A continuous futures chart joins data from contracts with different expiries into one historical series. It is not a single, unchanging contract that could have been traded throughout the displayed period. With historical price adjustment, or back-adjustment, enabled, previously visible levels may have been recalculated. Before using them for an order or strategy test, check the specific contract, the roll rule and the adjustment mode. A smooth price curve does not itself prove that trades could have been executed at those prices.
The practical task is narrower than evaluating an entire strategy: establish what to compare when the historical chart and the tradable contract show different prices. Start with one contract-roll boundary and a saved description of the settings. Until the price difference is explained, do not adjust entry rules simply to make the test look more convincing.
What a continuous futures chart joins together
An individual futures contract has a defined expiry. The continuous-series provider chooses which contract's data to display in each part of the history and when to switch to the next. The short symbol name does not explain that methodology. A series may use the nearest expiry, the most active contract or another documented rule.
For example, CME's description of continuous price series distinguishes Active and Front contract views and specifies the use of settlement prices. That is the methodology of a particular data product. Do not automatically attribute it to another intraday chart merely because its name also contains “continuous”.
Comparison requires the full symbol identity. Record the exchange, data source and contract expiry rather than just the underlying instrument's name. If the platform lets you open a continuous symbol and an individual contract, keep them side by side. A difference between those views is the starting point for investigation, not automatic proof that one source is wrong.
Also distinguish a chart switching contracts from rolling a position. CME's explanation of rollover describes closing an existing position and opening a new one in another contract. The chart's automatic switch is not, by itself, such a trade in your account. Check what the broker actually does and which contract appears in the order window.
How back-adjustment changes a historical level
When contracts are joined, their price difference may appear as a jump at the series boundary. This is not necessarily a market movement within one contract. Adjustment aims to handle that joining difference, but the method chosen determines how the earlier chart changes.
TradingView's documentation states that its back-adjustment adjusts previous contracts by the difference between the new and old contracts' daily-bar Close prices at the switching boundary. This is usually the daily bar before the switch. Check the boundary used rather than trying to explain the adjustment with an arbitrarily chosen intraday price.
Consider a simplified arithmetic example: the old contract's comparison price is 100, while the new contract's price in the same comparison is 105. Adding the difference of 5 to the earlier part of the series may make the transition look smoother. It does not mean that the old contract could then have been bought at the adjusted price of 105. These figures are demonstration data, not the history of a particular instrument.
The example separates a calculation result from an actual quotation. If you marked a historical line before a subsequent adjustment, its visible price context may later differ. A saved screenshot without the symbol and settings is then incomplete evidence. Do not judge signal quality until you know whether you are looking at the same data construction.
Which price to use for an actual contract
When preparing an order, check the identity and current data of the contract selected for execution. A level in the continuous history may be an analytical reference, but it cannot be transferred unchecked to a contract with another expiry. If a strategy uses such a mapping, its conversion logic must be defined in advance and verifiable.
This does not mean an individual-contract chart guarantees execution at the displayed price. Chart prices and actual fills are also affected by the data and execution conditions. This check only removes an additional error: using the wrong contract or a recalculated historical level. It does not replace checking the order type and risk.
If a level comes from longer-term analysis, record its origin. Is it an individual contract's historical high, a mark on the continuous series or an indicator calculation? Different origins may need different validation. Calling all of them “support” is not enough for another person to reproduce the calculation.
Do not calculate risk from an unclear price distance. If the stop reference comes from another series, resolve that mismatch before using position-sizing and volatility principles. A wider stop or smaller position does not make an incorrect data mapping valid.
Comparing two charts around one rollover
Choose a documented contract transition and open the continuous series alongside the relevant individual contract. First align the time zone, bar interval and session coverage. If one view shows a different part of the trading day, the price comparison may be misleading even without adjustment.
Near the transition, record old and new contract prices using the same price field. Do not mix the last traded price with a settlement price or another session's close. If the methodology uses a particular daily bar, locate that exact bar. Record the observation time too, so it is clear later which data version you used.
Then inspect the continuous series with and without adjustment, if the platform offers that option. Compare one earlier price point and the transition itself. The aim is to understand which value changes and why, not to choose the most attractive curve. If the feature is unavailable for that symbol, record this rather than assuming adjustment is disabled.
Return to the individual contract afterwards. Check whether the difference matches the documented recalculation. If it does not, avoid immediately changing the strategy. First rule out mismatches in contract, price field, time zone and session. If the methodology remains unclear, state this as an unresolved data limitation in the test conclusion.
|
Observation |
What to check first |
What cannot yet be concluded |
|
Prices differ only in the earlier part of the series |
Whether historical adjustment was applied |
That the old contract traded at the adjusted price |
|
A jump coincides with the contract change |
Comparable prices of the old and new contracts |
That the jump was an executable profit opportunity |
|
One day's closing values differ |
Price field, session and time zone |
That the data provider definitely made an error |
|
An old indicator signal changes |
Input series, adjustment and calculation settings |
That the signal was visible in exactly that form at the time |
Save the comparison table with the images rather than keeping screenshots alone. A contract and mode recorded in a table help separate observation from interpretation. For example, “the level moved” is not a cause. “The adjustment mode changed; all other comparison settings were retained” already makes the check repeatable.
In the record, distinguish what you observed on screen from what you checked in the documentation. The screen shows a price; the methodology source explains its origin. If the two cannot be linked, the result is not yet a verified price transformation. Agreement at one point, for example, does not explain how the entire earlier history was treated. Mark that boundary so a narrow check is not later presented as broader evidence of quality.
It is useful to finish the comparison with a short reproduction check: close the additional view and reopen it from the saved settings. Can you locate the same contract and price point? If so, the record is usable. If not, add the missing identifying information. This check requires neither a new trade nor strategy retuning.
If the platform does not disclose a necessary field, do not guess it. Record exactly where you looked and what still needs to be clarified with the data provider. Until there is an answer, do not treat that field as a reliable test assumption. “Unknown” is important audit information, not a cosmetic gap to fill. It tells another person why the comparison remains unfinished.
Even a successful check needs a narrow conclusion: at the selected transition, the price difference is explained under the saved settings. That does not establish that all other contract rolls behave identically or that trade simulation is correct. A precise conclusion supports further testing in stages without losing the link between each claim and its check.
Why a test result can change
A test calculates signals from its input data. When historical prices are transformed, rule outcomes may also change, particularly when they use absolute price levels or cross a series-joining boundary. This does not imply that all indicators react in the same way. Check the specific calculation and period.
Separate two practical questions: does the chosen analytical series produce the signal, and could the trade be modelled in the individual contract? A test using continuous history needs an explicit way to handle contract rolls. Otherwise, one result may combine a market movement with the effect of switching the data series.
In the simplified example, the price difference of 5 must not automatically be counted as profit on one held contract. It describes a comparison between two contracts. Assessing a position roll requires separate execution logic and assumptions about its costs. This article does not calculate the performance of a particular strategy.
When rechecking, change only one setting at first. If you switch the session, adjustment and contract selection simultaneously, the result difference cannot reliably be attributed to back-adjustment. Keep the original test separately and give the new version a clear name. Overwriting the old result destroys the very comparison needed to explain the problem.
After checking the data construction, return to the backtest and forward-test process. A better historical result in one adjusted series is not evidence of future profit, however. It is a result based on specific data and assumptions, which need to remain visible.
What to record in the test journal
To make the check repeatable, save the full symbol name, exchange and data provider. Record whether you used an individual contract or a continuous series. If the series switches contracts, include the available roll rule and methodology source. Mark an unclear field as unknown rather than filling it in from memory.
Record the price-adjustment mode, bar interval, time zone and session separately. Add the test period and the date the data was obtained. If the data version can be retained, do so in accordance with the service's terms of use. This helps explain a later difference, but does not itself establish that the data is correct.
The journal also needs a conclusion. Is adjustment enough to explain the difference, or does a mismatch remain? Which specific price point shows it? “Everything is fine” is too broad if only one transition was checked. Record the boundary tested and what the exercise does not cover.
A short checklist is useful:
- The symbol and execution contract are identified separately.
- The data provider and series-joining principle are known.
- The adjustment mode is recorded.
- Matching times, sessions and price fields have been compared.
- The selected transition table and evidence of the settings are retained.
- Unexplained differences are not hidden in the final test result.
If an old check can no longer be reproduced
Start with the saved data description rather than retuning strategy parameters. Check that the same symbol is open and that the adjustment mode has not changed. Then compare the session and price field. Only then examine the signal calculation. This sequence helps distinguish a data change from an error in the rule itself.
If the original settings were not retained, the honest result may be “the previous check cannot be reproduced exactly”. That is more useful than finding a configuration that displays the desired profit curve again. Give the new test a clear data description and do not call it confirmation of the old version.
The conclusion should make sense to another trader too. From the journal, they should understand which contract to open, which period to compare and what difference was found. If that requires the author's verbal explanation, the evidence set is incomplete. Add the missing setting rather than a general expression of confidence in the platform's reliability.
The practical next step is to check your symbol's contract and adjustment settings and record them in the test journal. Choose one rollover boundary and explain its price difference. This does not demonstrate a strategy edge, but helps avoid decisions based on mixed-up price series.
This material is educational, not individual investment advice. Futures trading involves a risk of loss; checking the data does not guarantee a trade's outcome.