Managing The Futures Roll

A futures contract has an expiration date, but most desks want continuous exposure that doesn't stop at expiry. To keep the position alive, the desk has to "roll" it: sell the expiring contract and buy the equivalent size in the next contract month, ideally close together in time so the exposure doesn't gap. This sounds like a bookkeeping chore, but it's a real trade with real execution risk, because everyone holding that contract is rolling around the same window of days, and liquidity in the outgoing month thins out fast as expiry approaches.

The core decision is when to roll. Roll too early and you give up basis or carry the older contract would have earned; roll too late and you're competing for a shrinking pool of liquidity in a contract most other participants have already left, often leaving you with worse fills and, in physically-settled contracts, a real risk of ending up in a delivery process you never intended to be part of. Desks typically pick a roll window — a handful of days before expiry, often standardized across an asset class — and either roll all at once or work the position out gradually across that window using a calendar spread order, which trades the price difference between the two contract months directly rather than legging into each side separately.

Worked example

A desk holds 1,000 long contracts in a commodity future expiring in five trading days. The front month's open interest has started declining as other holders roll out, while volume in the next-month contract is climbing. The desk works a calendar spread order — sell 200 front-month, buy 200 next-month, in five clips over the roll window — rather than dumping all 1,000 contracts on one day. The calendar spread trades at a small negative price (the next month is priced slightly above the front month, called contango), so each clip of the roll costs a few ticks versus rolling instantaneously at the mid, but spreading it out avoids being the largest seller in a thinning front-month book on any single day.

Two things commonly go wrong: rolling without checking the calendar spread market first, which can mean paying a much wider effective cost than trading the legs separately would suggest, and forgetting that different exchanges and products have different final trading days and delivery notice periods, so a roll calendar built for one product can't be copy-pasted onto another without checking its specific rules.