Skip to article content

Candidate → hypothesis → reaction → deletion

ES Intraday Support and Resistance: A Level Workflow

An ES chart can offer yesterday's high, overnight low, settlement, open, VWAP, swing points, round numbers and several profile references before the first order. If every line is “important,” any reversal can be explained afterward. A usable level is a timestamped hypothesis with a reason, rank, trigger, invalidation and deletion rule.

Use only values knowable at the decision time

Create a Bounded Candidate Set

A level source is a way to generate a hypothesis, not evidence that price must react. Fix the session definitions, data feed and calculation rules before using any candidate.

Candidate sourceTimestamp rulePrimary useCommon contamination
Prior RTH high/lowComplete only after the defined prior RTH closePreviously observed session extremesPlatform RTH template differs
Prior settlementUse exchange/vendor settlement field and trade dateOfficial daily accounting referenceConfused with cash close or last trade
Overnight high/lowKnown only as of the current timestamp until the overnight window endsCurrent-session extremesLater high/low leaks future information
Cash openAvailable after the opening print under the chosen definitionOpening referenceFutures open substituted for cash open
VWAPRecomputed using declared reset, price and volume inputsEvolving volume-weighted benchmarkDifferent platforms use different resets or prices
Predefined swingConfirmation must use only completed data under a frozen ruleRecent structural pivotFuture bars required to label the pivot
Round incrementMechanical grid declared before the sessionCandidate onlySelective attention after price turns

Limit the number of active candidates. For example, a plan may allow no more than five primary references and three secondary references. The exact cap is a policy choice to test; the important control is that the set cannot expand whenever price approaches a new line.

Rank before price arrives

Score Relevance Without Inventing Precision

A ranking rubric makes candidate selection auditable. It does not turn an ordinal score into a probability.

Dimension012
Definition qualityAmbiguous or platform-dependentDefined but weakly verifiedReproducible from stored data
AgeOutside permitted lookbackOlder but still allowedCurrent or immediately prior session
Independent confluenceNoneSame calculation familyDifferent source/clock agrees within tolerance
Prior interactionAlready repeatedly crossedOne ambiguous testUntested under the written definition
Execution qualitySpread/depth failsBorderlinePasses for intended quantity
Do not double-count aliases.

A prior close, settlement and platform “close line” may be the same or nearly the same input. Calling that three independent confirmations exaggerates evidence.

rank = declared rubric applied before first testestimated win probability

A price is not always a one-tick wall

Declare the Test Region and Clock

ES outright prices move in 0.25-point increments. A research definition may test an exact price or a bounded tolerance around it, but the tolerance must be fixed before observing the reaction and expressed on the valid tick ladder.

Exact touch

Trade at the reference tick

Simple to label, but it can treat one tick above and one tick below as fundamentally different without evidence.

Fixed band

Predeclared tick radius

For example, a hypothetical ±2-tick band. This is a test parameter, not a recommended zone.

State-scaled band

Forward-known volatility rule

The scale must be available before contact, capped, rounded to ticks and frozen for evaluation.

  • First-touch clock: define whether a touch means trade, bid/offer contact or bar range.
  • Entry clock: define whether action can occur on contact or only after a completed confirmation.
  • Expiry clock: delete or downgrade the candidate after a stated time, event or number of tests.
  • Crossing rule: decide whether price traversing the full band invalidates, changes side or merely records a test.

Observation before action

Define Acceptance, Rejection and No-Decision Branches

Price enters the band

Does the reaction meet the prewritten trigger?

Yes: continue to market-quality and risk gates. No: wait; a touch alone is not an entry.

Price holds beyond the band

Has acceptance been defined?

Yes: invalidate the rejection thesis or evaluate a separate acceptance setup. No: delete the level; do not reverse automatically.

Price oscillates through

Has the test count or time expired?

Yes: retire the candidate. No: retain only under the stated decay rule.

ObservableWhat it can supportWhat it cannot prove
Price displacementDirection and distance after contactWho caused the move
Trade flowExecuted volume at price over timeParticipant identity or future intent
Displayed depthVisible queued quantity at timestampsGuaranteed future liquidity
ReplenishmentDisplayed quantity returning after executionsOne hidden participant or permanent support
Spread/impactCurrent execution qualityProfitable reaction

A valid hypothesis can still be untradeable

Run Execution Gates at the Moment of Action

Displayed book state changes through new orders, cancellations and trades. CME Market by Price aggregates quantity at price; Market by Order can provide anonymous order-level granularity, but neither identifies the customer or guarantees displayed size remains.

Spread

Maximum ticks

Reject action if the inside spread exceeds the plan's stated threshold.

Depth

Quantity-aware

Estimate price impact for the intended order, not for one contract unless size is one.

Staleness

Feed-age limit

Block on delayed data, dropped sequence or mismatched contract timestamps.

Event

State override

Scheduled release, halt, roll migration or abnormal cancellation can invalidate ordinary cost assumptions.

A line must be allowed to die

Specify Invalidation Before Contact

Invalidation typeExample definitionAction
PriceCompleted acceptance beyond the far side under a fixed ruleDelete rejection hypothesis
TimeNo qualifying contact before stated session timeExpire candidate
TestsMaximum number of traversals reachedRetire or downgrade
EventScheduled release or verified unscheduled shock changes stateRebuild candidate set
ContractLiquidity migrates to a different ES expiryRecalculate from the new dated contract
DataSession template, source or calculation is wrongVoid the level and affected decisions
Invalidation is not automatic reversal.

Failure of a support hypothesis only says that hypothesis no longer authorizes a long. A short requires its own context, trigger, risk and execution permissions.

Preserve misses and deletions

Use a Level Ledger, Not an Annotated Victory Chart

FieldRecord
IdentityLevel ID, dated ES/MES contract, source, price/band and creation timestamp
CalculationSession template, raw data reference, formula and tick rounding
RankEvery rubric component before contact
ContactFirst eligible touch timestamp and market-quality snapshot
BranchReject, accept, wait, expire or data-invalid
ExecutionDecision, arrival, order, fill and exit fields if a trade was authorized
DeletionExact invalidation, time, test count, event or contract-roll reason

Review all candidates, including untouched levels, ambiguous reactions, no-trades and deleted hypotheses. Selecting only lines that preceded a visible turn manufactures accuracy after the fact.

Final state

No reaction, no trade

A high-ranked level is still only a candidate. Without the declared reaction, liquidity, risk and order gates, the correct output is wait or delete.

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

Sources were reviewed August 28, 2026. This page defines a forward-safe workflow and reports no original level study, hit rate, reversal probability, optimal tolerance or trading edge. All level IDs, prices, times and rubric examples are hypothetical.