Roll-down return (or "riding the yield curve") is the capital gain a bond earns simply by aging in an upward-sloping yield curve environment. As time passes, a bond's remaining maturity shortens, and if the curve is upward-sloping, it "rolls down" to a lower yield, which means a higher price.
For example, consider a 10-year Treasury with a yield of 4.25%. If the 9-year point on the curve is at 4.15%, then over the next year (assuming the curve shape doesn't change), the bond rolls from the 10-year point to the 9-year point, capturing approximately 10 bps of yield decline. The price gain from this roll is approximately:
Roll-down return = (10Y yield - 9Y yield) x modified duration
Roll-down is a component of total expected return alongside carry (coupon income minus financing cost). Together, carry and roll-down form the breakeven rate move: how much yields must rise before a long position loses money.
Roll-down is largest where the curve is steepest. In a normal curve environment, the 2-year to 5-year sector often offers the best roll-down per unit of duration risk. Roll-down provides zero benefit in a flat or inverted curve.
Portfolio managers use roll-down analysis to identify the most attractive maturities for buy-and-hold strategies and to compare relative value across the curve.
Measuring roll-down takes three steps.
The second step is where most of the work sits. A bond held for three months moves from the 10 Yr point to the 9.75 Yr point. The Treasury does not publish a 9.75 Yr yield, so the curve must supply it.
The third step is an approximation. It ignores convexity, which is small over short horizons and small yield moves. Over a long horizon or a steep segment, a full repricing gives a more accurate answer.
Every roll-down figure rests on one assumption. The curve does not move. That assumption is never true. It is still useful, because it separates the return that comes from the passage of time from the return that comes from a change in rates.
Roll-down depends on the slope of the curve at the bond's own tenor. It does not depend on the level of yields. A curve at 5% and a curve at 2% deliver the same roll-down if both have the same local slope.
This has a direct consequence. The best roll-down is found on the steepest segment of the curve, not on the highest yielding part of it. A long bond can offer the highest yield on the curve and almost no roll-down, because the curve is nearly flat between 20 Yr and 30 Yr.
The right comparison is roll-down per unit of duration. A bond with more duration converts a given yield decline into a larger price gain, so raw roll-down flatters long bonds. Dividing by modified duration removes that effect and shows which tenor pays best for the risk taken.
An inverted segment reverses the sign. A bond aging into a higher yield loses price as it rolls. Roll-down is negative there, and it adds to negative carry rather than offsetting it.
The unchanged curve assumption hides a bet, and it is worth stating plainly.
An upward sloping curve implies forward rates that sit above today's spot rates. Under no arbitrage, those forwards are the rates the curve would need to deliver for every tenor to return the same amount. If the curve moves to match its own forwards, roll-down disappears. The bond ages into a lower point on the curve, but the whole curve has risen by an offsetting amount.
So a roll-down trade is a bet that the forwards do not come true. That is not a reckless bet. Forward rates have historically overstated how much short rates actually rose, which is the same evidence that supports a positive term premium. It is a bet all the same, and calling it one keeps expectations honest.
This is why roll-down is quoted next to carry and never on its own. The two together give the breakeven, which states how far yields must rise before the position loses money.
It requires the segment around the bond's own tenor to slope upward. The rest of the curve does not matter. A curve that is inverted at the front and upward sloping past 5 Yr still offers positive roll-down to a bond in the 7 Yr sector.
Yes, through duration rather than through the curve. Two bonds at the same tenor roll down the same yield decline, because the curve does not care about the coupon. The price gain differs, because it depends on modified duration, and a higher coupon shortens duration at a given maturity. A high coupon bond therefore converts the same yield decline into a smaller price gain.
Because roll-down comes from the yield difference between two nearby points on the curve. If the curve is flat, a bond that ages by three months rolls to a point with the same yield. There is no yield decline, so there is no price gain.