Returns · rolling windows · regimes

6A and Gold Correlation: A Testing Framework

Putting a 6A chart beside gold does not measure correlation. The clocks may differ, the futures may roll on different days, and both USD-quoted prices can react to the same dollar shock. Align the intervals, use returns, publish the window and test whether the result survives regimes and shared-dollar controls.

Coefficient here
Not reported
Main input
Aligned returns
Shared factor
U.S. dollar
Causality
Not implied
Correlation auditAlign before calculate
1

Series

Dated contracts

2

Clock

Same endpoints

3

Return

Same interval

4

Control

Regime and USD

No lead claimNo causal shortcut

Direct answer

The Relationship Is Measurable, Conditional and Not Automatically Tradeable

Calculate correlation from synchronized changes in 6A and a defined gold benchmark. Report the sample dates, interval, roll method, estimator, coefficient, confidence interval and observation count. Then repeat the estimate across market regimes and after controlling for common factors. Until that work is done, “gold up means 6A up” is a hypothesis, not evidence.

What correlation answers

Whether returns co-moved in a stated sample

A positive value means the aligned returns tended to share a sign; a negative value means they tended to oppose. The magnitude summarizes linear association and can change when the window changes.

What it does not answer

Cause, lead, execution or future stability

Common U.S. dollar news can move both. A coefficient does not identify the transmission channel, which market reacted first or whether the gross association can clear costs.

Contract anchor

Standard 6A represents 100,000 Australian dollars and is quoted in U.S. dollars per AUD. Its current Globex outright tick is 0.00005, worth $5. A 0.0001 pip is two ticks, or $10. Final settlement is physical. Gold futures use different units, ticks and delivery rules, so compare percentage or log returns rather than raw point changes.

Instrument choice

Define “6A” and “Gold” Precisely

A valid test starts with observable instruments. For futures-to-futures research, pair dated 6A contracts with dated COMEX Gold futures and publish the roll rule for each. For macro robustness, compare an official AUD/USD series with the LBMA Gold Price, but acknowledge that their fixing and observation times do not necessarily match an exchange settlement.

ChoiceUseRequired disclosureMain risk
Dated 6A and GC contractsExecutable futures relationshipExpiries, timestamp, roll and settlement fieldsIndependent rolls and liquidity migration
Documented continuous futuresLong history and easier researchMapping and price-adjustment methodStitch artifacts or look-ahead
RBA AUD/USD and LBMA goldOfficial macro robustness checkObservation times, holidays and missing-date rulesNon-synchronous observations
Intraday futures barsEvent and lead-lag researchUTC endpoints, sampling and data cleaningMicrostructure noise and common-news timing

CME describes Gold futures as a 100-troy-ounce contract and provides its own contract and trading specifications. That is useful context, not permission to divide a 6A point change by a GC point change. Standardize the changes first.

Core calculation

Use Synchronized Returns, Not Two Trending Price Lines

For each instrument, calculate the log return as the natural logarithm of the current aligned price divided by the prior aligned price, or use a consistently defined simple return. Correlate those two return series. Do not correlate the price levels unless the statistical design explicitly addresses non-stationarity and a defensible long-run relationship.

Same endpoints

Sample both markets at the same UTC timestamps. If using official settlements with different methodologies, disclose the mismatch and rerun with synchronized bars.

Same interval

Daily with daily, hourly with hourly. Do not pair a daily gold move with one 6A session unless that asymmetric hypothesis was predeclared.

Same calendar

Join only observations with valid data for both assets. Explain holidays, missing values and stale prices rather than forward-filling silently.

Documented rolls

Roll 6A and gold independently according to observable rules. Flag transition observations and test whether excluding them changes the coefficient.

Uncertainty

Publish count and interval estimates. Serial dependence and volatility clustering can make naive significance claims too confident.

Costs separate

Correlation describes returns, not a spread trade. Any execution rule needs its own entries, exits, hedge ratio, commissions and slippage.

One full-sample coefficient compresses every environment into one number. Add a rolling estimate only after selecting the window from the decision horizon. A twenty-observation window reacts quickly and noisily. A multiyear window is smoother and slower. Show sensitivity rather than choosing the window with the prettiest line.

Conditional analysis

Control the Shared Dollar and Split the Economic Regimes

Both quoted series can contain a U.S. dollar channel. 6A is USD per AUD; gold is commonly observed in USD per ounce. A broad dollar move can therefore generate co-movement without gold transmitting anything to the Australian dollar. Use the Federal Reserve broad dollar index, defined and lagged without look-ahead, as one control or residualize both returns against the same dollar factor.

RegimeObservable labelWhy the relationship may change
Broad-dollar movementFederal Reserve dollar index returnA common denominator can dominate both USD quotes
Real-yield movementU.S. Treasury real-yield changeGold and rate-sensitive FX can react through different channels
Australian relative ratesPredefined Australian-minus-U.S. rate measure6A can move on bilateral policy repricing unrelated to gold
Commodity breadthRBA commodity price index returnGold may diverge from iron ore, energy and the broader export basket
Risk stressPredeclared volatility or credit measureGold's defensive demand and AUD risk sensitivity can conflict
Volatility stateFrozen realized-volatility thresholdCorrelation estimates become unstable in changing variance

The RBA's own research is a warning against permanent commodity stories: relationships among the Australian dollar, commodity prices and other drivers vary through time. Use the gold transmission guide for economic mechanisms. Keep this page's measurement separate from that causal narrative.

Reproducible workflow

Build the Test in Eight Steps

This page reports no original 6A-gold coefficient. The workflow is the evidence standard for calculating one.

1

Write the question

Define contemporaneous association, event response or lead-lag research before choosing data and lags.

2

Select instruments

Name the exact futures expiries or official series and retain raw observations.

3

Align timestamps

Convert to UTC, match endpoints and document holiday, closure and missing-data rules.

4

Construct returns

Apply the same return formula and audit outliers, stale marks and roll observations.

5

Estimate association

Report full-sample and predeclared rolling correlations with counts and uncertainty.

6

Add controls

Test the shared-dollar factor, real yields, relative rates, commodity breadth and risk state.

7

Validate forward

Freeze windows and thresholds, then evaluate unseen dates without retuning after every failure.

8

Test the trade separately

If a rule remains, specify orders, hedge ratio, costs, latency, margin and loss limits. Correlation is not a backtest.

Failure modes

Correlation Fails Fast When the Dataset Is Sloppy

Most strong-looking charts become less impressive after the clocks, rolls and common factors are handled correctly.

Price-level illusion

Two trending series look connected even when their returns are not.

Timestamp mismatch

One market includes information that the other observation had not yet seen.

Shared-dollar confound

A common USD shock is mislabeled as a direct gold-to-AUD mechanism.

Window shopping

The analyst tests many lengths and publishes only the strongest coefficient.

Roll contamination

Independent expiry transitions create artificial return pairs.

Lead-lag fantasy

Repeated lags find a winner that vanishes after common-news and search corrections.

Blunt conclusion

If the coefficient disappears when prices become returns, clocks align or the broad dollar is controlled, that is the result. Do not rescue the thesis by changing the window after the test.

Frequently asked questions

6A and Gold Correlation Questions

Are 6A futures positively correlated with gold?

The sign and strength can change by sample, return interval and macro regime. No original coefficient is reported on this page, so the defensible answer is to calculate synchronized 6A and gold returns and publish the estimate with uncertainty.

Should I correlate 6A and gold price levels or returns?

Use aligned returns for the main test. Trending price levels can create a persuasive but spurious relationship, while log or simple returns compare changes over the same interval.

What rolling window is best for 6A and gold correlation?

There is no universal best window. Choose a horizon that matches the decision, predeclare it, and show sensitivity across reasonable alternatives because short windows are noisy and long windows can hide regime changes.

Does gold lead 6A futures?

Contemporaneous correlation cannot establish a lead. A lead-lag claim requires synchronized granular data, predefined lags, controls for common news, multiple-testing correction and validation on unseen observations.

Why can the 6A-gold correlation break down?

The assets can respond differently to U.S. dollar moves, real yields, Australian rate expectations, commodity-specific supply, risk sentiment and contract-roll effects. A changing economic regime can overwhelm any historical average.

Sources, method and editorial disclosure

No original coefficient, rolling correlation, causal estimate, lead-lag result or backtested trade is reported here. Formulas, controls and windows are research methods, not findings. Sources and time-sensitive facts were reviewed August 13, 2026. This is original, unsponsored editorial analysis.