Introduction
Tactical view: sell rallies, not spot. The carry cushion is real, but the next material surprise is more likely to strengthen the yen than to re-accelerate the cross.
Investment conclusion
AUD/JPY is still supported by a 335bp policy-rate gap and a hawkish RBA. That support is why we prefer to wait for a rally rather than short at market. The asymmetry has nevertheless turned: the BOJ is considering a September hike, intervention risk rises as USD/JPY approaches 160, and leveraged-fund positioning still resembles a long-AUD/short-JPY carry expression. These forces can unwind faster than carry accrues.
| Scenario | Probability | 1–3m endpoint | What drives it |
|---|---|---|---|
| Base — orderly normalization | 50% | 109.5 | BOJ tightens or stays credibly hawkish; risk remains broadly stable |
| Bull — carry extends | 25% | 115.5–116.5 | BOJ delays; global risk-on persists; RBA reprices more hawkishly |
| Bear — nonlinear unwind | 25% | 104–106 | BOJ hike/intervention coincides with China or global-risk weakness |
Market snapshot: RBA exchange-rate table; RBA and BOJ policy decisions; Reuters, 13–14 August 2026. Scenario endpoints and probabilities are Mortise estimates, not forecasts from the cited institutions.
What would prove us wrong
• A sustained weekly close above 116.5, especially if accompanied by lower Japanese rate expectations and improving China/commodity breadth.
• A clear BOJ retreat from near-term normalization while the RBA reprices toward additional hikes and volatility remains contained.
The rate gap supports carry; the catalyst gap favors the yen
The relevant question is no longer whether carry is positive—it is whether carry compensates for a policy/intervention discontinuity.
| Central bank | Rate | Latest signal | Near-term read-through |
|---|---|---|---|
| Reserve Bank of Australia | 4.35% | Held on 11 Aug; inflation still too high; policy somewhat restrictive | Hawkish hold limits AUD downside and makes a short-at-market less attractive |
| Bank of Japan | 1.00% | Held 31 Jul; one dissenter preferred 1.25%; core inflation seen above 2% from H2 FY2026 | A September hike is plausible; any acceleration challenges short-yen carry |
Source: RBA, BOJ. A policy-rate spread is not the same as executable spot/forward carry; forward points, basis, hedging costs and roll must be refreshed from a live terminal before execution.
Priced versus underpriced
| Theme | Largely priced | Potentially underpriced |
|---|---|---|
| Relative rates | RBA stays restrictive; BOJ normalizes gradually | A faster BOJ cadence or a hawkish September surprise |
| FX policy | Verbal concern around yen weakness | A second intervention episode near USD/JPY 160, especially around a BOJ shift |
| Positioning | Carry remains attractive in calm markets | The speed of a crowded short-JPY unwind if volatility rises |
| Growth | Australia benefits from resilient nominal exports | China/industrial slowdown hitting AUD as JPY strengthens |
One cross, two macro engines
AUD/JPY falls fastest when the growth-sensitive AUD leg and the funding-sensitive JPY leg move in the same direction.
| Shock | AUD leg | JPY leg | AUD/JPY implication |
|---|---|---|---|
| Global risk-off | AUD/USD usually weakens | USD/JPY often falls as carry is reduced | Strong downside amplifier |
| China / industrial slowdown | Terms-of-trade and growth expectations weaken | JPY may strengthen if global risk also deteriorates | Downside; magnitude depends on breadth |
| BOJ hike / intervention | Little direct AUD effect | JPY strengthens, potentially abruptly | Downside driven by Japan leg |
| RBA hike / hot inflation | AUD strengthens | Limited direct JPY effect | Upside; key thesis risk |
| Commodity supply shock | AUD may rise despite risk-off tone | JPY response can be mixed | Cross signal becomes unreliable |
Australia: a high-beta currency with a policy cushion
• The RBA's 4.35% cash rate and upside inflation language support the AUD and make carry expensive to fight without a catalyst.
• China remains the dominant export destination: A$195.6bn in 2025, versus Japan A$65.1bn, the US A$59.9bn, Korea A$44.2bn and India A$32.2bn. China/commodity breadth remains the cleanest macro confirmation for the AUD leg.
• The relevant bear case is not 'Australia is weak'; it is that growth-sensitive exposure de-rates faster than the RBA can offset through relative rates.
Japan: the funding currency is becoming policy-sensitive
• Japan's 2025 international investment position was ¥1,805.6tn of external assets, ¥1,243.9tn of liabilities and a ¥561.8tn net asset position. This supports the yen's safe-haven narrative, but does not guarantee immediate repatriation in every risk-off episode.
• The nearer-term transmission is mechanical: higher Japanese yields, intervention risk and short covering raise the cost of funding carry in yen.
• The BOJ's July outlook—core CPI clearly above 2% from H2 FY2026 and underlying inflation around target through FY2027—makes normalization a live catalyst rather than a distant tail.
Source: ABS 2025 international trade data; Japan Ministry of Finance 2025 international investment position; RBA; BOJ.
The carry trade is not gone; that is the point
Positioning is a source of upside carry in calm markets and downside convexity when policy or volatility forces deleveraging.
Source: CFTC Traders in Financial Futures. AUD and JPY positions are separate futures markets and are only a proxy for a paired AUD/JPY trade. Leveraged funds: AUD +40,637 contracts; JPY −60,825. Non-commercial net positions were AUD −33,190 and JPY −45,473.
Trade construction
| Element | Level / rule | Rationale |
|---|---|---|
| Spot reference | 112.57 | RBA 4pm AEST rate on 14 Aug 2026 |
| 2026 range | 104.7–114.8 | Public market-data range; refresh before execution |
| Entry | Sell 113.5–114.5 | Use a carry-supported rebound to improve payoff; scale rather than chase |
| Target | 109.5 | Below the first support zone; above the tail-event range |
| Stop / invalidation | Weekly close >116.5 | Signals that carry/risk-on has overwhelmed the policy-risk thesis |
| Tail objective | 104–106 | Requires BOJ/intervention plus broader risk-off confirmation |
| No-chase rule | No new short <111.0 | Payoff deteriorates and short-covering risk rises |
Implementation principles
• Start at 0.5x intended risk in the entry zone; add only if BOJ repricing or multi-asset risk breadth confirms. Define loss in portfolio-risk terms before entry.
• Treat the 335bp policy spread as a directional proxy—not a substitute for live 1M/3M forward points, cross-currency basis, implied volatility and event premium.
• For options, prefer structures with defined downside and limited carry bleed around the BOJ window; refresh skew and vol before choosing strikes.
Risk-off sensitivity is real—but it is not a fair-value model
The regression validates the direction of the amplifier. Its low explanatory power is the reason we embed it inside a broader rates, positioning and catalyst framework.
Source: FRED DEXUSAL, DEXJPUS and VIXCLS. AUD/JPY is constructed from AUD/USD × USD/JPY; 2,886 common observations. The chart uses non-overlapping daily VIX-change buckets.
| dVIX bucket | n | Mean | Median | AUD/JPY down |
|---|---|---|---|---|
| ≤ −3 | 83 | +0.291% | +0.353% | 33.7% |
| (−3, −1] | 480 | +0.282% | +0.286% | 31.0% |
| (−1, 1) | 1,830 | +0.023% | +0.038% | 47.5% |
| [1, 3) | 381 | −0.274% | −0.223% | 63.8% |
| [3, 5) | 73 | −0.551% | −0.571% | 75.3% |
| ≥ 5 | 39 | −1.149% | −0.988% | 82.1% |
Both legs must align for the strongest signal
The cross behaves as an amplifier when AUD/USD and USD/JPY fall together—not as a standalone fear gauge.
Source: FRED; Mortise-defined event windows. Windows are descriptive and pre-specified, not a high-frequency causal identification design.
| Episode | Window return | Max drawdown | VIX peak | Dominant mechanism |
|---|---|---|---|---|
| China shock (2015) | −8.8% | −10.7% | 40.74 | China/growth shock; AUD weakness |
| Volmageddon (2018) | −3.6% | −4.1% | 37.32 | Volatility deleveraging |
| COVID (2020) | −13.6% | −14.9% | 82.69 | Global funding and growth shock |
| Yen unwind (2024) | −14.4% | −14.4% | 38.57 | BOJ repricing plus carry unwind |
| Tariff shock (2025) | −5.1% | −7.8% | 52.33 | Growth and trade uncertainty |
Two-leg decomposition
| Regression leg | Beta per +1 VIX point | Interpretation |
|---|---|---|
| AUD/USD | −0.064% | Risk-sensitive AUD leg weakens |
| USD/JPY | −0.049% | JPY strengthens / funding leg reverses |
| AUD/JPY | −0.113% | Sum of the two log-return legs |
A stable sign, an unstable magnitude
The negative relationship survives every subsample, but position size and conviction should rise only when market breadth confirms.
| Metric | Full | 2015–19 | 2020–23 | 2024–26 YTD |
|---|---|---|---|---|
| Common days | 2,886 | 1,243 | 994 | 649 |
| Correlation | −0.298 | −0.351 | −0.274 | −0.282 |
| Beta (% / VIX pt) | −0.113 | −0.169 | −0.090 | −0.098 |
| HAC t-stat | −7.71 | −4.99 | −6.87 | −2.87 |
| R-squared | 8.9% | 12.3% | 7.5% | 7.9% |
Source: FRED; Mortise calculations. Rolling correlation ranged approximately from −0.52 to −0.04; latest was around −0.23. The relationship is directionally persistent but not a fixed hedge ratio.
Live confirmation dashboard
| Signal | Bearish confirmation | Thesis warning |
|---|---|---|
| BOJ / JGBs | September hike odds or 2y JGB yields rise | BOJ delays and Japan rates fall |
| USD/JPY | Re-test of 160 followed by intervention or reversal | Clean break higher with no policy response |
| AUD leg | AUD/USD, CNH and industrial metals weaken together | AUD/USD holds up despite softer risk |
| Risk breadth | VIX up with credit wider and equities/cyclicals lower | Isolated VIX spike without breadth |
| Positioning | Short-JPY exposure remains large into a catalyst | Carry already materially de-risked |
A catalyst calendar with explicit kill switches
The thesis is tactical and should be exited when the policy or cross-asset evidence changes—not defended as a structural story.
| Window / trigger | What to watch | Expected AUD/JPY effect |
|---|---|---|
| September 2026 BOJ window | Hike decision, vote split, inflation language and future cadence | Hawkish surprise: downside; delay: upside squeeze |
| USD/JPY near 160 | MOF language, intervention prints and liquidity conditions | Intervention can generate abrupt JPY strength |
| Each RBA meeting / CPI release | Whether inflation forces a renewed hiking bias | More RBA hikes: upside / thesis risk |
| China and commodities | CNH, iron ore, equities and industrial cycle breadth | Broad weakness: downside through AUD leg |
| Global volatility | VIX level/change, term structure, credit and equity breadth | Confirmed stress: nonlinear downside |
Kill switches
| Trigger | Action | Reason |
|---|---|---|
| Weekly close above 116.5 | Exit short | Price invalidation overrides narrative |
| BOJ hike odds collapse + RBA reprices hawkishly | Cut / do not add | Relative-rate asymmetry reverses |
| AUD/JPY falls but AUD/USD and risk breadth hold | Re-diagnose | Move may be Japan-only and prone to reversal |
| Forward carry / vol makes payoff unattractive | Use options or stand aside | Spot thesis may not translate into executable return |
| Intervention occurs before entry | Do not chase | Much of the discontinuity may already be realized |
Risk map
• Upside risk: another Australian inflation shock, a broader commodity rally or a decisive BOJ delay can push the cross through recent highs.
• Timing risk: carry and positive roll can erode P&L while the thesis waits for a catalyst; this is why entry discipline matters.
• Model risk: the VIX relationship is bivariate and time-varying. It is evidence of stress transmission, not a stand-alone valuation anchor.
• Execution risk: intervention gaps, thin liquidity and option-volatility repricing can make realized exits materially worse than screen levels.
What the statistics do—and do not—say
The empirical section is reproducible, but its scope is intentionally narrow.
Dataset and construction
• Daily sample: 5 January 2015 to 7 August 2026; 2,886 common trading days after aligning FRED DEXUSAL, DEXJPUS and VIXCLS.
• AUD/JPY is constructed as AUD/USD × USD/JPY. Log returns are first differences of log price; dVIX is the first difference of the VIX index.
• Main specification: rₜ = α + β·dVIXₜ + εₜ. Reported t-statistics use Newey–West standard errors with maximum lag 5.
• Stationarity check: the ADF test cannot reject a unit root in the AUD/JPY level (t = −1.07) but strongly rejects one in returns (t = −21.64). Regressions therefore use returns, not levels.
Timing and robustness
• FRED FX observations are noon New York rates; VIXCLS is the US close. Asian-session events can map across dates. Same-day correlation is −0.298; returnₜ versus dVIXₜ₋₁ is −0.147; the remaining ±2-day correlations are close to zero.
• The full-sample Pearson correlation is −0.298 and Spearman correlation is −0.286. The negative sign is not driven only by a handful of outliers.
• For dVIX ≥ 5, the reported bootstrap resamples the 39 daily observations 10,000 times. Because crisis days cluster, a block bootstrap would be a stricter extension.
• Selected event windows are descriptive stress episodes. They are not an event-identification design and should not be read as causal estimates.
Known limitations and next analytical upgrades
| Limitation | Why it matters | Upgrade before live deployment |
|---|---|---|
| No live forward curve | Policy spread overstates executable carry precision | Add 1M/3M forward points, basis and roll-down |
| No implied-vol surface | Spot target may not map to option value | Add ATM vol, risk reversals and event premium |
| Bivariate VIX model | Omitted rates, commodities and USD factors | Estimate multivariate / regime-switching model |
| Daily timestamps | FX and VIX closes are not synchronized | Use intraday prices around BOJ/RBA events |
| Separate CFTC markets | Positions are only a cross proxy | Add prime-broker/client flow if available |
Primary sources, market data and disclosures
All market levels are as of 14 August 2026 unless otherwise noted.
| Source | Use in this report |
|---|---|
| RBA — Exchange rates | Spot reference and daily AUD exchange rates |
| RBA — Cash rate decision, 11 Aug 2026 | 4.35% cash rate and policy language |
| RBA — Cash rate target overview | Current effective cash-rate target |
| BOJ — Statement on Monetary Policy, 31 Jul 2026 | 1.00% policy rate, vote split and outlook |
| Reuters — BOJ considers September hike | Current market pricing and policy debate |
| Reuters — Yen after intervention | September hike odds and intervention context |
| CFTC — Traders in Financial Futures | Leveraged-fund and non-commercial positioning |
| FRED — DEXUSAL | USD per AUD noon exchange rate |
| FRED — DEXJPUS | JPY per USD noon exchange rate |
| FRED — VIXCLS | Cboe VIX daily close |
| ABS — International trade, calendar year 2025 | Australia exports by destination |
| Japan MOF — International Investment Position 2025 | External assets, liabilities and NIIP |
| BIS Bulletin 124 | Carry positioning and policy-tightening amplification |
| RBA — Drivers of the AUD | Terms of trade, rates and risk channels |
| IMF — The Curious Case of the Yen as a Safe Haven | Safe-haven transmission background |
Disclosures
This report is for informational purposes only and does not constitute investment advice, an offer, a solicitation or a recommendation to transact. Mortise Capital may do or seek to do business with issuers, markets or counterparties discussed in its research. Market prices, estimates and scenario probabilities can change without notice.
Scenario targets are analyst estimates. Public policy rates are used as a carry proxy; executable returns depend on forward points, basis, transaction costs, liquidity, funding and volatility. CFTC positions are separate AUD and JPY futures series and do not identify a specific AUD/JPY position.
Prepared 20 August 2026. Historical performance and statistical relationships do not guarantee future results.
