Between 30 July and 26 August the Japanese Ministry of Finance spent 15.4 trillion yen — about $98.66 billion — buying its own currency, jointly with the United States. Reuters reports it as the largest single-month intervention operation on record, and the first coordinated US–Japan action since 2011. It dragged USD/JPY from near 164 to 155.20 by 3 August. This morning, with no official bid anywhere in the market and New York shut for Labor Day yesterday, the pair traded through that level and kept going — to 153.53 on the Reuters wire, and towards 153.00 through the London session. The state paid a hundred billion dollars for a level the market has now given away for free. That is not a policy story. That is a positioning story, and it changes how much you should be carrying.
What 15.4 trillion yen actually bought
The receipt is public. Japan’s foreign reserves fell from $1.287 trillion at the end of July to $1.208 trillion at the end of August — a drop of $79.6 billion, or 6.18%, the biggest ever recorded. MUFG’s read of the same release puts the securities component down $87.8 billion to $839.6 billion, with roughly 70% of reserves held in foreign securities, mostly US Treasuries. Tokyo and Washington also signalled that Japan could draw on a COVID-era Federal Reserve facility to raise dollar liquidity without outright Treasury sales, which is the detail that tells you how carefully the operation was funded.
What it bought was nine big figures and about five weeks. The yen reached 155.20 on 3 August and then drifted back out towards the mid-150s over the rest of the month. By the start of last week USD/JPY was near 160 again. Measured as a trade, the MOF was underwater on its own intervention within a month.
Now read the same fact the other way, because this is the part that matters to you. The level that cost ninety-eight billion dollars in July and August has been taken out this week by nothing more than private flow. When a market clears a defended level without the defender showing up, the defence is not what changed. The supply of sellers did.
The shorts came back — and that is the fuel
MUFG’s note this morning makes the mechanism explicit: yen shorts started to rebuild after the joint intervention at the end of July, and the scale of the recent move suggests a reduction in that position rather than the establishment of a new one. The CFTC data supports it. Speculative yen positioning went to a net short of 92,227 contracts in the week to 4 September, deepened from roughly 63.3 thousand the week before. Traders were adding to short yen into a rally that had already started.
That is the entire story of the last six sessions. A short base built on the assumption that intervention marks a ceiling, run into a market that then went straight through the ceiling. Tony Sycamore at IG, quoted on the Reuters wire, called the drop “a sharp unwind of yen shorts” after the break of the supports around 155 that had held in August and May. He is describing a squeeze, and a squeeze has a specific property that a view does not: it is self-reinforcing until the position is gone, and then it stops abruptly.
You cannot see when the position is gone. Nobody can, until the following Friday’s COT release. What you can see is that as of the most recent count there were still more than ninety thousand contracts of it, and the price has moved about 4% against them in six sessions. That is a lot of pain that has not yet been reported.
Monday’s biggest move happened with New York shut
The single largest day of this rally was Monday, roughly 1.2%, and US markets were closed for Labor Day. ING’s FX daily says it plainly: thin US holiday liquidity amplified the moves. That is not a footnote. A 1.2% day in USD/JPY on a holiday is a smaller amount of real money than a 1.2% day on a normal Tuesday, and it deserves less weight in whatever your system uses to decide that a trend exists.
If you run a breakout or momentum filter on daily bars, Monday’s bar is currently being counted at full weight alongside every other bar in your lookback. It should not be. A holiday session is a lower-information observation of the same variable, and treating it as equal is how a system convinces itself a move is stronger than the flow behind it. You do not need a clever fix; you need to know which of your recent signals rest on that bar.
13:00 UTC today is the first fully staffed test
The New York open at 13:00 UTC is the first time since Thursday’s close that the full US dollar market is in the building at the same time as a Tokyo move of this size. Everything between then and now has been priced by Asia, by London, and by a skeleton US crew. That makes today’s afternoon the first honest read on whether this is a repricing or an unwind that has run out of people to squeeze.
The practical version, and it is a sizing answer rather than a direction one. If you are short yen anywhere — in USD/JPY, in EUR/JPY near 178.75, in GBP/JPY down towards 208 — you are not holding three positions. You are holding one position, and the whole week’s move has been against it. Net your JPY exposure across every pair and look at the number as a single line. If that number is larger than you would open fresh right now, at these levels, with a squeeze in progress and a Bank of Japan meeting on 18 September, then the market has sized you rather than the other way round. That is the trade to fix before 13:00 UTC, not the directional one.
The dollar itself is barely involved. The dollar index sat at 98.83, the euro at $1.1628, sterling at $1.3549, the Aussie at $0.7219 — all of them within a rounding error on the day. ING’s framing is the right one: this is a yen story, not a dollar rejection. If your system is reading it as broad dollar weakness and putting on correlated shorts elsewhere, it is reading a one-currency event as a two-sided one.
What this does not tell you
It does not tell you the squeeze is over, or that it has further to run. Positioning data is a week stale by the time you see it; the 92,227 figure describes 4 September, and four sessions of violent price action have happened since. The number could be half that by now.
It also does not settle the repatriation question, which is the other explanation being offered for the move. Part of the market is trading an expectation that Japan’s GPIF shifts allocation home — the fund managed 293.6 trillion yen, about $1.81 trillion, as of March 2026, split roughly a quarter each across domestic bonds, foreign bonds, domestic equities and foreign equities. But Reuters reported in July that government sources said no overhaul of those medium-term targets was planned, with one official noting markets had reacted far more than expected, and the minister said again this week that a review is still only being examined. A theme the government keeps declining to confirm is a weak thing to hold a position on.
And it does not rule out the MOF coming back. Nothing about a record August operation prevents a September one, and a yen that is now strengthening does not need defending in the same direction. But a ministry that has just spent 6% of its reserves does have less room, and that asymmetry is worth remembering before you assume anyone will catch this for you.
Related
- FX events calendar — the BOJ meeting on 18 September and this week’s US releases, in UTC.
- Signals — how we frame exposure when a move is positioning-driven.
- Expert advisors — netting correlated JPY exposure across pairs.
