Nasdaq says it plans to introduce a 9:00 p.m.–4:00 a.m. ET overnight session on December 6, creating 23 hours of continuous weekday trading with a one-hour nightly pause for processing and trade-date rollover. The announcement follows an SEC roundtable on operational preparations for broader overnight trading.

LSEG’s September 24 market-data analysis frames the transition more sharply: the old assumption of a clearly inactive overnight period is becoming unreliable. End-of-day processing, reconciliation, reference-data updates and scheduled maintenance have historically depended on that quiet interval. Compressing it can change not only execution, but also the meaning and timing of familiar datasets.

The trading day becomes a model choice

A daily bar looks objective until a market trades through most of the clock. Researchers must then decide which session owns a 1:00 a.m. trade, which close anchors a return, and whether overnight observations belong to the prior U.S. date, the next calendar date or a distinct session.

Those choices can alter daily returns, volatility estimates, gaps, volume profiles and event labels. A model trained on historical regular-session bars can appear to degrade when the real break is a changed session definition. Comparisons across vendors may also diverge when consolidated and venue-level overnight coverage do not begin at the same time.

OpenQFR view: Session boundaries should become explicit research parameters, not invisible defaults. Any reported edge that changes materially under two reasonable trading-day definitions is evidence of fragility.

Thin hours are not unimportant hours

Overnight trading is likely to carry less volume than the core session, but low volume does not imply low information content. Corporate announcements, macro releases and geopolitical events can arrive while liquidity is sparse. A small number of trades may therefore move prices sharply while spreads and market impact remain larger than a backtest based on daytime costs assumes.

This creates a familiar quantitative trap: more timestamps can look like more opportunity, while executable capacity may remain limited. Overnight signals need their own spread, depth, fill-rate and slippage estimates. A strategy that pools all hours together may learn the wrong relationship between apparent return and attainable return.

Corporate actions and reference data become live problems

The SEC chairman’s roundtable remarks highlighted unresolved questions around corporate actions, material-information dissemination and whether issuer systems such as EDGAR should operate for longer hours. These are not administrative details for a systematic strategy. Splits, dividends, symbol changes and announcements determine whether historical prices and live positions are interpreted correctly.

When the maintenance window contracts, a delayed reference update can meet a live market rather than a closed one. Pipelines should record when a change became known, when a vendor applied it and which session used it. Otherwise, research can accidentally apply corrected information earlier than it would have been available in production.

A practical quant checklist

What the evidence does—and does not—say

The published material supports a structural conclusion: U.S. equity infrastructure is preparing for much longer trading hours, and the industry expects a substantial change in data operations. It does not establish that overnight trading will provide persistent alpha, adequate capacity or daytime-quality execution. Those claims require observed spreads, depth, outages and fills after the new sessions operate at scale.

For now, the research advantage is defensive. Teams that make their hidden time assumptions visible are less likely to confuse a data-boundary error with a market signal.

Primary sources

This report is educational journalism based on public information. It is not investment, legal or tax advice.