Rolling the Hedge Forward

Every hedge so far assumed a futures contract that lasts exactly as long as the exposure. Real exposures are often much longer than any single futures contract's life: an airline wants to hedge jet-fuel costs three years out, but liquid oil futures rarely trade with more than 12–18 months to expiry; an oil producer wants to lock in prices on production it hasn't pumped yet, five years from now. No single contract reaches that far — or if one nominally does, it trades so thinly that using it would be like accepting a much worse basis just to get the maturity right. The standard solution is stack and roll: hedge with a short-dated, liquid contract, and just before it expires, close it out and open an equal-sized position in the next available near-month contract — repeating the cycle until the real exposure finally arrives.

How a stack-and-roll hedge is built

Say a firm needs to hedge an exposure that unwinds in three years, but the deepest liquid futures market only goes out 12 months. The rolling program looks like this:

Today: buy (or sell, per the exposure direction) enough near-month contracts to cover the entire three-year notional — the whole "stack" is put on at once, sized for the full exposure, not just the first year.
Just before the near contract expires (roughly 11 months later): close out that position and simultaneously open a new position of the same total size in the next available near-month contract. This is the roll.
Repeat every ~11 months until the real exposure is realized — roughly two more rolls for a three-year hedge built from 12-month contracts.

Each individual leg is a short-horizon hedge, exactly as before — while it's on, its P&L offsets the spot price move over that stretch, subject to the ordinary basis risk of an imperfect asset or timing match. Chained together, the sequence of legs approximates a single long-dated hedge that doesn't actually exist in the market.

Worked example: rolling a refiner's crude hedge

A refiner locks in the cost of Q = 900{,}000 barrels of crude it will need over the next three years, using a 12-month futures contract of 1,000 barrels each.

Leg	Action	Contracts (size Q/1{,}000)	Roll date
1	Open near-month future, full notional	900	~month 11
2	Close leg 1; open next near-month, full notional	900	~month 22
3	Close leg 2; open next near-month, full notional	900	~month 33 (expiry ≈ physical need)

At each roll date, the firm books whatever gain or loss the expiring leg has accumulated — that's ordinary hedge P&L, offsetting the spot price move over that period, subject to that period's basis. Over three rolls, the accumulated P&L approximates Q \times (S_1 - F_{0,1\text{yr}}): roughly the same insurance a true three-year contract would have provided, built by splicing three one-year contracts together.

Seeing it: every roll is a new dice throw on the basis

A single long-dated hedge held to expiry has one basis, resolved once, at the end. A rolled hedge instead realizes a new basis at every roll date — the near contract's basis just before it expires. If those roll-date bases are favorable, the rolled hedge can track the ideal long-dated hedge closely. If they're unfavorable — say, oil markets are in contango (near contracts persistently cheap relative to spot) exactly when each roll happens — the rolled hedge quietly bleeds value at every single roll, a cost with no analogue in the single-contract case. The chart adds one basis "jump" per roll date; drag the slider to see how a hedge that looked fine on paper can drift away from the ideal outcome as rollover costs compound.

In the early 1990s, the US subsidiary of German industrial conglomerate Metallgesellschaft sold customers long-term, fixed-price contracts to deliver oil products for up to ten years — a giant short exposure with no matching long-dated futures market to hedge it. Their solution was a textbook stack-and-roll: buy a huge stack of near-month NYMEX futures, sized for the entire multi-year commitment, and roll it forward every month.

Then oil prices fell sharply and the futures market moved into steep contango. Two things went wrong at once. First, each roll realized a loss on the expiring stack (selling low into a cheap near contract), bleeding cash. Second — and this is the part that actually broke the firm — futures are marked to market daily: every price drop triggered immediate margin calls on the futures leg, in real cash, right away. The offsetting gain on the long-term supply contracts was real too, but it wouldn't show up as cash for years, as deliveries were made. Metallgesellschaft's US arm faced over a billion dollars in margin calls it could not fund from a hedge that was, in an economic sense, working roughly as intended. The parent company stepped in with an emergency bailout and unwound the positions at a huge loss — a hedge that was directionally sound but destroyed by a timing mismatch between when losses hit as cash and when the offsetting gains would have arrived.

Two mistakes recur whenever a hedge program spans more than one contract's life:

"The hedge ratio was right, so nothing can go wrong." Sizing (N^{*}, or a beta calculation) answers "how many contracts," not "will the roll-date basis be favorable." Even a perfectly sized rolled hedge can lose real money if the market is persistently in contango (or backwardation, for a short hedger) at every roll — a cost stacking a chain of correctly-sized hedges does nothing to prevent.
Ignoring the cash-flow mismatch. As Metallgesellschaft discovered, futures losses are paid in cash immediately via daily margin, while the offsetting gain on the underlying physical exposure often shows up only much later (or not as cash at all — an unrealized reduction in a future cost). A hedger who is economically fully protected can still run out of liquidity funding margin calls before the offsetting benefit ever arrives. Sizing a rolling hedge program means budgeting for this cash-timing risk, not just getting the number of contracts arithmetically correct.