What Data Is Needed to Construct a Chart?

To construct reliable charts, the technical analyst must be sure of the trustworthiness of the data. During a normal trading day, many errors appear on the tape, from which most data originates, that someone must screen out and adjust. Some trade reports show the wrong price or volume; some trades are in error and must be broken; and some trades occur out of order. When a price error occurs at a high or a low for the trading day, the error is especially troublesome because it affects daily calculations of averages and oscillators using highs and lows. It is, therefore, important that any data used for charting is extremely “clean” and reliable.

In addition to trading errors, other data errors can occur through stock splits, dividends, offerings, and distributions. In the commodities markets, because contracts have a settlement date at which trading halts, incorrect calculation of the time and price linkage between contracts may affect longer-term technical patterns and trends. Calculation of these linkages into time series data for a longer-term perspective is never precise. The results are called either “nearest future,” “perpetual series,” or “continuous series” and are provided by many data vendors and exchanges (see Box 11.3). There are other methods of joining the different contracts into a continuous series, but all have serious problems for the analyst wanting to test longer periods of data. Schwager (1996) recommends the continuous contract as the best method for such tests.

Box 11.3 Linked Contracts

For short-term testing and analysis of futures contracts, you can use the actual contract data.

The test period must be extremely short, however, because the liquid portion of any contract, the period when active trading occurs, is only a portion of the contract time span. Thus, the data available for realistic study is short, and to test a number of signals, it must be separated at short intervals.

Once the trading signal horizon exceeds an hour, however, one must calculate a linked contract to provide sufficient data for testing. There are three basic types of linked contracts: the nearest future, the perpetual, and the continuous. Each has its advantages and disadvantages.

The nearest future method is just a plot of each futures contract as it expires and is replaced by a new contract. Unfortunately, during the transition between old and new contract, a gap always exists at the rollover into the new contract. Thus, this method is useless for testing trading systems, even though the contracts plotted are historically correct. It is the least preferable method of analyzing longer-term futures price moves.

The perpetual contract (also known as the “constant-forward”), to avoid the nearest future rollover problem, uses a constant forward price—namely, the anticipated price some specific period ahead. This forward contract comes from the interest and foreign exchange markets where constant-forward contracts are traded. The adjustment to the futures markets assumes a price that adjusts for the time between a contract series that is current and one that is beyond the specific constant-forward period. The perpetual price then is the price of the current contract and the forward contract each weighted by the amount of time remaining in the hypothetical forward period. Thus, the perpetual price avoids the problem of the rollover by smoothing the two contract prices—one near and one far out—in a gradual manner as time progresses. The problem with this method is that it is not real. The actual prices recorded in the perpetual contract never occurred. It is, thus, not a suitable method for testing a technical trading system.

The third method, the continuous contract, is more realistic but is useless for calculating percentage changes over time. It adjusts for the premium difference between the current contract prices—the price of the previous contract and the contract into which the trader rolls his position at a specific rollover date, say 15 days before expiration—to avoid the trading bias that occurs as the contract nears its end. This continuous contract then carries the adjustment into the future. It reflects exactly what would have occurred to a portfolio that invested in the first contract and rolled over each contract at its rollover date. It is, thus, a realistic expression of the history of the futures contract and can be used for testing past data. It has two problems, however. One is that because the adjustments are additive, the continuous contract cannot be used for percentage returns; and two, the ending price of the continuous contract is not the same as the current price of the current contract. The continuous contract is the primary contract used for testing trading systems.

Source: Kirkpatrick II Charles D., Dahlquist Julie R. (2015), Technical Analysis: The Complete Resource for Financial Market Technicians, FT Press; 3rd edition.