One contract · two study regimes · one continuous trade date

ES Overnight vs RTH: A Market-Quality Comparison

By Kyle ParrottUpdated August 28, 2026

ES can move while the U.S. cash core is closed, then meet a different liquidity regime when constituent stocks enter their core session. It is one futures contract, not two markets—but a fair comparison must account for unequal duration, scheduled information, holiday clocks, depth and execution cost.

Set boundaries Check failure cases

Overnight: Study-defined complement
RTH label: Cash-core aligned
Comparison: Normalized
Always leads: Neither

RTH is an ambiguous label unless defined

Declare the Session Split and Preserve the Full CME Schedule

For this protocol, “cash-core-aligned RTH” means 9:30 a.m.–4:00 p.m. America/New_York on a normal full NYSE day. “Overnight” is the remaining eligible ES trading time inside the same CME trade date after removing official pauses. This is a research convention, not a claim that every platform’s RTH template uses the same boundary.

Field	Cash-core-aligned RTH	Overnight complement	Integrity rule
Clock	9:30 a.m.–4:00 p.m. New York on normal days	Other matchable CME periods in the trade date	Retain UTC and timezone identifier
Schedule source	NYSE core-session calendar	CME product/holiday schedule	Use dated official calendars
Pauses	Any official halt/closure recorded	Maintenance and nonmatching states removed	Do not manufacture zero-return minutes
Contract	Same dated ES contract and trade date wherever possible		Flag roll migration and expiry
Early close	Separate regime	Recomputed complement	Never treat as a full normal day

CME markets U.S. equity-index futures as nearly 24-hour products, but current hours, pauses and holidays must be checked on the official CME schedule. The NYSE session guide supplies the normal cash core clock. U.S. daylight-saving rules can shift UTC offsets, so do not hard-code them.

A six-and-a-half-hour window cannot be compared to a longer complement by totals alone

Normalize Market Quality by Time, Activity and Order Size

Report raw totals, then comparable rates and distributions. Volume per session, for example, confounds duration with activity. A strategy cares about the book and price available for its order, not which interval accumulated the larger total.

Activity

Rates and clustering

Trades and quantity per eligible minute, inter-trade duration and concentration around scheduled events. Keep totals for capacity context.

Price

Comparable returns

Return distributions at equal horizons, realized variation, range and jump contribution. Avoid comparing raw point ranges over unequal hours.

Book

Width and depth

Time-weighted spread, depth through fixed tick bands, replenishment and size-conditioned sweep cost.

Execution

Experienced quality

Fill rate, delay, partial fills and shortfall for the same order policy, size and urgency in each state.

Equal-horizon returnr_t,h = ln(P_t / P_t-h) for the same h in both regimesUse synchronized, valid prices and disclose sampling. Overnight-to-open returns are a separate endpoint measure, not interchangeable with intraregime returns.

CME’s data catalog offers trades, top-of-book, depth and market-by-order records, while its equity execution statistics distinguish width, depth, fill percentage and slippage. Bar data alone can support return and range comparisons but not a queue or executable-capacity claim.

Information crosses the boundary before and after 9:30

Treat Event and Auction Handoffs as Separate States

Many scheduled U.S. releases occur before the cash core opens; FOMC statements are released on their announced calendar during the U.S. day. The resulting overnight or RTH label is a clock classification, not a causal conclusion. Preserve the prior, timestamp and liquidity response.

Before information

Baseline

Capture price, return state, spread, depth, active contract and cash-market status.

At timestamp

Repricing

Measure jumps, quote depletion, trades and recovery using source time; direction remains unknown ex ante.

Cash open

Handoff

Record opening auction/cash-core activation and whether ES retains, extends or reverses the prior move.

Afterward

Attribution limit

Separate observed path from any mechanism not supported by primary contemporaneous evidence.

Use the responsible primary calendars: BLS releases, BEA releases and FOMC communications. Their timestamps support event labels; they do not establish which session “should” lead.

The comparison should change permissions, not produce a slogan

Map Session Evidence to the Actual Task

Build separate distributions for urgent hedging, passive execution, stop exposure and research observation. A session can offer attractive typical spreads but worse adverse-tail depth. Require the current state to match the validated sample.

Task	Compare	Permit	Reject when
Urgent risk reduction	Tail sweep cost, fill delay and capacity	Execute / reduce / wait	Current depth is out of distribution
Passive order	Queue turnover, fill rate and adverse selection	Join / improve / cancel	Event repricing invalidates the state
Overnight hold	Gap path, maintenance exposure and stop slippage	Hold / hedge / flatten under policy	Stress loss exceeds risk authority
Session strategy	Forward net results inside matched regimes	Enable only after holdout	Costs or calibration fail

Most session conclusions fail at the boundary

Know When the Comparison Is Invalid

Fail closed when the data cannot distinguish the regimes or when results do not survive fair controls. “Overnight is thinner” and “RTH is best” are not sufficient conclusions without a defined sample, order and uncertainty.

Clock failure

Timezone conversion, holiday, early close or maintenance state is unresolved.

Output: unscorable day

Composition failure

Events, roll days or volatility regimes explain the apparent session difference.

Output: conditional only

Execution failure

Return differences do not survive state-matched spread, depth, slippage and fees.

Output: no tradable evidence

Both regimes use the same dated contract and trade-date logic.
Minutes when the market cannot match orders are not treated as quiet liquidity.
Totals and normalized rates are shown together.
Event, holiday, volatility and roll strata are fixed before comparison.
Order size, type, side and latency are identical across matched tests.
No session is declared a permanent price-discovery leader.

Research status as of August 28, 2026

No original ES overnight-versus-RTH return, liquidity, price-discovery or trading-edge result is reported. This article defines a comparison protocol and failure cases.

Sources, methods and editorial disclosure — reviewed August 28, 2026

CME U.S. Equity Index futures and CME ES product page for the nearly 24-hour futures-market context.
CME current trading-hours and holiday calendar for changeable schedule controls.
NYSE trading-session FAQ for the normal cash core-session clock.
CME futures and options data catalog and CME Equity Execution Statistics FAQ for measurement layers.
U.S. DOT uniform-time guidance for timezone and daylight-saving controls.

Sources were reviewed August 28, 2026. The overnight and cash-core-aligned RTH labels are explicit research conventions, and no original result is claimed.