Futures rollover dates: why your chart disagrees with CME

You opened your NQ chart this morning and the level you drew last week is sitting at a price nobody ever traded. Nothing got deleted. You did not drag the drawing. The chart rolled to a new contract, and every bar behind the switch was arit

You opened your NQ chart this morning and the level you drew last week is sitting at a price nobody ever traded.

Nothing got deleted. You did not drag the drawing. The chart rolled to a new contract, and every bar behind the switch was arithmetically shifted to make the join look smooth.

Futures rollover is the point where a continuous chart stops showing one contract and starts showing the next. Most guides to it publish a calendar and stop, which would be fine if there were one date to publish.

There are three. They belong to three different systems, they land on three different days, and the one on your screen is not the one the exchange publishes. The roll calendar most sites print is not CME's roll calendar, your chart switches on neither of them, and the price history behind that switch has been rewritten.

CME's roll date is the Monday before the third Friday

Start with the exchange, because this is the one with an authoritative answer and it gets quoted wrong constantly.

CME Group's equity index roll dates page puts it in one sentence: "Equity products roll date is the Monday prior to the third Friday of the expiration month."

Monday. Not Thursday. Search the phrase and you will find the Thursday version repeated across a stack of trading blogs, and it does not match the exchange that lists the contract.

CME's own table for 2026 gives the US index contracts a 14 September roll against an 18 September expiration, then a 14 December roll against an 18 December expiration. Nikkei 225 and TOPIX are the footnoted exception, rolling the Monday prior to the second Friday.

What that date actually does is narrow. CME states the roll date "determines what contract month is listed for trading during the CME Globex session," and that afterwards "it is customary to identify the second nearest expiration month as the 'lead month'" because the expiring contract "will terminate soon and will have a less liquid market."

So it is a listing and convention date. It is not a deadline, it does not touch your position, and it is not when your chart changes.

Your chart switches on a rule fitted to historical averages

The symbol you are probably looking at is NQ1! or ES1!, and it is not a contract at all.

TradingView is blunt about what those are in its continuous contracts documentation: "Continuous contracts are synthetic instruments, they may not accurately represent the actual traded prices." The 1! series tracks the front month, 2! tracks the second nearest, and only 1! is tradeable, on CME and EUREX. The 2! series is analysis-only.

The switch date is decided by a separate mechanism. TradingView documents the main condition as the next contract's daily volume exceeding the current contract's, which sounds like a live measurement.

It is not one. The platform studies when volume historically migrates, converts that into a fixed rule, and then applies the fixed rule regardless of what this quarter's volume is doing. The support page states the consequence plainly: "Since the rule is based on average values, there may be situations when the continuous future has already switched, and the volume on the previous contract is higher than on the new one, and vice versa."

Read that again if you trade from a 1! chart. Your chart can be quoting the back month while the liquidity is still in the front month, or the reverse, and nothing on screen tells you which.

Back-adjustment rewrites every bar behind the switch

Two contract months do not trade at the same price. Different expiration, different instrument, different quote. So on switch day the stitched series has a hole in it.

The fix is back-adjustment, which TradingView describes as "the ability to adjust the data of previous contracts in a continuous in order to remove the roll gap". The coefficient is "the difference between the Close of the new and old contracts for the nearest daily bar to the switching point in the continuous."

In plain English: the platform measures the gap, then shifts every older bar by that amount until the line joins up.

Your chart is now continuous. It is also displaying historical prices that are off by the running total of every roll adjustment behind you. Not off by a tick. Off by the accumulated sum of many quarterly gaps, which is exactly why a multi-year back-adjusted index chart can show levels the index never printed.

Switch back-adjustment off with the B-ADJ toggle and every price becomes real again, at the cost of a visible step at each roll that your trend lines now cross.

Neither setting is correct. They break in opposite directions, and the job is picking which breakage you would rather manage.

Every level you drew before the roll now points somewhere else

This is the part that costs money rather than just being trivia.

Support, resistance, anchored VWAP, profile nodes, whatever you mark up: those objects are pinned to price coordinates. Back-adjustment moves the price coordinates of the history underneath them and the drawing does not follow, so a line that sat on three clean touches last Friday is now floating in the middle of nothing.

The size of the displacement is quoted in index points, and index points are money. A Micro E-mini Nasdaq-100 contract is $2 times the index with a minimum tick of 0.25 index points, so a thirty-point roll gap moves your entire chart history by sixty dollars per contract of displacement that was never profit or loss for anyone.

Volume profile work takes the worst of it, because the profile aggregates traded volume by price across a lookback window, and the roll changes what a price on that axis means partway through the window. The high-volume node you have been treating as structure is partly built from a contract that no longer exists at that price.

Backtests inherit the same defect. Running a strategy over a back-adjusted continuous series means the engine is filling orders at prices that were never quoted, on top of the default assumptions that already flatter almost every result in the strategy tester.

A back-adjusted chart is a very convincing drawing of a market that does not exist. It is useful for trend context and worthless as a record of what anything cost.

— Internal note on futures data handling, Tradoki desk

Nothing rolls the position for you

The exchange lists a new lead month, your chart quietly changes symbol, and neither of those events moves your contracts anywhere.

Topstep's explainer on rollover frames it as a decision the trader makes: when contracts near expiration, traders with open positions "have to choose whether to close out their positions entirely or roll them into a new contract month." The firm also notes that markets behave differently during roll week, because institutional flow is busy repositioning rather than doing whatever it normally does.

There are two ways to execute it. A calendar spread sends both legs as one order at a quoted difference. Legging means closing the front month, then opening the back month as a separate trade, and wearing whatever the market does in the seconds between them. Topstep names legging as the riskier of the two, which matches the mechanics.

On a funded account this stops being an academic distinction. The round trip carries commission and spread like any other trade, and a trailing drawdown floor does not distinguish between money lost to a bad read and money lost to an operational roll. Both push the same number in the same direction.

Check the rule page rather than assuming, too. Prop firm rulebooks differ on nearly everything measurable, and expiring-contract handling is exactly the kind of clause that sits three clicks past the pricing table.

Roll week is an operations problem with a fixed checklist

None of this needs a view on the market. It needs a routine that runs four times a year.

Know both dates before the week starts. The CME roll date tells you when the lead month convention changes; the third-Friday expiration tells you when the contract you are holding stops existing. They are four days apart and they mean different things.

Check the volume yourself rather than trusting the switch. Pull up both contract months and look at which one is actually trading, because your platform's switch is a statistical guess that its own documentation admits can be early or late.

Pick a back-adjustment posture and keep it. Consistency matters more than the choice: flipping B-ADJ mid-quarter means your levels, your journal screenshots and your backtests are measured against three different price series.

Redraw levels on the contract you are trading, not on the synthetic series. Then log the roll in the journal with the exit and re-entry prices, because four rolls a year of commission and slippage is a real line item, and it is the kind of cost that only becomes visible once it is written down.

The uncomfortable summary is that the smoothest-looking chart on your screen is the one that has been edited most. That is a reasonable trade for long-term trend context and a bad basis for precision work, and the four days between CME's roll date and the third Friday are when the difference between those two uses stops being theoretical.

● FAQ

What is the actual futures rollover date for ES and NQ?

CME Group publishes it as the Monday prior to the third Friday of the expiration month, which is not the Thursday date repeated across most trading blogs. For 2026 that gives a 14 September roll against an 18 September expiration, then 14 December against 18 December. Nikkei 225 and TOPIX are the documented exception and roll on the Monday prior to the second Friday instead.

Does the CME roll date force me out of my position?

No. CME describes the roll date as what determines the contract month listed for trading during the Globex session, and as the point after which the second nearest expiration is customarily treated as the lead month. It is a listing and liquidity convention. Your position is your responsibility right up to the contract's own expiration.

When does TradingView switch NQ1! to the next contract?

Not on the CME roll date. TradingView documents the main condition as the next contract's daily volume exceeding the current contract's, but it applies that as a fixed rule derived from statistical averages rather than checking live volume. Its own support page warns the continuous series can switch while the old contract still carries more volume, and the reverse.

Should I turn back-adjustment on or off for futures charts?

Both settings are wrong in opposite directions and you pick the failure you prefer. Back-adjustment shifts every historical bar by the roll gap so the series joins cleanly, which means older prices on the chart are not prices that traded. Leaving it off keeps every price real and leaves a visible step at each roll that your trend lines cross.

Do prop firms roll futures positions for me?

Treat the answer as no unless your firm's rule page says otherwise in writing. Topstep frames the roll as a decision the trader makes when contracts near expiration, either closing out or moving into the next contract month. On a funded or evaluation account the round-trip cost of that roll lands on your equity curve like any other trade.