Reset · price input · availability clock

ES Session Highs, Lows, and VWAP: Timestamp-Safe Usage

By Kyle ParrottUpdated August 26, 2026

Two platforms can display different ES VWAPs without either calculation being broken. One may reset at the CME trade date and use every trade; another may reset at the U.S. cash open and approximate price from bars. The labels look identical while the inputs are not.

Specify the benchmark Build the review log

Reset: Named session event
Price: Trades or declared proxy
Volume: Eligible ES quantity
Final high/low: Known after session

A benchmark is a complete recipe

Specify Session, Price and Volume Inputs

Store the dated contract, timezone and reset event. CME and NYSE schedules can change around holidays; a platform template is not a permanent authoritative calendar. Use raw trades when the claim is trade-weighted VWAP, and label any bar-based calculation as an approximation.

Trade-based session VWAP through time tVWAP_t = Σ_i≤t(p_iq_i) / Σ_i≤tq_ip_i is an eligible ES trade price and q_i its quantity after the declared reset. Corrections and zero/missing quantity require a written rule.

Choice	Option A	Option B	Why results differ
Reset	CME trade-date/session event	9:30 a.m. ET cash core open	Different observations and overnight inventory
Price	Every eligible trade	Bar close or typical-price proxy	Aggregation loses within-bar path and size detail
Contract	One dated ES expiry	Continuous series	Roll mixing and adjustment can change levels
Corrections	Apply source correction state	Ignore corrections	Numerator and denominator can diverge
Missing data	Invalidate interval	Silently continue	Benchmark may look precise but be incomplete

CME’s historical data catalog distinguishes time-and-sales, top-of-book, depth and market-by-order records. Use the current CME schedule and, for a cash-core reset, the NYSE session guide.

Developing and final levels are different variables

Keep Every Level on Its True Availability Clock

At 11:00 a.m., the developing session high, low and VWAP are knowable through 11:00 after source and processing latency. The final session high, low and VWAP are not. Using final levels earlier creates look-ahead bias.

Level	Available at decision time t	Valid research use	Leakage example
Developing high	Maximum eligible price from reset through t	Candidate live reference after latency	Using the day’s eventual high at noon
Developing low	Minimum eligible price from reset through t	Candidate live reference after latency	Using the eventual low before it forms
Developing VWAP	Eligible price-volume observations through t	State feature or execution benchmark	Using full-session VWAP for an earlier entry
Prior-session final	Known after prior session completion and corrections	Next-session reference with version timestamp	Substituting current-session final value
CME settlement	Officially disseminated under CME procedure	Separate official benchmark	Calling it the same as platform VWAP or last trade

VWAP and CME daily settlement are not interchangeable.

CME determines official settlements under product procedures. A user-calculated session VWAP depends on the declared input interval and data. Keep the names, timestamps and uses separate.

A benchmark does not generate a trade by itself

Turn Each Use Case Into a Falsifiable Test

Define the state at t, the eligible order after t, the forward horizon and the outcome. Avoid flexible phrases such as “VWAP held” or “the high rejected” unless the price, quantity and time rules are machine-readable.

VWAP interaction

Cross, touch or rejection?

Specify bid/ask/trade side, minimum distance, dwell or recross condition, latency and target horizon. Test continuation and reversal as separate hypotheses.

Developing extreme

Break or retreat?

Freeze the high/low at t, then measure an executable extension and later path. Never move the reference to the eventual session extreme.

Prior-session level

Arrival response

Measure spread, depth, traded quantity and forward return when price approaches a known prior level. A touch does not prove support or resistance.

Minimum observation record

For every candidate, retain decision time, level type and version, reset, contract, price source, distance in points and volatility units, spread, depth, event state, session state, order policy, fill record and forward path. Report non-events and repeated touches under a declared independence rule.

Condition the analysis on scheduled releases, gap state, volatility, time of day, holiday and roll. Use a chronological holdout. If a benchmark appears useful only after selecting one reset, one band and one horizon from many, report the multiple-testing problem rather than a clean edge.

A visible line is not executable capacity

Apply Spread, Depth and Order-Side Cautions

VWAP, session highs and lows are computed prices, not standing orders. Before acting, inspect the current book and simulate the declared quantity. A stop near a reference can gap through; a passive order can sit unfilled while a last-trade chart touches.

Eligible

Level and state valid

The reference is timestamp-safe, source data are complete and current spread/depth fit the validated sample.

Reduce

Capacity supports less

A smaller whole-contract quantity passes the preregistered sweep and risk stress.

Wait

Benchmark still forming

The range or VWAP is changing rapidly, or an event transition requires a new observation.

Reject

Reference contaminated

Wrong reset, contract mix, missing data, stale quote or future-known level makes the signal invalid.

For actual risk authorization, send the executable entry, invalidation distance and stressed cost to ES Position Sizing by Account Balance. A benchmark never overrides the account’s risk limit.

Version the benchmark and the conclusion

Maintain a Calculation and Review Log

A reviewer should be able to recreate the plotted value and the order decision. Archive the exact code version, data revision, reset, instrument, rules and outcome. Never update only unfavorable observations.

Dated ES contract and roll rule are explicit.
Reset event, timezone and holiday treatment are explicit.
Trade-based VWAP and any bar proxy have different names.
Every high, low and VWAP level carries an as-of timestamp.
Corrections, missing intervals and feed changes have visible status.
Tests include executable side, latency, capacity, slippage and fees.
Training choices are locked before the untouched forward period.

Platform mismatch

Values diverge because hidden inputs cannot be reconciled.

Action: do not combine series

Leakage

A final high, low or VWAP enters before session completion.

Action: invalidate result

Economic failure

Forward behavior does not survive executable costs.

Action: no trading claim

Research status as of August 26, 2026

No original ES VWAP, session-high or session-low reaction probability or after-cost edge is reported. This page defines calculations, tests and audit controls only.

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

CME futures and options data catalog for official trade, top-of-book, depth and market-by-order data categories.
CME daily settlement procedures for the boundary between official settlement and a user-calculated VWAP.
CME E-mini S&P 500 product page and CME current schedule for contract and session context.
NYSE trading-session FAQ for the normal cash core reset used in one disclosed variant.
CME Equity Execution Statistics FAQ for width, depth, fill and slippage measurement categories.

Sources were reviewed August 28, 2026. The VWAP formula is conventional, while the benchmark cases are test specifications; this article reports no tested level reliability or trading edge.