Objective price structure · 6A
6A Chart Patterns: Define, Test, Then Trade
A pattern name is not a strategy. "Flag," "triangle" and "double top" are visual labels until you define the points, tolerances, confirmation, entry, invalidation and time limit. If two people cannot mark the same events without seeing the right edge, there is nothing to test yet.
- Detection
- Causal
- Parameters
- Frozen
- Sample
- Complete
- Result
- Net of costs
candidate→confirmed→failed or resolvedEvery candidate stays in the dataset, including the ugly ones that hindsight wants to erase.
No universal patternNo claimed win rate
Direct answer
No 6A Pattern "Works" Until a Frozen Rule Survives New Data
Price structure can organize decisions. It can define where a premise becomes invalid, distinguish compression from expansion and create repeatable event labels. None of that proves positive expectancy. Performance depends on market context, contract selection, bar construction, execution and exits. A clean drawing is a hypothesis; a timestamped, costed, out-of-sample result is evidence.
Useful role
Compress observable information
A rule can summarize swing direction, range boundaries, volatility contraction and boundary tests. It gives the trade an explicit state and a point where the thesis is wrong.
Dangerous role
Explain every chart after the move
If boundaries move, pivots repaint or confirmation changes with the outcome, the method cannot fail. That also means it cannot be validated.
Standard 6A represents 100,000 AUD and its outright Globex minimum increment is 0.00005, worth $5. Express tolerances and stops in current ticks, not vague pips. A 0.0001 pip is two ticks and $10 per standard contract.
Objective taxonomy
Classify the Information, Not the Artwork
Most familiar patterns combine a small set of measurable features. Start there and add the market-specific context later.
| Structure family | Measurable ingredients | Possible state | Subjective trap |
|---|---|---|---|
| Trend continuation | Directional impulse, bounded retracement, reduced range or slope | Forming, broken, accepted, failed | Calling any pause a flag after continuation |
| Compression | Contracting high-low span, converging boundaries, falling realized range | Compressing, expanding, expired | Redrawing a triangle around future bars |
| Repeated boundary | Predefined touch count and price tolerance at support or resistance | Holding, breached, reclaimed | Treating near-touches differently by outcome |
| Reversal candidate | Prior directional move, swing failure or boundary rejection, structural break | Candidate, confirmed, invalidated | Equating two peaks with a completed reversal |
| Range | Upper and lower bounds, duration, excursion tolerance | Balanced, testing, accepted outside, failed break | Expanding the range whenever price escapes |
A breakout deserves its own entry and failure workflow; use 6A Breakouts. A reversal using volume needs venue-specific definitions; use 6A Volume Reversals.
Specification
Write the Rule as a State Machine
The detector must know only what was available at that timestamp. If a swing high requires three bars on the right, it becomes observable three bars later, not at the high itself.
Freeze data
Specify venue, symbol, actual or continuous contract, bar source, interval, session template and treatment of missing or duplicate observations.
Define geometry
Write pivot strength, minimum movement, boundary-fitting method, touch tolerance, maximum duration and whether wicks or closes count.
Define states
Separate candidate, complete, breakout, acceptance, failure and expiry. Record the first timestamp each state becomes knowable.
Define trade
Specify entry order, fill model, stop, target or exit, time stop, event filter, position size and what cancels an unfilled order.
Example definition, not a recommendation
A compression candidate might require a falling rolling high-low range plus at least two causally confirmed pivots on each boundary within a fixed lookback. Confirmation might require a close outside by a set number of current ticks. Those values are placeholders: pre-register a small parameter family and validate it rather than tuning until the chart looks good.
Store the detector version with every signal. If a later code change alters historical labels, rerun the full development sample and keep the earlier result. Silent label changes create a moving denominator and make an honest comparison impossible. A small versioned event table is more valuable than hundreds of annotated screenshots.
- Clock
- Decision timestamp
- Output
- Every candidate
- Expiry
- Predeclared
Validation gate
Test the Detector Before Testing the Money
A strategy can appear profitable because its pattern labels are wrong. Audit the labels first.
Have a second implementation or blinded reviewer mark a sample. Investigate every disagreement.
Replay bar by bar and confirm that pivots, boundaries and states appear only when knowable.
Keep expired, overlapping and failed candidates. Do not count only textbook survivors.
Choose definitions and parameters on earlier data, freeze them, then evaluate later untouched periods.
Use bid-ask and slippage assumptions appropriate to order type, time and event conditions.
Vary nearby parameters, contract-roll choices and session definitions. A one-setting spike is fragile.
| Metric | Why it matters | Required context |
|---|---|---|
| Event count | Shows how much evidence exists | Unique and overlapping candidates disclosed |
| Expectancy after costs | Combines wins and losses in money or risk units | Fill and fee assumptions |
| Drawdown | Shows path risk hidden by average return | Position sizing and order sequence |
| Parameter stability | Tests dependence on one tuned value | Predeclared sensitivity grid |
| Out-of-sample result | Tests generalization after the rule is frozen | Untouched dates and no retuning |
Failure cases
Hindsight Can Manufacture a Beautiful Pattern Library
The biggest threat is not choosing the wrong shape. It is letting the outcome decide which shapes existed.
Future bars change earlier swing labels or the time those labels become available.
Lines move after a breakout so the winning touch looks exact.
Only resolved textbook examples reach the screenshot folder.
A continuous-series stitch creates a false gap, range or boundary break.
Dozens of shapes, sessions and exits are tried, but only the winner is reported.
Most profit comes from one volatility burst or macro episode.
No primary exchange or regulator source establishes that a named 6A chart pattern has a universal win rate. Any such figure needs the exact code, dated sample, data provenance, complete event set, costs and untouched validation results. Without those, the honest answer is unknown.
Frequently asked questions
6A Chart Patterns: Quick Answers
Which chart pattern works best in 6A?
There is no defensible universal winner without a dated, reproducible study. A pattern's result depends on its exact definition, bar construction, session, volatility, trend context, entry, exit and costs. Compare frozen rules on an untouched sample instead of ranking pattern names from memory.
How do I define a 6A chart pattern objectively?
Translate every visual term into measurable conditions: data source, bar interval, swing algorithm, number of touches, tolerance, lookback, breakout rule, invalidation and expiry. The definition should let two independent implementations mark the same events without seeing future bars.
Is a chart pattern complete before the breakout?
Only if your written definition says so. A forming structure, a confirmed close beyond a boundary and post-break acceptance are different states with different information. Label them separately and never backfill a breakout condition into an earlier entry.
Should volume confirm a 6A chart pattern?
Volume can be a predeclared contextual feature, but it does not reveal participant motive or guarantee confirmation. Use venue-specific relative volume aligned to the same contract and time bucket, then test whether it improves out-of-sample results after costs.
Why do chart patterns look better in hindsight?
Future bars make swing points and clean boundaries obvious, failed candidates disappear from memory, and continuous charts can hide roll effects. Prevent that bias with causal detection, complete event logs, fixed parameters and screenshots or signals saved at decision time.
Sources, method and editorial disclosure
- CME Group FX Product Guide 2026 for current 6A size, quote, tick and physical-settlement terms.
- CME Rulebook Chapter 255: Australian Dollar Futures for authoritative contract rules.
- CME Globex Notice 20201116 for the 6A outright tick change to 0.00005.
- CME Group daily settlements for exchange settlement data access.
- CFTC customer advisory on exchange-traded currency futures and OTC forex distinctions.
This article is a research protocol, not a pattern-performance study. No named pattern, parameter, win rate or return is endorsed. Example definitions are illustrative and must be frozen, independently implemented and validated on licensed data with realistic costs. Sources and contract facts were reviewed August 13, 2026.