Nasdaq starts 23-hour trading on December 6
← All posts
marketsmarket-structureagents

· 6 min read

Nasdaq starts 23-hour trading on December 6

The overnight session runs 9pm to 4am ET, with no market orders, 20% price bands, and a trade date that rolls forward three hours before the calendar does. The traders best equipped for those hours are the ones that already run on a timer.

Nasdaq's Global Trading Hours start on December 6. From that date US equities trade 23 hours a day, five days a week, closed only from 8pm to 9pm ET for maintenance. NYSE Arca is targeting the same date for its own overnight session, subject to SIP and DTCC readiness.

The clearing side is already done. NSCC went live with a 24x5 schedule on June 29, running Sunday 8pm to Friday 8pm ET, which is the piece that had to land first. Everything since has been exchange plumbing and testing, with UAT weekends running through December 4.

Seven new hours of US equity trading, five nights a week. The question is who fills them.

The mechanics are not the same as the day session

Four rules in the Nasdaq filing change how a strategy has to be built.

Market orders are not accepted overnight. An order sent into the 9pm to 4am session has to carry a limit price, so any system that leans on market orders for certainty of execution needs a different path before December.

Orders do not rest across the session boundary. Nasdaq cancels orders outstanding in the night session when it ends, so an order with a time-in-force reaching past 4am gets cancelled rather than carried into the early session and has to be re-entered. This is a rule about orders sitting at the exchange overnight, not about the stop held in your broker's system, which is unaffected and gets released into sessions the way it always has. It is worth knowing which of those two you actually have. On Nasdaq this is partly moot already: it stopped accepting GTC order entry on January 1 and decommissioned the attribute on February 2, so an order resting at that exchange indefinitely is not a thing there any more.

Price bands replace the auction. Orders more than 20% above or below the adjusted official closing price are rejected, with a minimum band width of $3.00 for stocks priced at a dollar or more. There is no opening auction on the overnight session to establish a reference price, so the previous close does that work for seven hours.

The trade date rolls at 9pm, three hours ahead of the calendar. A fill between 9pm and midnight carries the next trading day's trade date, so an execution at 10pm on Monday is a Tuesday trade. From midnight through 8pm the trade date and the calendar date agree again. Settlement follows the assigned date, which means that for three hours every evening anything computing a daily return has to know which of the two dates it is using.

Overnight favors systematic participants

A human desk cannot cover 9pm to 4am without hiring for it. Most cannot justify the headcount for a session that will start thin. Software has no such constraint. A strategy that already runs on a schedule does not care whether it wakes at 10am or 2am, and the marginal cost of the extra hours is close to zero.

That asymmetry is why the overnight session is likely to be disproportionately automated from day one. Not because machines are better traders, but because they are the only participants for whom showing up costs nothing.

The tape those hours produce will look different from the day session. Thinner books, wider spreads, fewer participants setting prices, and no auction to anchor the open. A 300-share print can move a quote further than it should. Systems calibrated on regular-session liquidity will read those moves as signal when they are mostly noise, and the first months will sort out which ones can tell the difference.

What changes for an automated strategy

Indicators need recalibrating. Most technical signals are computed from bars built out of regular-session prints. Adding seven hours of low-volume tape changes what a bar contains. A 14-period RSI on hourly bars covers a different amount of real trading than it did in November, and a threshold tuned against the old bars will fire at different times against the new ones.

Risk rails need rethinking. A fixed percentage stop is reasonable against a liquid book. In an overnight session with a 20% band as the only structural protection, a thin print can trigger a stop that the morning would never have touched. Whether that reaches you depends on how your broker handles stops outside regular hours, which is worth asking before December rather than after. Operators will either widen the stops, gate them by session, or accept getting filled at prices almost nobody else traded at.

Decision frequency becomes a real choice rather than a default. Today a closed market is a free filter on how often a strategy acts. Extend the day from 16 hours to 23 and a system checking every fifteen minutes gets 28 more decision points, most on thinner information. More opportunities to act on less evidence is how a sound strategy turns into churn, and the operators who do well overnight will be the ones who deliberately do less rather than more.

Crypto already ran this experiment

Continuous markets are not new. Crypto has traded 24/7 for a decade, and automated trading grew up there precisely because no human schedule matches an always-open venue. The overnight sessions are quiet, the books are thin, and the participants are overwhelmingly systematic. US equities are about to acquire a version of the same shape, five nights a week.

The other half of that convergence is coming from the opposite direction. The SEC's tokenized stock exemption, issued in September, opens a path for US equities to trade onchain on venues that never had a closing bell in the first place. One route extends the exchange day. The other puts equities on infrastructure that has no concept of a day at all.

December 6 is the date the first one starts.

Sources: Nasdaq Equity Trader Alert 2026-46, NYSE extended hours, DTCC on NSCC 24x5 go-live, SIFMA extended trading hours.