, balance-sheet charge). Often quoted as yield minus financing rate.
Roll-down — price effect from aging along today's curve shape, holding yields at each maturity fixed.
Curve shift — repricing because yields at one or more tenors moved (parallel shift, twist, or butterfly).

Risk systems often estimate roll-down with horizon yield or roll-down yield: the yield of the bond at horizon maturity minus current yield, multiplied by modified duration. More precise desks use key rate duration to attribute roll-down to specific curve points rather than assuming a parallel slide.

For a quick sanity check on a bullet portfolio:

Roll-down (approx) ≈ −ModDur × (Y_horizon − Y_current)

where Y_horizon is the yield of a bond with maturity equal to (remaining maturity minus holding period). Sign matters: on a normal curve Y_horizon is usually below Y_current, so the term is positive.

When roll-down is large, small, or negative

Steep upward-sloping curve

Roll-down is strongest when the curve is steep between your bond's current tenor and the horizon tenor. Long-end term premium elevated? Intermediate bullets (5–7 year Treasuries, 7–10 year investment-grade corporates) often harvest the most roll-down per unit of duration.

Flat curve

If 2s, 5s, and 10s all yield roughly the same, roll-down approaches zero. Carry dominates; curve positioning adds little.

Inverted curve

When short yields exceed long yields, aging moves a bond up the curve into higher-yield buckets. Roll-down becomes negative: price drifts lower as the bond “rolls up” toward higher yields. Managers who rode steep curves into inversion often shorten duration or switch to front-end carry trades.

Callable bonds and MBS

Negative convexity on callables and MBS breaks clean roll-down math. Prepayment speeds change as rates move; effective maturity shortens unpredictably. Use OAS-based horizon tools, not Treasury roll-down formulas, for those sectors.

Roll-down vs duration risk

Roll-down is a static-curve benefit. Duration measures sensitivity to parallel yield moves. A bond can earn positive roll-down while losing on curve shift if long-end yields rise faster than roll-down accrues.

Break-even analysis: roll-down plus carry must exceed expected duration loss from anticipated rate rises. If modified duration is 6, a 25 bp parallel rise costs roughly −1.5% price; roll-down of +0.4% over three months may not compensate if the Fed is still hiking.

Convexity adds curvature: for option-free bonds, positive convexity helps when rates fall sharply and hurts less when they rise. Roll-down does not replace convexity analysis on large moves.

Harbor Capital refactor (worked example)

Harbor Capital's core sleeve held a barbell (2-year + 20-year) against an Agg-like benchmark with a 6-year average maturity. In a steep 2024 curve, the barbell underperformed on roll-down: the 20-year leg had minimal aging benefit while suffering long-end volatility; the 2-year leg had high carry but almost no roll-down path.

Refactor: shift 35% of the barbell's long end into a 7-year Treasury bullet, keep 2-year liquidity bucket, hedge residual duration with 5-year Treasury futures instead of holding 20-year cash bonds.

Measured over 12 months (static-curve roll-down model plus realized returns):

Roll-down contribution: 0.4% → 1.8% (annualized)
Carry: 4.1% (unchanged within 10 bp)
Tracking error vs benchmark: 42 bp → 28 bp
Max drawdown on +50 bp parallel shock: −3.1% → −2.4%

The trade failed in a simulated 2022-style bear steepener: long-end yields jumped 80 bp while the belly underperformed. Harbor added a rule: if 2s10s spread falls below 25 bp, cut bullet weight by half and revert toward barbell neutrality.

Technique decision table

Approach	Strength	Weakness	Use when
Curve-riding bullet	Harvests roll-down on steep curves	Underperforms in inversions and bear steepeners	Normal steep curve; stable Fed; belly cheap vs history
Barbell (short + long)	Convexity; reinvestment optionality	Low roll-down on both legs	Uncertain direction; want twist exposure
Front-end carry only	High carry; low duration	Minimal roll-down and term premium	Inverted curve; hiking cycle near peak
Immunization / liability match	Matches cash-flow dates	Roll-down not an objective	Pension payouts; defined horizons
Active curve twist trades	Express 2s10s, 5s30s views directly	Requires derivatives; basis risk	Strong view on steepening or flattening

See portfolio immunization when the goal is matching liabilities rather than harvesting roll-down.

Common pitfalls

Confusing roll-down with rate cuts — roll-down works on a static curve; you do not need the Fed to ease.
Using parallel duration on twisty moves — belly yields can rise while roll-down accrues; use key-rate attribution.
Ignoring inversion — roll-down flips sign; riding the curve becomes riding into losses.
Callable and MBS roll-down — effective maturity shifts with prepayments; Treasury formulas mislead.
Over-levering carry plus roll-down — repo-financed bullets amplify curve-shift drawdowns.
Stale curve snapshots — roll-down models need current on-the-run yields at each tenor, not month-old marks.
Tax and accounting noise — amortize premiums correctly or roll-down attribution will not reconcile to P&L.

Production checklist

Decompose horizon return into carry, roll-down, and curve shift weekly.
Plot 2s5s10s30s spreads; flag when roll-down regime is weak (flat/inverted).
Compute roll-down yield from horizon maturity yield minus current yield.
Cross-check with key rate duration roll-down attribution if available.
Stress-test: parallel +50 bp, bear steepener, bull flattener scenarios.
Document break-even rate rise where curve shift wipes roll-down plus carry.
Separate Treasury roll-down models from credit OAS roll-down for corporates.
Rebalance bullet weights when 2s10s spread crosses policy thresholds.
Reconcile model roll-down to realized price change ex-coupon monthly.
Educate stakeholders: roll-down is not free alpha in every rate environment.

Key takeaways

Roll-down is the price drift from aging down an upward-sloping yield curve.
Total short-horizon return decomposes into carry, roll-down, and curve shift.
Harbor Capital lifted roll-down contribution from 0.4% to 1.8% with a 7-year bullet on a steep curve.
Inverted and flat curves erase or reverse roll-down; adjust positioning accordingly.
Compare curve-riding bullets against barbells and immunization at equal duration risk.