SHOULD FUTURES CHARTS BE SPREAD-ADJUSTED AT CONTRACT ROLLOVERS?—A DIFFERENCE OF OPINION

Posted August 17, 2017

In his latest Factor Weekly Update report, my friend and veteran trader Peter Brandt answers a reader’s question as to whether futures charts should be adjusted for the price spread between contracts at contract rollover points. This Q&A is reproduced below:

Question:

I was wondering if your futures charts are roll adjusted? And if so, is it necessary to use it like that? Your response is highly appreciated.

–B.K., Istanbul, Turkey

Answer:

I highly detest and object to price-adjusted futures charts. I have never used them and never will. They are highly deceptive. The attempt to get rid of roll gaps results in charts showing historical prices with no resemblance to historical reality.

How’s that for an unequivocal answer? “Highly detest.” “Object to.” “Highly Deceptive.” Not much doubt where Brandt stands on this issue. Well, while I respect Brandt’s opinion and acknowledge his superior experience and skill in applying chart analysis to trading, I do take issue with his reply. This disagreement is a rarity. I remember listening to a podcast interview of Brandt on one of my daily walks and being struck by how remarkably similar Brandt’s replies were to those I would have given to the same questions. This impression was reinforced when we interviewed each other in a three-part audio series (https://www.peterlbrandt.com/market-wizards-jack-schwager/). So the fact that here was a subject we significantly disagreed on was both surprising and notable.

Although, there are indeed situations where spread-adjusted futures charts, also called continuous futures charts, are inappropriate, there are times where they are far preferable to unadjusted linked contracts charts (also called nearest futures charts) and times when both types of charts are useful. To explain why I believe in this more nuanced, case-dependent view on whether spread-adjusted prices should be used, it is useful to begin with an explanation of why linking contracts is essential to the chart analysis of futures.

THE NECESSITY OF LINKED-CONTRACT CHARTS

Many chart analysis patterns and techniques require long-term charts—often charts of multiyear duration. This observation is particularly true for the identification of top and bottom formations, as well as the determination of support and resistance levels. A major problem facing the chart analyst in the futures markets is that most futures contracts have relatively limited life spans and even shorter periods in which these contracts have significant trading activity. For many futures contracts (e.g., currencies, stock indexes) trading activity is almost totally concentrated in the nearest one or two contract months. For example, in Figure 1, there were only about two months of liquid data available for the March 2016 Russell 2000 Index Mini futures contract when it became the most liquid contract in this market as the December 2015 contract expiration approached. This market is not particularly unusual in this respect. In many futures markets, almost all trading is concentrated in the nearest contract, which will have only a few months (or weeks) of liquid trading history when the prior contract approaches expiration.

Figure 1

The limited price data available for many futures contracts—even those that are the most actively traded contracts in their respective markets—makes it virtually impossible to apply most chart analysis techniques to individual contract charts. Even in those markets in which the individual contracts have a year or more of liquid data, part of a thorough chart study would still encompass analyzing multiyear weekly and monthly charts. Thus, the application of chart analysis unavoidably requires linking successive futures contracts into a single chart. In markets with very limited individual contract data, such linked charts will be a necessity in order to perform any meaningful chart analysis. In other markets, linked charts will still be required for analyzing multiyear chart patterns.

METHODS OF CREATING LINKED-CONTRACT CHARTS

Nearest Futures

The most common approach for creating linked-contract charts is typically termed nearest futures. This type of price series is constructed by taking each individual contract series until its expiration and then continuing with the next contract until its expiration, and so on.

Although, at surface glance, this approach appears to be a reasonable method for constructing linked-contract charts, the problem with a nearest futures chart is that there are price gaps between expiring and new contracts—and quite frequently, these gaps can be very substantial. For example, assume the September coffee contract expires at 132.50 cents/lb and the next nearest contract (December) closes at 138.50 cents/lb on the same day. Further assume that on the next day December coffee falls 5 cents/lb to 133.50—a 3.6 percent drop. A nearest futures price series will show the following closing levels on these two successive days: 132.50 cents, 133.50 cents. In other words, the nearest futures contract would show a one-cent (0.75 percent) gain on a day on which longs would actually have experienced a huge loss. This example is by no means artificial. Such distortions—and indeed more extreme ones—are quite common at contract rollovers in nearest futures charts.

The vulnerability of nearest futures charts to distortions at contract rollover points makes it desirable to derive alternative methods of constructing linked-contract price charts. One such approach is detailed in the next section.

Continuous (Spread-Adjusted) Price Series

The spread-adjusted price series known as “continuous futures” is constructed by adding the cumulative difference between the old and new contracts at rollover points to the new contract series. An example should help clarify this method. Assume we are constructing a spread-adjusted continuous price series for gold using the June and December contracts. (The choice of the combination of contracts used to create a continuous futures chart is arbitrary. One can use all or any combination of actively traded months.) If the price series begins at the start of the calendar year, initially the values in the series will be identical to the prices of the June contract expiring in that year. Assume that on the defined rollover date (which need not necessarily be the last trading day) June gold closes at $1,200 and December gold closes at $1,205. In this case, all subsequent prices based on the December contract would be adjusted downward by $5—the difference between the December and June contracts on the rollover date. Assume that at the next rollover date December gold is trading at $1,350 and the subsequent June contract is trading at $1,354. The December contract price of $1,350 implies that the spread-adjusted continuous price is $1,345. Thus, on this second rollover date, the June contract is trading $9 above the adjusted series. Consequently, all subsequent prices based on the second June contract would be adjusted downward by $9. This procedure would continue, with the adjustment for each contract dependent on the cumulative total of the present and prior transition point price differences. The resulting price series would be free of the distortions due to spread differences that exist at the rollover points between contracts.

The construction of a continuous futures series can be thought of as the mathematical equivalent of taking a nearest futures chart, cutting out each individual contract series contained in the chart, and pasting the ends together (assuming a continuous series employing all contracts and using the same rollover dates as the nearest futures chart). Typically, as a last step, it is convenient to shift the scale of the entire series by the cumulative adjustment factor, a step that will set the current price of the series equal to the price of the current contract without changing the shape of the series.

Comparing the Series

It is important to understand that a linked futures price series can only accurately reflect either price levels, as does nearest futures, or price moves, as does continuous futures, but not both—much as a coin can land on either heads or tails, but not both. The adjustment process used to construct continuous series means that past prices in a continuous series will not match the actual historical prices that prevailed at the time. However, a continuous series will accurately reflect the actual price movements of the market and will exactly parallel the equity fluctuations experienced by a trader who is continually long (rolling over positions on the same rollover dates used to construct the continuous series), whereas a nearest futures price series can be extremely misleading in these respects.

NEAREST VERSUS CONTINUOUS FUTURES IN CHART ANALYSIS

Given the significant differences between nearest and continuous futures price series, the obvious question in the reader’s mind is probably: Which series—nearest futures or continuous futures—would be more appropriate for chart analysis? To some extent, this is like asking which factor a consumer should consider before purchasing a new car: price or quality. The obvious answer is both—each factor provides important information about a characteristic that is not measured by the other. In terms of price series, considering nearest futures versus continuous futures, each series provides information that the other doesn’t. Specifically, a nearest futures price series provides accurate information about past price levels, but not price swings, whereas the exact reverse statement applies to a continuous futures series.

Consider, for example, Figure 2. What catastrophic event caused the instantaneous 165- cent (24 percent) collapse in the nearest futures chart for corn from July 12 to July 15, 2013? Answer: absolutely nothing. This “phantom” price move reflected nothing more than a transition from the old crop July contract to the new crop December contract. Figure 3, which depicts the continuous futures price for the same market (and by definition eliminates price gaps at contract rollovers), shows that no such price move existed—corn was actually little changed from July 12 to July 15. Clearly, the susceptibility of nearest futures charts to distortions caused by wide gaps at rollovers can make it difficult to use nearest futures for chart analysis that focuses on price swings.

Figure 2

Figure 3

On the other hand, the continuous futures chart achieves accuracy in depicting price swings at the sacrifice of accuracy in reflecting price levels. In order to accurately show the magnitude of past price swings, historical continuous futures prices can end up being very far removed from the actual historical price levels. In fact, it is not even unusual for historical continuous futures prices to be negative (see Figure 4). Obviously, such “impossible” historical prices can have no relevance as guidelines to prospective support and resistance levels.

Figure 4

Figure 4

The fact that each type of price chart—nearest and continuous—has certain significant intrinsic weaknesses argues for combining both types of charts in a more complete analysis. Often these two types of charts will provide entirely different price pictures. For example, consider the nearest futures chart for lean hogs depicted in Figure 5. Looking at this chart, it would be tempting to conclude that hogs were experiencing a period of severe price dislocation and volatility in 2013, peaking sometime in early July. Now look at Figure 6, which shows the continuous version of the same market. This chart shows that hog prices were in a consistent uptrend that began in April and peaked at the end of October. It is no exaggeration to say that, without the benefit of the chart labels, it would be virtually impossible to recognize that Figures 5 and 6 depict the same market.

Figure 5

Figure 6

CONCLUSION

For any application of technical analysis in which the accurate representation of price moves is essential (e.g., price swing analysis, moving averages and most other technical indicators, etc.), continuous futures, as opposed to nearest futures, is the only viable choice for depicting price series that extend across multiple futures contracts. However, in the case of applications in which the accurate representation of price levels is essential (e.g., support and resistance on longer-term charts where continuous futures price levels can even be negative), nearest futures charts are essential. Finally for some applications, such as determining short-to-intermediate-term support and resistance levels, the choice of a preferred series is not clear-cut. In the latter case, traders need to experiment with both types of charts to determine if they prefer one over the other, or, for that matter, if they find consulting both of these charts the most effective method.

This article has only discussed the selection of appropriate prices series for chart analysis. As a final word, I would note that for the purposes of trading system development and testing, continuous futures are the only correct choice for data. Any use of any other linked series, such as nearest futures or constant forward prices (e.g., 3-month forward FX prices) will lead to highly distorted results. Readers interested in a detailed discussion of this related topic are referred to Chapter 18 of the 2017 edition of A Complete Guide to the Futures Market.

——————————————————–

This is the third in a series of articles that will be released approximately every two weeks that are based on extracts from the the 2nd edition of A Complete Guide to the Futures Market published in January 2017. (The first edition, which was also my first book, was published in 1984.)

Jack Schwager, Chief Research Officer, FundSeeder

Information on FundSeeder: http://fundseeder.com

admin