Skip to content
Trading Technology

The 23-Hour Trading Day Starts On 6 December: What Near-Round-The-Clock US Equities Actually Break

On 6 December 2026, Nasdaq opens an overnight session from 9pm to 4am ET, taking US equities to nearly 23 hours a day, five days a week. The SEC approved it in April; NYSE Arca has preliminary approval for a 22-hour day and Cboe is working toward 24x5 on EDGX. The rationale is straightforward - global demand for US stocks and competition from crypto venues that never close. The consequences are not. Almost every piece of trading infrastructure ever built assumes a daily close: end-of-day marks, overnight batch runs, risk recalculation windows, corporate action processing, and the humble question of what date a trade belongs to. Here is what actually breaks, and what desks and platform teams should be doing about it now.

AlchmAI Editorial13 min read

6 Dec

Nasdaq's overnight session launches, running 9pm to 4am ET, Sunday evening through Friday

23 hours

Daily US equities trading once the day session (4am-8pm ET) and night session combine

10 Apr 2026

Date the SEC approved Nasdaq's proposal to extend trading for NMS stocks and ETPs from 16 to 23 hours

22x5

NYSE Arca's preliminary-approved trading day, with Cboe continuing toward 24x5 on EDGX

The most consequential change to US equity market structure in a decade will arrive with almost no fanfare, because it is scheduled for a Sunday evening in December. From 6 December 2026, Nasdaq will run an overnight session from 9pm to 4am ET, which combined with its existing 4am to 8pm day session takes US equities to nearly 23 hours a day, five days a week. The SEC approved the proposal on 10 April 2026. NYSE has preliminary approval for a 22-hour day on Arca, subject to SIP and DTCC readiness, and Cboe continues toward 24x5 on EDGX pending approvals.

The stated rationale is uncontroversial and almost certainly correct: demand for US equities from investors in Asian and European time zones has grown to the point where making them wait is leaving money on the table, and crypto venues have spent several years demonstrating that a market which never closes is a viable product. What has had far less attention is the engineering consequence, which is substantial. We build trading platforms, charting systems and market data infrastructure, and our honest assessment is that a great deal of the industry's software has a daily close baked into assumptions that nobody wrote down, and those assumptions are about to become visible in an unhelpful way.

What Actually Breaks

Based on what we are already seeing in client environments preparing for this, the failure modes cluster into five areas. None is intractable; all are considerably more work than the trading desk expects.

  1. 01The closing price stops being obvious. An enormous amount of finance is defined by reference to a closing price: fund NAVs, index levels, margin calculations, benchmark comparisons, client statements, structured product fixings. If a stock trades at 2am, what is 'the close'? The answer is that the official close remains the 4pm primary-market close, and overnight prints are something else - but every system in your estate has to agree on that, and any that naively takes 'last trade' will quietly produce a different number.
  2. 02The trade date becomes ambiguous. A trade executed at 11pm ET on Tuesday is, in most conventions, a Wednesday trade. Systems that derive trade date from a local timestamp, or from the server's date, will book it to Tuesday. This is the single most common bug we expect to see, because it is invisible until settlement breaks and then it is everywhere.
  3. 03Overnight batch windows disappear. Position files, risk recalculation, margin runs, regulatory reporting extracts and reconciliations were scheduled at 2am for the excellent reason that nothing was happening at 2am. Something is now happening at 2am. Either these move to incremental real-time processing or they run against a moving target.
  4. 04Risk and margin go stale in a way nobody notices. If intraday risk runs during the day session and a full recalculation runs overnight, a portfolio that changes materially at 1am is carrying risk numbers computed against a position that no longer exists. In thin overnight liquidity, that is exactly when you would most want them accurate.
  5. 05Corporate actions collide with live trading. Splits, dividends and symbol changes have traditionally been applied in the overnight window precisely because the market was shut. Applying them while the market is open requires either a brief halt or genuinely careful sequencing, and getting it wrong means trades executed against a stale reference price.

The Charting And Data Problem Nobody Has Priced

This is our particular corner of the problem, so we will be specific. Charting a 23-hour session is not the same chart with more bars on it. Several things change at once, and every one of them shows up as a support ticket from a trader who thinks the chart is broken.

  • Session boundaries stop being visual furniture and start being information. When there is a one-hour gap between 8pm and 9pm and another between 4am and 4am the next day, the chart needs to render those gaps honestly rather than collapsing them, or the time axis lies about how long a move took.
  • Volume profile becomes bimodal and misleading. Overnight liquidity is a fraction of the day session, so a naive volume-weighted indicator computed across 23 hours will be dominated by the day and will misrepresent overnight conditions precisely when a trader is relying on it.
  • Intraday indicators built on a 6.5-hour day quietly change meaning. A 20-period moving average on five-minute bars covers a materially different span of market activity when the session is 23 hours. Every default period in your indicator library was chosen for a shorter day.
  • Data volume rises substantially, but not proportionally. You are not getting 3.5x the messages, because overnight is thin - but you are getting 3.5x the wall-clock time during which your ingestion, storage and rendering must be live, with no window to compact, re-index or deploy.
  • Gap detection and bar alignment need rethinking. Systems that infer a session break from an absence of trades will see the thin 3am tape as a break. Session structure has to come from a calendar, not from the data.

The Human Problem Is Harder Than The Technical One

Software can be made to run for 23 hours. People cannot, and this is where we think the industry is least prepared. A market that is open at 3am needs someone who can answer when the risk system alerts, when a client calls about a fill, when connectivity to a venue drops, or when an algo behaves unexpectedly. Investment banks with global footprints can genuinely follow the sun; a mid-sized asset manager or a regional broker cannot, and hiring a night desk to cover a session with a fraction of the volume is difficult to justify commercially.

This is, we think, the most underappreciated driver of AI automation in trading operations over the next two years, and it has nothing to do with alpha. When you cannot staff the hours, the automation is not a productivity choice - it is the only way to offer the session at all. The workload is well suited: monitoring, exception detection, escalation and routine triage are exactly the high-volume, rules-heavy, checkable tasks where automation is reliable. The judgement calls still wake a human; the difference is that the human is woken for the twelve things that matter rather than watching for all of them.

“Extended hours will do more to drive automation in trading operations than any productivity business case ever has, because the alternative is not a slower desk - it is no desk at all.”

What Firms Should Be Doing Between Now And December

  1. 01Decide, in writing, what 'the close' means in each of your systems, and make them agree. This is a policy decision before it is an engineering one, and the finance, risk, product and technology answers must match.
  2. 02Audit every place a date is derived from a timestamp. Anywhere trade date, value date or reporting date is computed from a local clock rather than from an exchange session calendar is a settlement break waiting for December.
  3. 03Find your batch jobs and classify them. Which genuinely need a quiet market, which merely assumed one, and which can move to incremental processing? The middle category is usually the largest and the cheapest to fix.
  4. 04Test your charting and analytics against a simulated 23-hour session before traders do. Indicator periods, session shading, volume profiles, gap handling. This is a day of work that prevents a month of tickets.
  5. 05Plan the coverage model honestly. If you intend to offer overnight access, decide now what is monitored by a person, what is monitored by automation, and what triggers a call - and test the escalation path at 3am at least once before it matters.
  6. 06Watch the follow-on approvals. NYSE Arca and Cboe are on different timelines with different hours. Building for 'Nasdaq's session' rather than for a configurable calendar means doing this work again twice.

Is This Actually Good?

Worth asking, since the industry has largely treated it as inevitable rather than desirable. The case in favour is genuine: investors in Singapore and London should not have to trade US stocks through a thin, fragmented, retail-hostile after-hours venue simply because of geography, and access is a real benefit. The case against is about liquidity quality. Spreads overnight are wider, depth is thinner, and a retail investor reacting to news at 2am may execute considerably worse than one who waits for the open - while believing they have been prudent by acting quickly. Whether extended hours are good for the person using them depends almost entirely on whether they understand the liquidity they are trading into, and that is a disclosure and interface design problem as much as a market structure one.

For platform builders, that has a practical consequence worth taking seriously: if you are giving clients access to the overnight session, showing them the same interface you show them at 11am is arguably a poor design. Displaying prevailing spread, available depth and a clear session indicator is not regulatory theatre; it is the difference between offering access and offering a trap.

The Bottom Line

On 6 December the US equity market stops having a daily close in any practical sense, and a lot of financial software is about to discover how much it depended on one. The extension itself is trivial; the removal of the quiet window is not, and it lands on trade dating, closing prices, batch processing, risk recalculation, corporate actions, charting defaults and the simple question of who is awake. None of this is unsolvable, and firms that treat it as a calendar-service problem rather than a hundred small configuration problems will get through it cleanly. The more interesting consequence is the second-order one: a market open 23 hours a day cannot be staffed the way a 6.5-hour market was, which makes automation of trading operations a requirement rather than an efficiency. That is the work we do as a trading platform and workflow automation developer in London, and December is closer than the project plans we are being shown assume.

References & Further Reading

Trading Platform DevelopmentReal-Time Data Ingestion & PlottingHigh-Frequency Charting SolutionsTrading Workflow automationmarket structure24x5 tradingInvestment Banking Trading Applications
Share Email
AI

AlchmAI Editorial

Research and analysis, London

The AlchmAI team writes about the markets, technology and regulation we work with every day. We build trading platforms, real-time charts and AI analysis tools for brokers, prop firms and fintech teams from our office in Mayfair, London. Every article lists its sources. Nothing we publish is investment advice.

This article is general information and commentary. It is not investment advice or a recommendation to buy or sell any investment. Important information

Keep reading

All articles