Market quality · order size · executable evidence

6N Liquidity Guide: Measure the Market You Can Trade

By Kyle ParrottUpdated August 17, 2026

Two traders can view the same 6N chart range and face different costs. One asks for one contract with time to rest a limit; the other must cross several levels immediately. Price range is identical. Executable liquidity is not.

Build the measurement record Use the decision card

Displayed: Spread + depth
Traded: Volume + prints
Experienced: Fills + slippage
Fixed best hours: Not claimed

Evidence starts at the feed

Match the Data Fields to the Liquidity Claim

Daily volume can describe participation but cannot reveal an intraday spread. Top-of-book quotes can measure the inside market but cannot prove queue position. The claim must fail when the feed lacks the fields it requires.

Evidence layer	Required fields	Supports	Does not prove
Trades	Exchange timestamp, price, quantity, sequence/correction fields	Turnover, trade count, size distribution	Bid, ask, resting depth or missed orders
Top of book	Bid, ask, displayed quantities, update time	Quoted spread and displayed inside depth	Depth beyond best price or actual fill priority
Depth	Multiple price levels with update ordering	Hypothetical marketable sweep at observation time	Future cancellations or latency path
Order level	Order identifiers, priority events, modifications and executions	Queue-aware research where schema permits	Your unobserved broker and network latency
Own orders	Decision, submission, acknowledgment, fill and cancel timestamps	Implementation shortfall and realized fill quality	Counterfactual fills for orders never submitted

CME DataMine offers distinct settlements, Market by Order and PCAP datasets. That catalog is evidence that dataset types differ; it is not evidence that any particular 6N sample is complete. No market-data order was placed for this page.

Three clocks, one observation

Build an Event- and Daylight-Saving-Aware Clock Map

6N trades on an exchange clock while New Zealand and U.S. information arrives on local civil clocks. The offsets change on different dates. Hard-coded “New York time equals New Zealand time plus X” rules will eventually misclassify windows.

Source

Preserve timestamp, timezone name, precision and sequence.

Resolve

Apply historical timezone rules, including ambiguous or skipped local times.

Normalize

Store UTC plus original local time and CME trade date.

Tag

Attach official event, holiday, open, pause, roll and expiry states.

Bin

Assign only after the entire clock policy is frozen.

Scheduled hours are not a liquidity guarantee.

CME’s trading-hours page states that hours can change and publishes holiday schedules. An open venue can still have insufficient market quality for a given order.

Compare like with like

Design Measurement Bins Around Information States

A permanent “best time” label hides the conditions that produced the estimate. Use bins that separate ordinary windows from scheduled announcements, holidays, roll migration and unusual market states.

New Zealand information window

Local clock

AnchorOfficial release time

CompareMatched non-event days

U.S. information window

Eastern time

AnchorOfficial release time

ControlOverlap and daylight saving

Ordinary intraday window

UTC

AnchorFixed clock bin

ExcludeRegistered event buffers

Contract migration window

Dated 6N

AnchorExpiry-specific activity

RecordFront/next participation

Use the RBNZ publication guidance, the Stats NZ release calendar, and relevant U.S. agency calendars as versioned inputs. A present-day calendar must not be projected backward over historical releases.

Confounds that change the denominator

Control Events, Holidays, Rolls and Stale Books

The measurement table needs an eligibility flag before it needs an average. A bin that silently drops the hardest observations can look liquid precisely because its failures vanished.

Event control

Separate anticipation from reaction

Define pre-event, immediate, recovery and ordinary windows. Do not mix them into one time-of-day statistic.

Contract control

Measure the intended expiry

Track volume, open interest and quotes by dated contract. A generic continuous symbol can hide where trading migrated.

Data control

Reject unusable states

Flag crossed or locked quotes, sequence gaps, one-sided books, feed resets and intervals without heartbeat evidence.

Retain zero-trade bins. Zero observed trades and missing data are different states.
Version holiday schedules. Early closes and pauses cannot be inferred from a normal-week template.
Keep every expiry. The front contract is an outcome of a roll rule, not a timeless symbol.
Report sample attrition. Show how many windows each control removed and why.

A panel, not one number

Measure Spread, Depth, Turnover and Price Impact Together

Each metric captures a different failure mode. Publish distributions and tails by order size and information state; an overall median can conceal precisely the conditions in which an order becomes dangerous.

Quoted

Inside spread

Ask minus bid at valid two-sided updates. Report time-weighted and observation-weighted versions with update rules.

Displayed

Depth curve

Quantity available through each price level for declared contract counts. Displayed size can cancel before arrival.

Traded

Volume and cadence

Contracts, trade count, intertrade duration and size distribution. Prints do not reveal unfilled demand.

Experienced

Implementation shortfall

Side-adjusted fill price versus the predeclared decision or arrival benchmark, plus fees and unfilled quantity.

side × (fill − benchmark) × 100,000 NZD × filled contractsUSD fees + declared USD opportunity cost for unfilled quantityImplementation shortfall in USD

Sign and unit convention: use +1 for a buy and −1 for a sell so a worse execution is positive cost. Fill and benchmark are USD per NZD; multiplication by the 100,000 NZD contract unit and filled-contract quantity converts their difference to U.S. dollars before dollar fees and opportunity cost are added. Choose the benchmark and unfilled-quantity policy before observing the fill.

Liquidity is task-dependent

Replay the Order You Actually Intend to Send

A one-contract market order, a patient limit and a multi-contract stop do not consume the same evidence. Define size, urgency, order type, latency, time-in-force, cancellation logic and partial-fill handling before applying any threshold.

Marketable entry

Walk declared size through contemporaneous depth, then stress adverse book changes during latency.

Passive entry

Require queue evidence appropriate to the feed. A touched price is not automatically a fill.

Protective stop

Model trigger-to-order conversion and gaps. Trigger price is not a guaranteed execution price.

Exit or roll

Charge each leg and any overlap. Include the risk of holding the old contract while seeking the new fill.

Thresholds before outcomes

Calibrate Early and Validate on Later Market States

Use a development block to set data cleaning, binning and market-quality thresholds. Lock the full policy, then assess later chronological data that includes ordinary sessions, announcements, holidays and roll windows.

Manifest

Data source, schema, checksum, gaps, corrections, clocks and expiries.

Calibrate

Limited bins, order tasks, distributions and thresholds in early data.

Lock

Freeze eligibility, costs, gates and permitted order actions.

Validate

Report all later states, failures, tails and threshold breaches.

Research boundary

No original 6N liquidity study, order-book sample, spread distribution, fill ledger or preferred trading window is reported on this page. It is a protocol, not a finding. Numerical thresholds must come from a separately documented study and the trader’s own size and risk limits.

Order gate

Go, Reduce, Wait or Reject

Evaluate all required fields at the intended decision time. A missing feed field is a rejection, not permission to infer good liquidity from a recent candle.

GoSpread, depth, data integrity, event state and stressed execution cost all pass for the declared order.

ReduceA smaller integer size passes every gate; the original size does not. Recompute risk and economics.

WaitTemporary event, roll or book condition fails now, and the strategy permits delay without changing its premise.

Reject the order when any hard gate fails

Data: Missing, stale, crossed, incomplete or unmapped contract evidence.
Market: Spread, depth or stressed shortfall breaches the frozen limit.
State: Unapproved event, holiday, expiry or roll condition is active.
Execution: Required size cannot be modeled with the available book evidence.
Risk: Stop, gap stress or total loss exceeds the independent budget.
Economics: Expected edge is not positive after conservative non-overlapping costs.

Sources, methods and editorial disclosure — reviewed August 20, 2026

CME Rulebook Chapter 258: New Zealand Dollar/U.S. Dollar futures for dated-contract identity and rule context.
CME DataMine historical-data catalog and CME Market by Order FAQ for distinctions among settlements, trades, books and order-level data. No dataset was purchased or analyzed for this article.
CME Group trading-hours and holiday schedules for versioned venue-state controls.
Reserve Bank of New Zealand release guidance and RBNZ decision-date schedule for official event timing.
Stats NZ official release calendar for New Zealand statistical-event controls.

Sources and methods were reviewed August 20, 2026. This unsponsored article distinguishes exchange facts, market-quality measures, hypotheses about conditions, inferences from a future study and order-level applications. It reports no original liquidity finding.