Bond Future Rolls and Calendar Spreads

Treasury futures contracts don't run forever — each one expires and settles in a specific delivery month, with a fresh contract listed for the next quarter. A fund that wants continuous exposure, say a pension fund duration-hedging with futures indefinitely, has to sell the expiring contract and buy the next one before expiry. That combined trade, done as a single order, is the roll, and the price difference between the two contracts is the calendar spread.

What drives the spread

If financing the CTD bond in repo is cheap relative to the bond's coupon (positive carry), holding the near contract's underlying bond is attractive, which tends to make the near contract trade rich relative to the far one — the roll (near price minus far price) trades wide. If repo is expensive relative to coupon (negative carry, common when short rates are high relative to bond yields), the near contract cheapens relative to the far one, and the roll trades tighter or even inverts. On top of carry, the relative value of the delivery option embedded in each contract, which depends on how much time remains and how volatile the deliverable basket's relative pricing is, also affects the spread.

$$\text{Roll} = \text{Near contract price} - \text{Far contract price}$$

In words: the roll is simply the price gap you pay (or receive) to swap the near-month exposure for the far-month exposure, and that gap should, in a no-arbitrage world, roughly equal the net cost of carrying the CTD bond over the period between the two expiries.

Worked example

A fund holds a long position of 1,000 contracts in the March 10-year Treasury futures and needs continuous exposure into June. The March contract trades at 118-16, the June contract at 118-04, so the roll (March minus June) is +12/32, or +0.375 points. Estimating the net carry on the CTD bond, coupon income minus repo financing cost over the roughly three months between expiries, the fund calculates a fair roll of about +10/32 given current repo rates. Since the market roll of +12/32 is 2/32 rich relative to fair value, the fund executes the roll as a single "roll trade" (sell March, buy June simultaneously) rather than legging into each side separately, both to save on execution risk and because the current pricing modestly favors selling the expensive near-month exposure.

Worked example: rolls under repo stress

Around quarter-end, repo rates for on-the-run Treasuries occasionally spike due to balance-sheet constraints at dealers (regulatory reporting dates make banks reluctant to hold inventory). If the CTD bond goes "special" in repo (very low or negative repo rates because everyone wants to borrow that specific bond), the net cost of carry on it turns sharply negative, and the roll can compress or invert versus its typical seasonal pattern. Funds that roll on autopilot without checking repo conditions can end up paying up meaningfully more (or receiving meaningfully less) than the roll's recent historical average, simply because they didn't account for a temporary repo squeeze.

What this means in practice

Rolls happen every quarter for every actively used Treasury futures contract, and desks that manage large futures books watch the roll's richness or cheapness relative to fair carry as a small but recurring source of P&L, distinct from the underlying duration view. Because rolls are large, concentrated trading events (most open interest migrates from near to far contract within a narrow multi-day window before expiry), the roll period itself can also see temporary price pressure unrelated to fundamentals.