Continuous data: Not so easy

July 15, 2008 11:44 AM

Futures contracts have finite lives. They start trading. They trade. They expire. This creates a natural hurdle for building long-term price charts. Analysis cannot extend backward through time beyond the start of trading in the contract.

While with certain contracts, like New York Mercantile Exchange energy contracts, one contract goes off the board and the next month becomes front month; others, like the CME Group contracts, the front month goes to spot and continues to trade for several weeks while the next option becomes the front month. The exchange announces the official roll date based on open interest.

For years, though, data vendors have developed a number of techniques for creating much longer series of price data for finite-lived futures contracts. In short, they string subsequent price series together.

However, this solution has its own problems. Often, when one contract goes off the board and another becomes the front-month, or active, contract, there is a significant price gap. The contracts may trade in tandem, more or less, on either side of that gap, but there is a gap nonetheless.

Here are ways that various vendors close the gap:

Back-adjusted. The gap between two contracts is measured. That value is then added or subtracted to all values in the prior contract to bring it in line with the most recent contract.

Forward-adjusted. Similar to back-adjusted, except the mathematical modifications are made going forward, not backward. Most traders prefer back-adjusted contracts, however, because they keep the current prices intact for more accurate recent references.

Perpetual. One problem with back-adjusted and forward-adjusted contracts is when you make the roll. It’s not always clear cut when one contract ceases to be the predominant contract on the board and when another takes over. This method attempts to alleviate that concern by displaying some type of average (usually a weighted average, the weights of which change as contract importance gradually shifts). The problem here is the prices are, in effect, fantasy. The defense is they are intended to be an analysis tool, not a tradable vehicle.

Gann. These charts recognize the seasonality of futures contracts. That is, November new-crop soybeans are affected by different economics than old-crop contracts. These charts avoid the problem of splicing together non-similar contract months by only linking the same months of contracts.

Percentage-adjusted. The gap between contracts is calculated as a percentage of the absolute level of the last contract’s value. That percentage correction is applied to all price points composing the prior contract. Percentage-adjusted data series are designed for percentage-based analysis methods, including software that has the ability to report and translate percentage-based performance reports into absolute numbers.

Which type of continuous data makes sense for your specific application depends on your specific application. Some are clear, however. If your strategy depends on the absolute level of past prices, then it would be folly to base your analysis on back-adjusted contracts, the prices of which that are listed prior to a contract splice may never, in fact, have traded at all.

About the Author