Relative value analysis in fixed income compares yields, spreads, or expected returns across securities to identify which are cheap (high expected return) or rich (low expected return) relative to their peers.
The Salomon Brothers yield curve primer (Part 6) lays out the standard relative value framework:
Common relative value metrics:
Relative value trading differs from directional trading in that it takes offsetting positions. A relative value trader might go long a cheap 7-year note and short the rich 10-year note, profiting from convergence regardless of the direction of rates.
This approach is the bread and butter of fixed-income proprietary trading desks and dedicated relative value hedge funds. It requires detailed data across the curve, and the charting tools on this site are built for exactly that kind of analysis.
No bond is cheap on its own. It is cheap against something, and naming that something is the first step of any relative value analysis.
The usual reference is a fitted curve. An analyst draws a smooth curve through the yields of many Treasuries, then measures how far each bond sits above or below it. A bond yielding more than the fitted curve predicts is cheap. A bond yielding less is rich. The distance is called the residual.
The choice of curve is not neutral. A flexible curve hugs the observed yields and reports small residuals. A stiffer curve reports large ones. Two analysts can therefore disagree about which bonds are cheap while both apply their own method correctly.
A second reference is history. A spread that held a narrow band for two years and now sits outside it looks like an opportunity. That reasoning assumes the band still applies. Regimes change, and a spread that never comes back was not mispriced. It was repriced.
Yield alone cannot rank bonds of different maturities, because it ignores both the funding cost and the passage of time. Total expected return fixes this by adding three components.
The first two are mechanical. Given a curve and a funding rate, they can be calculated exactly. Only the third requires a forecast, which is why carry and roll-down are quoted together as a breakeven. The breakeven states how wrong the forecast can be before the trade loses money.
Comparing the results still needs one adjustment. A 30 Yr bond will produce more expected return than a 2 Yr note simply because it carries more risk. Dividing by modified duration puts every tenor on the same footing and answers the question that matters, which is which part of the curve pays best for the risk taken.
One trap deserves its own sentence. The expected repricing must be measured against the forward curve, not against today's spot curve. The market has already priced a change, and a trade only makes money by beating what is priced.
Identifying a cheap bond is the easy half. The harder question is why anyone should expect it to stop being cheap.
Sometimes there is a clear mechanism. A newly issued security is usually the most liquid at its tenor and trades rich. As the next auction approaches, that status passes to the new issue and the old one cheapens toward the curve. The cycle is repeatable and the timing is known in advance.
Sometimes the cheapness is a payment for something real. A bond may be cheap because it is difficult to finance, or because it sits outside a widely tracked index, or because it is a small issue that few investors can trade in size. In these cases the discount is a fair price for a genuine disadvantage, and waiting for it to disappear is waiting for nothing.
The distinction matters more than the size of the residual. A trade with a small residual and a known convergence mechanism is better than a large residual with no explanation. A large unexplained residual is more often a sign that the reference curve is wrong than a sign that the market is.
This is also the risk that defines the strategy. Relative value positions have small expected moves, so they are usually held with leverage. That combination produces long stretches of quiet returns punctuated by episodes where funding tightens, positions are unwound at once, and spreads that had never diverged do so together. Sizing the position for that episode, rather than for the quiet period, is the whole discipline.
A directional trade profits when rates move in a particular direction. A relative value trade holds offsetting positions, so it profits when two instruments move toward each other regardless of the overall direction of rates. The second is not safer. It simply carries a different risk.
Yes. The Treasury market has hundreds of securities across the same credit, differing only in maturity, coupon, issue size, and financing. That is enough variation for a fitted curve to identify bonds that are rich or cheap against their neighbors, with no credit judgment involved at all.
Large enough to cover the cost of getting in and out and the cost of financing the position for the expected holding period. A residual of a few basis points on a trade that must be held for a year rarely survives those costs. This is why relative value desks focus on securities that finance well and trade in size.