Width · capacity · replenishment · impact
NQ Liquidity Windows: Build a Live Market-Quality Map
NQ volume can rise while displayed depth falls. That is more activity, but it may be less immediately available capacity for a marketable order. A useful liquidity window starts and ends when a declared stack of market-quality conditions changes—not merely when the clock reaches a popular hour.
- Spread
- Time weighted
- Depth
- Size conditioned
- Replenishment
- After depletion
- Volume alone
- Insufficient
Measure quoted, traded and experienced liquidity separately
Assemble the Live Metric Stack
Use source-sequenced NQ trades and book updates when the claim needs intrawindow execution quality. Bar volume can describe activity; it cannot reconstruct the spread, near-touch capacity or an order’s queue.
| Layer | Live measures | Calibration target | Hard boundary |
|---|---|---|---|
| Width | Inside spread, quote age, locked/crossed share | Time-weighted distribution by state | Message count is not time weighting |
| Capacity | Depth through K ticks, sweep cost for Q | Median and adverse tail by side/size | Displayed size can cancel |
| Durability | Executed fraction, cancellation, refill time | Post-depletion replenishment distribution | Public data do not reveal intent |
| Impact | Signed response after aggressive quantity | Size- and state-conditioned price response | Correlation is not actor attribution |
| Experienced | Fill rate, delay, shortfall, partials, rejects | Actual order-policy outcomes | Historical book is not your latency |
CME’s data catalog distinguishes top-of-book, depth and market-by-order products, while the Equity Execution Statistics FAQ separates width, depth, fill and slippage measures.
A window is a state interval, not a favorite hour
Construct Start, Persistence and End Rules
Calculate features in rolling windows using only completed source events. Require persistence to reduce one-message flips, but include the delay that persistence creates. The state becomes actionable only after computation and system latency.
Validate
Resolve dated contract, sequence, timestamp, schedule and book integrity.
Measure
Compute width, capacity, durability and impact over the registered lookback.
Enter
Open a window after all gates persist for p completed observations.
Monitor
Recompute with the same rule and compare realized fills to capacity.
Exit
Close after a hard breach, q failed observations or an event reset.
Store the full window record
Archive start decision time, source cutoff, contract, session/event state, lookback, every component value, gate version, order-size class, end time and end reason. A window that starts after a clock event may recur, but the clock event is only an overlay; the measured stack creates eligibility.
Estimate thresholds by task. A passive order may require replenishment and acceptable adverse-selection evidence. An urgent hedge may accept a wider spread if tail sweep cost remains inside authority. Never collapse those policies into one universal “liquid” flag.
Scheduled flow can invalidate the recent window
Overlay Events, Session Transitions and Roll State
Attach known calendar and contract events before live scoring. A window can be terminated at a scheduled timestamp and requalified afterward rather than assuming the pre-event distribution still applies.
Macro event
Known time, new distribution
BLS, BEA and FOMC calendars supply official timing. Close or suspend the routine state under the registered policy.
Cash transition
Constituents enter or leave core
Tag the cash open and close rather than blending the transition into the preceding state.
Quarterly roll
Capacity migrates by contract
Monitor dated-contract volume and depth. Do not carry front-contract thresholds into a new expiry without requalification.
Holiday/schedule
Clock exceptions
Use current CME calendars. Early closes and maintenance changes create their own state or exclusion.
The Nasdaq-100 methodology defines the index exposure that NQ wraps, but index concentration is not itself a live liquidity measure. Keep index-event hypotheses separate from the exchange-book evidence used to authorize an order.
Authoritative overlays come from the CME schedule, CME roll analyzer guide, BLS, BEA and Federal Reserve.
Convert the stack into four explicit outcomes
Go, Reduce, Wait or Reject
Score the declared order, not NQ in the abstract. Use the current side, quantity, urgency, route and latency. The decision rule should fail closed when any required field is absent.
Go
All gates represented
Current values sit inside validated support, modeled capacity covers Q and recent real fills remain calibrated.
Reduce
Smaller Q passes
A lower preregistered whole-contract tier clears depth, impact and risk limits.
Wait
State may requalify
Event recovery, spread normalization or replenishment has not persisted long enough.
Reject
Unscorable or expired
Sequence loss, stale quotes, unknown contract, out-of-distribution state or platform failure blocks execution.
It can change before an order arrives. Use conservative sweep and cancellation assumptions, then calibrate against acknowledgement and fill records. A dashboard green light is not a guaranteed price.
Every live map needs a kill switch
Expire the Window on Drift, Mismatch or Missing Evidence
Use both immediate and statistical expiry. A sequence gap or schedule transition can terminate eligibility instantly. Sustained cost or distribution drift can invalidate the calibration until a new frozen model passes forward validation.
Integrity expiry
Source sequence, timestamp, contract or market state cannot be verified.
Immediate rejectDistribution expiry
Spread, depth, replenishment or impact repeatedly leave registered bounds.
Rebuild calibrationExecution expiry
Realized shortfall exceeds the conservative model by the preregistered amount.
Disable orders- Thresholds and persistence rules are learned before the forward period.
- Every live feature has a source timestamp and freshness limit.
- Book reconstruction fails closed on sequence loss.
- Event, cash-transition, roll and holiday overlays can reset the state.
- Go/reduce/wait/reject outcomes are stored, including no orders.
- Actual fills calibrate the model by size, side, urgency and latency.
No original NQ liquidity-window clock, threshold, price-impact estimate or execution edge is reported. This page defines a live construction, permission and expiry protocol.
Sources, methods and editorial disclosure — reviewed August 28, 2026
- CME E-mini Nasdaq-100 product page for NQ context.
- CME futures and options data catalog for top-of-book, depth and market-by-order distinctions.
- CME Equity Execution Statistics FAQ for width, depth, fill and slippage categories.
- CME current trading-hours calendar and CME roll analyzer guide for schedule and migration states.
- Nasdaq-100 Index Methodology for the underlying index identity.
Sources were reviewed August 28, 2026. The metric stack is an evaluation protocol awaiting measured data; its thresholds and expiry rules are not empirical results.