Operational definitions · false-break controls · sealed validation
6S Volatility Compression: Define and Test Breakouts
A narrow high–low range, low realized variation and a tight Bollinger-style band can label three different 6S samples “compressed.” One can be quiet because quotes are stale, another because liquidity is deep, and another because a scheduled announcement has not arrived. Until the definition, data and clock are fixed, “compression leads to breakout” is not one testable claim.
- Compression
- Operational state
- Breakout
- Timestamped trigger
- Outcome
- Frozen horizon
- Original results
- None reported
The state exists only after the rule
Register an Operational Compression Definition
Choose a primary estimator, lookback, threshold, normalization and minimum-data policy before viewing breakout outcomes. Alternate definitions belong in a limited robustness set and remain separate results.
Candidate A
Range compression
Completed high–low range over a fixed window, scaled by a past-only reference such as price or earlier range. Sensitive to bad ticks and session boundaries.
Candidate B
Return compression
Realized variation from completed intraday returns at a declared sampling interval. Sensitive to microstructure noise and missing updates.
Candidate C
Band compression
Width of a fully specified past-only envelope relative to price or a training-period scale. Sensitive to lookback and estimator choice.
| Field | Required declaration | Failure condition |
|---|---|---|
| Data | Dated 6S contract, price field, timestamps, cleaning and roll policy | Continuous symbol lacks expiry mapping |
| Clock | Session endpoints, timezone, holiday and missing-bar treatment | Vendor defaults are undocumented |
| Estimator | Exact formula, sampling, lookback and scaling | Definition changes after outcomes |
| Threshold | Training-only quantile or fixed rule with minimum history | Full-sample percentile leaks later data |
| Duration | Minimum and maximum number of completed eligible intervals | State begins or ends by visual judgment |
A repeated last price or missing quote updates can mechanically lower return-based volatility. Require valid observation counts, quote-age limits and sequence continuity before an interval can qualify.
Leaving the box is not yet an executable order
Define the Candidate Breakout Trigger and Arrival Path
The trigger must occur on information available at a timestamp and the simulated order must arrive later. Register direction, buffer, confirmation, entry type, invalidation and re-entry policy. Values are test parameters, not recommendations.
Eligible
Data, event, liquidity and contract gates pass.
Compressed
Primary estimator remains below its frozen threshold.
Released
Completed price or variation crosses the registered boundary.
Confirmed
Optional past-only confirmation occurs within its time limit.
Ordered
Order arrives after processing and route delay.
Resolved
Fill, miss, false break, exit and costs are logged.
Candidate continuation
Trade in the break direction
Requires a frozen boundary, buffer and later entry. It must compete with a direction-neutral timing rule and no trade.
Candidate failure
Fade a false break
Requires an independent false-break definition and later trigger. It cannot use the future return to decide the original break was false.
A breakout needs an endpoint
Freeze Outcome Horizons and Failure Labels
A move can continue for five minutes and reverse by the session close. Report the path across a small registered horizon set rather than selecting the endpoint that makes the setup look successful.
| Outcome | Operational definition | Why it is separate |
|---|---|---|
| Continuation | Signed return from executable entry to a fixed later horizon | Measures direction, not maximum excursion |
| Adverse excursion | Worst signed path before the fixed exit | Exposes risk hidden by endpoint return |
| Favorable excursion | Best signed path before the fixed exit | Does not assume a fill at the extreme |
| False break | Registered boundary cross followed by a registered return inside before deadline | Must be defined without later discretionary relabeling |
| No resolution | Neither success nor invalidation before the time stop | Prevents open-ended waiting |
| Net order result | Realized fills and exits less spread, slippage, fees and roll cost | Connects path evidence to implementation |
outcome(h) = signed return after executable entry|h belongs to a preregistered horizon setCompression often sits before information
Control Events, Time of Day, Liquidity and Rolls
A quiet interval before an FOMC or SNB decision is different from an ordinary low-variation interval. Separate these states before estimating whether the compression label adds information.
Events
Scheduled information
Tag official FOMC, SNB and major data timestamps. Estimate event and non-event samples separately.
Clock
Intraday seasonality
Compare within matched UTC bins and historical DST states so quiet hours do not create the entire effect.
Book
Market quality
Require spread, depth, quote-age and update gates where the claim depends on executable liquidity.
Contract
Roll lifecycle
Preserve dated expiries and distinguish active trading from migration and expiry-proximate intervals.
Information arrival
A scheduled or unscheduled release can generate the move independently of the prior range.
Liquidity transition
Session handovers and reopening participation can alter both volatility and fill quality.
Common dollar factor
A broad USD move can drive CHF/USD while the compression state adds no incremental information.
Measurement artifact
Stale quotes, missing bars or a roll splice can manufacture both a quiet state and apparent break.
Reasonable alternatives should not reverse the decision
Stress Definitions Without Data-Mining a Rescue
Register the robustness family before the sealed sample. The goal is to learn whether a practical conclusion survives, not to locate one surviving parameter after dozens fail.
Definition
Does the state survive?
- Adjacent lookbacks and threshold bands
- Range versus return-based sensitivity
- Strict stale-data and bad-tick policies
Outcome
Does the path survive?
- Adjacent non-overlapping horizons
- Continuation, false break and no-resolution labels
- Entry delays and adverse-selection stress
Sample
Does the evidence survive?
- Chronological blocks and leave-one-event-out
- Event, clock, volatility and roll states
- Final holdout never used for selection
- Count the full family. Every estimator, lookback, threshold, buffer and horizon is part of the search.
- Use causal fills. Marketable orders cross the historical book; limit orders need queue-aware or conservative rules.
- Stress costs. Wider spreads, lower depth, latency and missed orders must not be hidden.
- Keep the null. If compression adds no information beyond event and time controls, reject it.
One row per registered generation
Compression-and-Breakout Validation Ledger
Never overwrite a failed version. The ledger links each conclusion to a definition, dataset, sample cutoff and executable policy.
| Record | Required contents | Decision |
|---|---|---|
| Definition | Estimator, threshold, duration, trigger, entry, exits, horizons and code hash | Reject if any field is discretionary |
| Data | Source, contract map, timestamps, corrections, gaps, events and costs | Reject if causal reconstruction fails |
| Search | All candidates, parameter choices, controls and robustness variants | Adjust inference for the full family |
| Evidence | Event count, outcomes, uncertainty, blocks, fills, misses and capacity | Limit if state-specific and pre-identifiable |
| Holdout | Single sealed result versus registered baselines and cost stress | Accept only if useful and stable |
| Failure | Sign reversal, unstable state, control dependence, no net value or missing data | Abandon; do not rename |
No original 6S compression sample, breakout rate, false-break rate, parameter, confidence interval or trading performance is reported here. No original result is reported. This page defines a protocol whose valid outcome may be rejection.
Sources, methods and editorial disclosure — reviewed August 21, 2026
- CME Rulebook Chapter 254: Swiss Franc/U.S. Dollar futures for dated-contract identity and exchange rules.
- CME Swiss Franc futures contract page for current product context; full mechanics are centralized in the 6S contract specification guide.
- CME DataMine historical-data catalog for distinctions among settlements, Market by Order and PCAP data. No data was purchased or analyzed for this article.
- Federal Reserve FOMC calendars and official materials for U.S. policy-event controls.
- Swiss National Bank time schedule for official Swiss policy and publication scheduling context.
- NIST/SEMATECH time-series analysis guidance for time dependence, diagnostics and validation context.
Sources and methods were reviewed August 21, 2026. This unsponsored article distinguishes contract facts, operational definitions, empirical hypotheses and trading application. No original result is reported.