A continuous futures contract is a constructed price history that links a sequence of individual futures contracts, each covering a different delivery month, into one unbroken series. It is not a tradable instrument. It exists because every real futures contract has a limited life, typically trading actively for only a few weeks or months before it expires, which is too short for long-term charts or statistical study.
Data vendors, charting packages and trading platforms offer continuous series for most futures markets, and analysts build their own. They are used in technical analysis, in research and above all in backtesting trading rules. The way the series is built affects the results, so the method has to suit the purpose.
Why splicing is needed
At any moment several delivery months of the same future trade side by side at different prices. The difference between them reflects interest, storage and expected supply, as described under cost of carry. When later months are dearer than nearer ones the market is in contango, and when they are cheaper it is in backwardation.
If the price history of an expiring contract is simply followed by that of the next one, the series jumps at the join by the amount of this difference. The jump is not a gain or loss that any trader experienced, because a holder who moved from one contract to the next sold the old one and bought the new one at their respective prices. Left in the data, such gaps create false signals in indicators such as a moving average and distort calculated returns.
Roll rules
The first choice is when to switch from one contract to the next, known as the roll. Common rules are to roll a fixed number of days before expiry or before the first notice day of the delivery month, or to roll when volume or open interest in the next contract overtakes the current one. The second rule follows where trading actually is. In markets where only certain months are liquid, the series skips the inactive ones. The dates at which the switches occur are sometimes called ratio adjustment points.
Adjustment methods
- Unadjusted. Contracts are joined with no correction. The series shows the actual price traded on each date but contains a gap at every roll.
- Difference adjustment. The price gap between the new and old contract on the roll date is added to all earlier prices, shifting history up or down to remove the jump. This is also called back-adjustment.
- Ratio adjustment. All earlier prices are multiplied by the ratio of the new contract's price to the old contract's price on the roll date.
- Perpetual series. A weighted average of the two nearest contracts is calculated each day, with weights shifting gradually, to represent a constant time to expiry. No roll date is needed.
For the two adjusted methods the corrections on the roll date are:
- is the price of the contract being rolled into.
- is the price of the expiring contract on the same day.
- is added to every earlier price under difference adjustment, and every earlier price is multiplied by under ratio adjustment.
Suppose that on the roll date the expiring contract closes at 100 and the next contract at 104. Then and . An earlier price of 80 becomes under difference adjustment and under ratio adjustment. The adjustments accumulate with each roll, so prices far back in the series can differ greatly from what was quoted at the time.
Choosing a method
Difference adjustment preserves point changes, so the profit or loss in currency terms of a rule tested on the series matches what holding and rolling real contracts would have produced. Its drawback is that percentage changes are distorted, and after many rolls in a market in contango early prices can be pushed below zero, which makes percentage calculations meaningless.
Ratio adjustment preserves percentage returns and keeps prices positive, which suits return-based statistics and comparisons across markets. It distorts point values, however, and cannot be used where prices themselves have been zero or negative, as happened briefly in crude oil in 2020. Unadjusted data remain necessary for anything that depends on true price levels, such as a round-number support level or contract limits.
Limitations
A continuous series is a model of history and embeds assumptions. The simulated roll occurs at a single price with no cost, while a real roll involves two trades, a spread and possibly slippage. Different vendors use different roll dates and methods, so two continuous series for the same market can disagree, and a strategy may look profitable on one and not on the other.
The series also behaves differently from the spot price of the underlying. An adjusted series reflects the return from holding futures, including the gain or loss from rolling, which is the same effect that separates the return of a commodity index from spot price changes. Charts of adjusted futures and of the physical commodity can therefore diverge widely over time.