When Yen Crosses Stop Moving Together: Reading the AUD/JPY Fracture Line

The Bank of Japan left its policy rate at 1.0% on 31 July, an 8–1 decision with the dissenter arguing for 1.25%, and warned that underlying inflation is likely to run clearly above target from the second half of its fiscal year. The rate itself is unchanged, but it sits at the highest level since the mid-1990s after June's hike. The night before the decision, the yen had been trading near its weakest against the dollar in roughly four decades — around 163.5 — before snapping back toward 158 on what was widely reported as intervention by Tokyo.

I want to write about the hour after that snap, because it was the cleanest example I have seen in years of something that is otherwise hard to demonstrate: what it looks like when the yen crosses genuinely become one trade.

For that hour, they were. USD/JPY, EUR/JPY, GBP/JPY and AUD/JPY all moved in the same direction, on the same clock, at broadly similar percentage scale. One driver, four instruments. If you were short the yen anywhere, you were short it everywhere, and the fact that you had spread the exposure across four "different" pairs bought you nothing at all.

That is the exception. Most weeks the yen crosses are not one trade, and the interesting question for a discretionary trader is not "is the yen weak" but "is the yen the thing that is actually moving right now, or is it the other side?"

Two regimes, one chart

Any cross is an arithmetic statement about two currencies. AUD/JPY going up can mean the Australian dollar is bid, the yen is offered, or some mix. The chart cannot tell you which, because the chart only shows the sum. This sounds obvious written down. It stops being obvious late in the New York afternoon when a candle is running and you are deciding whether to add.

In practice the yen complex spends its time in two rough states.

JPY-driven. Something happens to the yen — a policy meeting, a Ueda press conference, an intervention scare, a global risk-off flush that unwinds carry. Every yen cross moves together. Dispersion between them collapses. In this state the crosses are close substitutes, correlation between your positions goes to something near one, and the only real choice you are making is which counter-currency you would rather own while you express a yen view.

Base-driven. Nothing much is happening to the yen. AUD/JPY moves on Chinese data, iron ore, or an RBA repricing while EUR/JPY sits still. GBP/JPY moves on a gilt story that has nothing to do with Tokyo. In this state the crosses are genuinely different trades, and lumping them together as "yen shorts" is a category error.

The transition between those two states is what I have started calling the fracture line, and AUD/JPY is usually where it shows up first.

Why AUD/JPY fractures first

AUD/JPY has an unusual construction. The yen leg is the most policy-sensitive and most carry-sensitive major there is. The Australian dollar leg is the most commodity- and China-sensitive of the majors, with its own central bank on its own schedule — the RBA is on hold at 4.35% and does not meet again until 11 August, meaning that for the entire back half of July the Australian side of this cross had no RBA decision on the calendar.

So AUD/JPY is a cross where the two legs are driven by almost entirely separate news flows. When the yen story dominates, AUD/JPY joins the herd. When the yen story goes quiet for a few days, AUD/JPY is usually the first to peel off, because its base currency has the loudest independent voice. EUR/JPY and CHF/JPY tend to stay glued to the yen story longer, because their base legs are quieter and more correlated with each other.

This has a practical consequence I did not appreciate for a long time. If you want to know whether the yen complex is still trading as one thing, do not watch USD/JPY. USD/JPY is the yen's headline pair and it will look yen-driven almost by default. Watch the gap between AUD/JPY and EUR/JPY. When they diverge, the single-driver regime is over, whatever USD/JPY looks like.

Measuring it without an indicator

You do not need any tooling to do a first pass at this. Take a fixed window — the last four hours, or since the Tokyo open — and write down the percentage change of USD/JPY, EUR/JPY, GBP/JPY, AUD/JPY, CAD/JPY and CHF/JPY over that window. Not pips. Percent, so the pairs are comparable.

If all six have the same sign and cluster within a reasonably tight band, you are in the JPY-driven regime. If the signs are mixed, or one pair is doing three times the move of the others, you are not — something is happening to a base currency, and the cross carrying it is not a yen trade at that moment.

That crude test catches most of it. What it does not do is tell you which currency is responsible when the picture is messy, and it does not scale past the yen complex. That is the arithmetic that currency strength meters exist to automate: take the majors, compute each one's change against all the others over the same window, average it, and rank. The output is not a forecast. It is a decomposition — an attempt to split each cross into the part attributable to the base and the part attributable to the quote.

I run this on a MetaTrader 5 chart as a permanent background panel rather than as a signal source. In the screenshot above, the ranking is doing the only job I ask of it: telling me the spread between the strongest and weakest majors, and where the yen sits inside that spread. A yen pinned at one extreme with a wide gap to the rest is a JPY-driven tape. A yen sitting mid-table while one base currency runs away at the top is a fractured one.

What changes when you know which regime you are in

Three things, for me.

Position count stops lying to you. Three yen shorts in a JPY-driven regime is one position at triple size. That is a risk statement, not a diversification statement. The clean version of this test is to ask what a single Ueda sentence would do to the book. In late July the honest answer was "all of it, at once," and sizing should have reflected that before the intervention headline, not after.

Stop placement gets a reason. If AUD/JPY is moving because of the Australian leg, then the risk to that trade is Australian — data, China headlines, iron ore — and a stop sized to yen volatility is the wrong width. If it is moving because of the yen, the reverse. Same chart, same level, different correct distance depending on which currency is actually driving.

Expression improves. If I have a yen view and the tape is JPY-driven, the cross barely matters for direction but matters a lot for cost and for what else I am implicitly taking on. Buying the yen against a currency that is itself falling is a cleaner expression than buying it against one that is bid. This is where I use the pinned PICK mode rather than the automatic strongest-versus-weakest one: I want the panel to hold JPY against a specific counter-leg I have chosen, not to keep rotating to whichever pair happens to be extreme.

The tool I use for this is the free Currency Strength Duel panel — it ranks the eight majors live and lets you pin any two of them as a fixed comparison. I built it, so treat that as disclosure rather than a recommendation; the arithmetic is not proprietary and you can reproduce the ranking part in a spreadsheet if you prefer.

Where this breaks

Strength ranking is a description of what has already happened. It has no predictive content on its own, and it is worth being blunt about the ways it misleads.

It is lagging by construction. Every reading is an average of changes that are already in the price. On the night of the snapback the ranking would have put the yen at the top of the table — correctly, and far too late to be information.

It is basket-dependent. Ranking eight majors and ranking twenty-eight pairs including crosses give different answers to the same question, and neither is more correct. Change the timeframe and the ranking changes with it; a currency can be the strongest on H1 and the weakest on D1, and both statements are true.

It says nothing about cause. A currency can rank at the top because of genuine demand or because its counterparts are having a bad day. The ranking cannot distinguish those, and during a policy event the two are entangled anyway.

And it does not survive an intervention. When an official bid enters the market, the decomposition is still arithmetically valid but no longer economically meaningful — the yen is strong because someone decided it would be, not because of anything the other seven currencies did. The honest response to a tape like Thursday night's is smaller size and wider assumptions, not a sharper indicator.

What I am watching from here

The near-term question is whether the yen complex stays welded together or comes apart again. The 31 July hold leaves an inflation warning on the table without a rate move behind it, which is the sort of setup that keeps the yen twitchy for weeks rather than resolving anything. Meanwhile the Australian side gets its own event on 11 August.

If AUD/JPY starts moving on Australian news while EUR/JPY sits quiet, the fracture is back and the crosses are separate trades again. If the whole complex keeps moving as one block into the RBA, then it is still a single yen position wearing four different tickers, and it should be sized as one.

I wrote about the rates side of this same story from the metals angle a couple of weeks ago in The Real Yield Flip. The FX version is narrower but has the same shape: the interesting work is not predicting the policy path, it is knowing which leg of your position is actually exposed to it.