Displayed orders · executed trades · bounded inference
ES DOM and Time & Sales: What the Tape Can Prove
A large ES bid appears, price approaches, and the size vanishes without trading. The DOM proved that displayed quantity changed. It did not prove a buyer defended the price, a seller was trapped, or anyone spoofed. Useful tape reading begins by separating orders, cancellations, executions and price response on one clock.
Four views, four different claims
Know What Each Screen Observes
| Layer | Direct observation | Does not directly identify |
|---|---|---|
| DOM / order book | Displayed bid and offer prices, quantities and updates within the feed's depth | Undisplayed interest, participant identity, future cancellation or future execution |
| Time & sales | Executed trade price, time and size as represented by the feed | Full strategy, inventory, motive or whether the same participant returns |
| Own order state | Submitted, acknowledged, queued, partially filled, filled, cancelled or rejected states | Other customers' identities or intentions |
| Price response | Subsequent trades, quote movement, spread, impact and replenishment | A unique cause when several events coincide |
A trade has both sides. An aggressor-side classification can describe which resting quote was hit under a chosen rule; it does not prove whether either participant is opening, closing, hedging or expressing a directional view.
The DOM is a rendering of a data product
Resolve Market by Price, Market by Order and Platform Transformations
CME describes Market by Price (MBP) as aggregated quantity and order count at each price, with a maximum of ten disseminated levels. Market by Order (MBO) provides anonymous order-level granularity and full depth, including exchange order and priority identifiers. MBO still contains no customer-identifiable information.
MBP
Price-level aggregate
Useful for spread, top-of-book depth and changes by level. Individual order size and exact queue position cannot be inferred reliably from the aggregate.
MBO
Anonymous order detail
Can expose individual displayed orders and queue sorting. It does not reveal firm identity, beneficial owner, motive or off-book interest.
- Entitlement: confirm whether the account receives real-time or delayed data and which depth product.
- Platform: document aggregation, filtering, update throttling, trade bundling and color rules.
- Contract: ensure DOM, tape, chart and order ticket all use the same ES month and year.
- Sequence: detect feed gaps, reconnects and book resets; do not interpret an incomplete book.
- Clock: retain exchange or feed timestamps when available, not only the local display clock.
Order matters at millisecond scale
Put Book Updates, Trades and Your Order on One Timeline
Exchange event
Quote add, modify, cancel or trade occurs in venue sequence.
Feed distribution
The message travels through the market-data path.
Local receipt
The platform receives and processes the update.
Screen render
The DOM or tape visually updates after local processing.
Decision and order
The trader reacts; the order follows a separate outbound/acknowledgement path.
| Timestamp | Why retain it | Failure if substituted |
|---|---|---|
| Exchange/feed event | Orders venue events | Local latency is mistaken for market sequence |
| Local receive | Measures inbound delay and gaps | Display time hides processing backlog |
| Decision | Defines information actually available | Later book state leaks into rationale |
| Order submit | Starts outbound execution path | Reaction time and routing delay are merged |
| Acknowledgement/fill | Establishes authoritative order state | Screen intent is mistaken for venue acceptance |
If a replay cannot preserve or reconstruct these clocks, label the limitation. A screen recording can help review what the operator saw, but it is not a lossless substitute for sequenced order-book, trade and order-event data.
Translate visual impressions into tests
Write a Tape Hypothesis With Rival Explanations
Hypothetical replenishment test
Candidate: executed sell volume repeatedly reaches a bid while displayed quantity refreshes and price does not trade through for a declared interval. Rivals: several independent orders arrive, price is constrained by another market, trade classification is wrong, feed updates are missing, or the response is random. Invalidation: price traverses the level with specified impact or the data-quality gate fails.
- Window
- Predeclared
- Quantity
- Measured
- Response
- Forward only
- Finding
- Not reported
| Label | Minimum observable definition | Rival to test |
|---|---|---|
| Replenishment | Displayed quantity returns at a price after executions remove quantity | Independent new orders rather than one hidden order |
| Cancellation burst | Displayed quantity falls without matching executed volume in synchronized data | Feed reset, state change or routine order management |
| Price impact | Forward price change per aggressive quantity over fixed horizon | Concurrent news or cross-market repricing |
| Failed impact | Large classified flow with response below a threshold | Threshold chosen after seeing outcome |
| Queue persistence | Order/level remains displayed through a defined event count | Display throttling or aggregation artifact |
“Absorption,” “iceberg,” “trapped traders” and “buyers in control” are not executable definitions. State the data, window, threshold, response and invalidation.
Observation does not grant order permission
Run Risk and Execution Gates After the Signal
Hypothesis passes
Is the feed complete and current?
No: no order. Yes: confirm exact contract, event state and session.
Market quality passes
Does intended quantity fit the book and cost ceiling?
No: reduce to a recalculated whole-contract quantity or zero. Yes: choose the approved order branch.
Order submitted
Did acknowledgement and fill state reconcile?
No: execute the unknown-order incident procedure. Yes: protect actual filled quantity and log slippage.
| Choice | Potential benefit | Cost or failure | Required measurement |
|---|---|---|---|
| Resting limit | Price control and possible passive fill | Non-fill, adverse selection, queue uncertainty | Queue, fill rate and post-fill path |
| Executable order | Greater immediacy within order protections | Spread, book sweep and adverse movement | Decision/arrival/fill prices and depth |
| Stop limit | Trigger plus explicit limit | Can remain unfilled beyond the limit | Trigger, eligible range and residual state |
| Stop with protection | Exchange-defined protected range after trigger | Residual quantity can remain at protection limit | Current product support and protection points |
Platform names can represent synthetic broker behavior rather than the exact CME order type. Verify the current order definition and test the rejection, partial-fill and cancellation paths before live use.
Intent is not a DOM field
Do Not Diagnose Spoofing From One Visual Sequence
The Commodity Exchange Act prohibition concerns bidding or offering with intent to cancel before execution. CFTC guidance evaluates intent using market context, trading patterns, fill characteristics and other facts. Legitimate orders are routinely modified or cancelled. A retail display supplies neither participant identity nor the full evidence needed to determine intent.
You may say
Displayed size was cancelled
When synchronized order-book data show quantity removed without corresponding trade volume.
You may test
Cancellation predicted response
Using a frozen, repeatable protocol with rivals, costs and out-of-sample data.
You may not infer
A named manipulator or intent
Not from an anonymous DOM snapshot or a single cancellation pattern.
Use “displayed quantity was added/removed,” “trades occurred,” and “price responded.” Reserve legal conclusions for authorities with the required records and investigation.
Review the full decision path
Build a Replay Record That Can Falsify the Method
- Capture source identity.Provider, MBP/MBO entitlement, contract, trade date, sequence status and platform transformations.
- Freeze the hypothesis.Observable definition, window, threshold, rivals, expiry and invalidation.
- Mark the decision clock.Preserve exactly what was known before action.
- Reconstruct the order.Submission, acknowledgement, queue/partial states, fills, cancels and protection.
- Measure costs.Spread, arrival shortfall, adverse selection, fees and missed-trade outcomes.
- Keep failures.False signals, no-fills, data gaps, rejected orders and zero-contract decisions remain in the sample.
- Retire when falsified.If the hypothesis fails after costs or cannot be observed reliably, stop using it.
Decision standard
The tape must improve a defined decision after costs
A compelling replay clip is not an edge. Require a prospective protocol and independent evaluation before granting the observation live authority.
Sources, data boundaries and editorial disclosure — reviewed August 28, 2026
- CME Market by Order FAQ for MBO, MBP, full-depth, anonymous order and queue-information scope.
- CME About Time & Sales for the exchange report's trade-date, time, price, size and condition fields.
- CME Liquidity Tool Methodology for order-book construction, spread and depth measurements.
- CME Futures Order Types for current order types and protection mechanics.
- CFTC interpretive guidance on disruptive practices for the intent and context boundary around spoofing.
- ES/MES canonical mechanics for current contract arithmetic and lifecycle controls.
Sources were reviewed August 28, 2026. This page reports no original DOM study, predictive accuracy, fill advantage or trading edge. All sequences and thresholds are hypothetical examples; data entitlements and platform transformations must be verified for the actual setup.