Continuous futures contracts explained: how rolled price series are constructed and why they matter
Every backtest of a futures strategy depends on a price history that does not actually exist. Individual contracts expire every month or quarter, leaving a series of disconnected segments rather than a continuous record. Stitching those segments into one unbroken series is not a neutral data-engineering step—the method chosen has a direct, measurable effect on apparent returns, and two analysts building the same strategy from the same raw data can reach meaningfully different conclusions if they make different splicing choices.
What continuous futures contracts are
The methodological challenge of constructing continuous price series from rolling contracts—and the return distortions introduced by different splicing methods—was examined by Gorton and Rouwenhorst (2006) in Facts and Fantasies about Commodity Futures, Financial Analysts Journal. Their study used continuous series adjusted for roll returns (similar to the Panama backward-adjustment method) to document commodity futures returns across 36 years; the construction methodology materially affects apparent historical performance and is a prerequisite for any systematic backtesting of futures strategies.
A continuous futures contract is a synthetic price series constructed by chaining successive front-month contracts into a single data stream. No such instrument is directly traded. It is a construction: a data engineering decision that must be made before any analysis of futures performance can begin.
The need arises because futures contracts have finite lives. A crude oil contract expiring in May ceases to exist after its last trading day. The equivalent contract for June is a different instrument, typically priced differently. Treating them as consecutive observations of the same price series requires an explicit decision about how to handle the gap between them.
How continuous series are constructed
Two main construction methods exist, and they produce different results.
Unadjusted (stitched) method. Successive contracts are concatenated without any price adjustment. When the active contract expires, the series continues with the next contract at whatever price it is trading. If the next contract is priced USD 2 higher than the expiring one, the continuous series shows a USD 2 jump at that point. This jump is not a return earned by any investor; it is an artefact of switching instruments. Unadjusted series preserve correct absolute price levels but introduce false return events at every roll.
Backward-adjusted method (Panama method). Roll gaps are eliminated by applying each gap as an offset to all historical prices. When the series rolls from a contract at USD 80 to one at USD 82, a USD 2 adjustment is subtracted from every prior observation. Backtested period returns are correct, but absolute price levels become progressively distorted over time. On sufficiently long crude oil series, backward-adjusted prices can turn negative—which has no economic meaning. The adjustment is a modelling choice, not a reflection of actual prices.
Return-splicing method. Returns are calculated for each contract segment separately and chained rather than chaining price levels. This avoids distorted absolute levels while preserving return accuracy, but produces an index-like series rather than a tradeable price series.
Why the construction method matters
The choice of construction method and roll date are not neutral assumptions—they affect measured strategy performance. In a market with persistent contango, the backward-adjusted series will be progressively deflated relative to spot prices. Two analysts building what appears to be the same strategy on the same underlying can produce materially different backtest results simply because they used different roll dates or different adjustment methods.
A roll executed five days before expiry rather than ten days before will encounter different liquidity conditions and different spreads between contracts. The cumulative effect across years of monthly rolls is non-trivial: on a crude oil strategy with average monthly roll costs of USD 1–2 per barrel, a five-day difference in roll timing compounds to a return difference of several percentage points over a decade of backtesting.
Limitations
No construction method produces a series that is simultaneously correct on absolute prices and correct on returns. The backward-adjusted method is the industry standard for backtesting because return accuracy is what matters for strategy evaluation—but the distorted absolute levels mean the series cannot be used to answer questions about historical price levels.
There is no universal standard for roll timing. Different data providers roll on different days before expiry, use different contract selection rules, and apply different cost assumptions. When comparing backtest results across providers or publications, the construction methodology must always be verified before drawing conclusions.
Continuous futures in pfolio
pfolio's continuous futures chain builder constructs adjusted price series from raw contract data for any futures instrument in the platform. The roll date—the number of days before expiry at which the series transitions to the next contract—and the adjustment method are both user-configurable. Every backtest involving futures in pfolio works from a consistently constructed series whose parameters are explicit and reproducible. The construction parameters are visible in the strategy settings, not buried in a methodology document. See futures roll mechanics for the cost implications of each roll choice.
Related articles
Disclaimer
Get started now
