Skip to main content

Benchmark facts · observable books · execution guardrails

6S Around the London 4 p.m. Fix: A Liquidity Study

A 6S move near 4 p.m. London time does not reveal whether an asset manager bought francs, a dealer hedged a benchmark order, a U.S. release moved the dollar or ordinary liquidity shifted. Client instructions are generally absent from public futures data. The defensible task is narrower: anchor the official benchmark time, measure the 6S book around it and leave unobserved flow unlabelled.

Anchor
4 p.m. London
Observed
6S trades + quotes
Unobserved
Client intent
Routine reversal
Not claimed

Administrator evidence first

Separate the Benchmark From a Futures Pattern

FTSE Russell’s WMR benchmark page identifies its closing spot rates as fixed daily Monday through Friday at 4:00 p.m. London time and provides the current methodology. That is an authoritative timing and administration fact. It does not establish a predictable 6S direction, a client-flow imbalance or a post-fix reversal.

1

Established fact

A named WMR closing spot benchmark exists at the administrator’s stated London time under a published methodology.

2

Observable hypothesis

6S market-quality measures may differ near that anchor relative to matched adjacent and control windows.

3

Unidentified story

A particular 6S move was caused by client benchmark flow or dealer inventory.

4

Trading application

An execution schedule improves realized cost after delay, fees, misses and regime changes.

Do not call every 4 p.m. price a fix print.

WMR is a spot FX benchmark under the administrator’s methodology. CME 6S is a dated CHF/USD futures contract. A futures observation around the anchor is research evidence, not the benchmark value itself.

London local time moves in UTC

Construct Symmetric and Asymmetric Benchmark Windows

Resolve 4 p.m. Europe/London with the historical timezone database, then convert it to UTC and the exchange clock for each date. Register one primary window and only a few sensitivity windows before outcomes are inspected.

Pre[T−30m, T−5m)Core[T−5m, T+5m]Post(T+5m, T+30m]
WindowPurposeContamination rule
PreMeasure market quality approaching the anchorTag every scheduled release and unusual market closure
CoreMeasure the narrow benchmark-proximate stateDo not interpret direction as client flow
PostMeasure persistence, normalization or further repricingNo automatic “reversal” label
Adjacent placeboRepeat the same geometry at non-benchmark timesMatched by weekday, month state and event load
Control dayCompare like calendar and market statesSelected using past-only variables

Study the book, not a candle shape

Measure Spread, Depth, Slippage and Fill Quality

A close-to-close futures bar cannot reconstruct what a marketable order would have paid or whether a resting limit would have filled. Match the claim to the data layer and order-size rule.

Quoted

Inside spread

Time-weighted and arrival-time bid/ask spread, quote age and locked/crossed-market treatment.

Capacity

Displayed depth

Quantity at the inside and cumulative depth through declared price bands for fixed order sizes.

Traded

Print distribution

Volume, trade count, size distribution and directional classification with method uncertainty.

Experienced

Order outcomes

Arrival slippage, implementation shortfall, fill rate, time to fill, partials and cancellations.

side × (fill − decision) × 125,000 × filled contractsUSD fees = filled-order shortfall in USD

Use +1 for a buy and −1 for a sell. Record unfilled quantity separately unless a predeclared miss model converts it to a finite USD opportunity cost. Never add a raw price difference, a contract count and dollar costs as though they shared one unit.

CME DataMine lists Market by Order and PCAP separately from settlement datasets. The source choice determines whether spread, depth, message sequence or only a reference price can be tested. No CME dataset was purchased or analyzed for this article.

Month-end is not the only rival explanation

Control Calendar, Events, Contracts and Common FX Moves

Report month-end and ordinary dates separately, but do not stop there. A benchmark-window estimate can be driven by U.S. information, holiday-thinned liquidity, roll migration or a broad-dollar shock.

ConfoundRequired controlFailure signal
Month and quarter endPredeclared calendar classification and separate estimatesEffect exists only in one rebalancing subset
Scheduled informationOfficial FOMC, U.S., Swiss and European event timestampsA few release collisions drive the average
London/New York DST mismatchHistorical timezone conversion and explicit mismatch stateClock-fixed result moves with the offset
Contract migrationDated-contract mapping, volume/open-interest roll ruleSpread change occurs only near the splice
Broad-dollar movementIndependent timestamp-compatible USD factor6S direction vanishes after common FX control
Search freedomRegistered windows, metrics and sample partitionsOnly the best undisclosed window survives

Protocol question—not a result

Does executable 6S market quality around the declared WMR 4 p.m. London anchor differ from matched adjacent and control windows, and can any difference survive later data under a fixed order policy?

Anchor
Administrator time
Primary outcome
Market quality
Flow attribution
Not identified

Observation before participation

Use Execution Guardrails, Not a Directional Fix Trade

The study can support a scheduling or sizing decision only if the same observable state exists live and the order policy passed a chronological holdout. It cannot turn an unobserved flow story into a buy-or-sell signal.

Pre-order decision

Trade only the state that was tested

Verify the correct dated 6S contract, current spread, displayed depth, quote age, event status and maximum order size. If any required field is absent, the valid output is wait or reject.

  • Go: all live gates pass and the order size, urgency and venue match the validated policy.
  • Reduce: capacity is below the full-size gate but inside a validated smaller-size band.
  • Wait: an event collision, unstable book or transient quote condition makes the state ineligible.
  • Reject: the data, holdout or cost advantage is incomplete, unstable or economically immaterial.
  • No routine reversal assumption. Post-window direction must be measured and may be continuation, reversal or noise.
  • No client-flow label. Public 6S prints do not identify beneficial owner or benchmark instruction.
  • No zero-cost backtest. Include spread, impact, fees, latency, partials and missed orders.
  • No timeless window. Re-estimate after methodology, market-structure or participation changes.

Write down the failure

Benchmark-Window Validation Ledger

Record every result, including states where the benchmark window is indistinguishable from controls.

Accept, limit, or abandon

Definition
Benchmark version, timezone, windows, contract map, data layer, order size and metrics.
Controls
Month-end, events, holidays, DST state, roll state, common dollar factor and placebos.
Validation
Chronological holdout, uncertainty, all tried windows and conservative cost model.
Accept
Market-quality improvement is stable, live-observable and useful for the declared order.
Limit
Use only in a pre-identifiable state with adequate independent observations.
Abandon
Controls erase it, costs consume it, fill quality fails, or later data do not reproduce it.
Research status as of August 21, 2026

No original 6S London-fix dataset, spread estimate, depth result, reversal statistic, client-flow finding or trading performance is reported here. No original result is reported. The benchmark facts are sourced; the market-quality study remains a protocol.

Sources, methods and editorial disclosure — reviewed August 21, 2026

Sources and methods were reviewed August 21, 2026. This unsponsored article distinguishes benchmark facts, observable futures evidence, unobserved flow hypotheses and execution application. It reports no original market result.