Japan’s carry-trade unwind is not done. The yen tells you when.
USDJPY at 157.20 is 4 yen from the MOF's implicit intervention threshold. With Ueda's BoJ signalling October hike, the carry-trade unwind that started in August 2024 has another leg to go.
USDJPY printed 157.20 at the Tokyo close on Tuesday, weakening 1.4% on the week despite the broader dollar trading sideways. The cross is now 4.30 yen from the 161.50 level that triggered the May 2024 unscheduled MOF intervention, and roughly 2 yen from the 159 level that Vice Minister for International Affairs Masato Kanda flagged in his April briefing as “watched with strong concern.” The Bank of Japan under Governor Kazuo Ueda has telegraphed an October policy meeting hike — the first since the symbolic 10bp move in July 2024 — but the market is still skeptical, with overnight index swaps pricing only 13bp of cumulative hikes between now and December. That gap between BoJ guidance and market pricing is the fault line that the next carry-trade unwind will run along.
What is at stake is the unwinding of an estimated $20 trillion structural short-yen position built over a decade and a half of zero rates. The August 2024 unwind — when USDJPY dropped from 161 to 142 in three weeks and the Nikkei fell 12% in a single session — was the most violent single-asset deleveraging since the 1998 LTCM crisis. The IMF and BIS estimated that roughly $400-500bn of carry-trade exposure was unwound in that episode, against a total estimated short-yen position of $20tn across hedge funds, mutual funds, official sector reserves, and Japanese household FX deposits. In other words: less than 3% of the position liquidated in August 2024. The other 97% is still on the books, and a yen-positive policy surprise from the BoJ remains the most asymmetric risk in global macro.
The Ueda communication strategy is deliberately ambiguous
Governor Ueda’s 23 May Tokyo Foreign Correspondents’ Club address argued that “underlying inflation has been gradually heading toward the price stability target” and that “the conditions for policy normalisation continue to be met.” Translated from BoJ-speak: a hike is coming, probably at the 31 October meeting, with risk of acceleration to the 19 September meeting if USDJPY breaks 160. The ambiguity is the policy. Excessive precision on the timing would trigger the carry-trade unwind in advance and force MOF intervention to manage the rate of depreciation in the dollar leg. By keeping the timing fuzzy, Ueda preserves optionality and makes the market do the work of pricing in the move gradually.
The complication is that the Federal Reserve has shifted dovish at the same time. With CME FedWatch now pricing a 71% probability of a September Fed cut, the rate differential that drives USDJPY is narrowing from both ends — BoJ rising, Fed falling. That should be yen-positive on a fundamentals basis. The fact that USDJPY has weakened anyway — actually moved toward the intervention zone — tells you the carry trade is not yet responding to fundamentals. Speculative positioning data from the CFTC Commitment of Traders report shows leveraged funds still net short 64,000 JPY futures contracts as of last Tuesday, down from a 178,000 short peak in June 2024 but still meaningfully negative.
The MOF intervention math
Japan’s Ministry of Finance has $1.27 trillion in foreign currency reserves per the latest monthly disclosure, of which roughly $1.05 trillion is in US Treasuries and dollar deposits. That is the war chest. The April 2024 intervention spent approximately $62bn over two trading days to defend the 160 level. The May 2024 intervention spent approximately $36bn defending 161. The pattern: MOF intervenes in size around round-number psychological levels, lets the market trade in between, and re-engages when the trend resumes. The implicit playbook is that 160 is the soft floor and 162 is the hard floor — beyond which intervention escalates from yen-buying to coordinated G7 communication.
| Episode | Date | USDJPY peak | Est intervention spend | Effect (30d) |
|---|---|---|---|---|
| Black Wednesday TMS | Apr 29 2024 | 160.20 | ~$36bn | USDJPY -2.8 yen |
| Second intervention | May 1-2 2024 | 161.95 | ~$26bn | USDJPY -4.6 yen |
| August unwind | Aug 5 2024 | 146 (low) | $0 (market move) | USDJPY -15 yen in 3 days |
| Q4 2024 stealth | Oct-Dec 2024 | 156 | ~$18bn (suspected) | USDJPY -2 yen |
| Spring 2025 | Apr 2025 | 158 | ~$22bn | USDJPY -3 yen |
The structural short positioning is the issue
The carry trade has multiple layers. The top layer is speculative — leveraged funds and prop shops borrowing yen at near-zero rates and parking it in higher-yielding dollar, Mexican peso, Brazilian real, and Turkish lira assets. That layer is reflected in CFTC data and unwinds first under stress. The middle layer is Japanese retail FX margin accounts at Gaitame Online and similar venues, which historically run heavily long USDJPY and capitulate in cascading stop-loss waves when the cross moves more than 3 yen in a session. The bottom and largest layer is structural: Japanese pension funds (GPIF, Pension Fund Association), life insurers (Nippon Life, Dai-ichi Life), and corporate treasuries holding hundreds of billions in unhedged or partially hedged foreign assets to capture the yield differential.
The structural layer is the dangerous one. It does not unwind on a daily basis — it unwinds when the BoJ raises rates enough that the cost of hedging the FX exposure exceeds the additional yield earned on the foreign asset. That threshold is currently estimated at around 75bp of BoJ policy rate, which would push 3-month yen forward points to a level where dollar-denominated 10-year Treasury exposure becomes barely positive net of hedge cost. Once that threshold is crossed, the pension and life insurance hedging programmes ramp up systematically, mechanically buying yen and selling dollar in size. The August 2024 episode was the leveraged layer panicking; the structural unwind has not yet started.
The Nikkei is doing the early signalling
The Nikkei 225 closed Tuesday at 38,140, down 4.2% on the month and underperforming the S&P 500 by 6.1 percentage points. That dispersion is the cleanest early warning that the carry-trade unwind is in progress at the speculative layer. Japanese equities are typically a long-yen-funding trade for global allocators — funds borrow yen, buy Japanese equities, and earn the difference between equity returns and the funding cost. When the funding side gets uncertain, those positions unwind regardless of the fundamentals on the equity side. The Topix banks index, which traditionally benefits from rising rates, has underperformed the broader Topix by 2.8 percentage points this month — counter to what you would expect from a hawkish BoJ pivot. That tells you that the marginal seller is not adjusting for the rate path; the marginal seller is reducing yen-funded exposure across the board.
The cross with US equities matters. In August 2024 the Nikkei led the S&P 500 by approximately 24 hours into the unwind. If the Nikkei’s current dispersion continues — particularly if it accelerates on a session where USDJPY moves more than 1 yen — the read-across to US equities and risk assets globally is one trading day. That is the signal to watch. The Nikkei 38,000 level is the next meaningful technical floor; a break below would likely correspond to USDJPY moving past 159 and MOF intervention becoming likely.
What this means for crypto and global duration
The August 2024 carry-trade unwind took Bitcoin down 18% in three sessions, despite no Bitcoin-specific catalyst. The mechanism was deleveraging — funds unwinding yen-funded long positions sold whatever was liquid, and BTC was among the most liquid risk assets in the global book. The same dynamic applies in 2026, with the additional wrinkle that BTC is now meaningfully more institutional and therefore more correlated to the equity tape that gets hit first. If the carry-trade unwind resumes in earnest before the October BoJ meeting, expect BTC to underperform on a beta basis for 5-10 sessions before stabilising. The cleanest hedge is long-dated yen calls against the BTC position, sized roughly 15-20% notional to capture the asymmetric tail.
The duration read-across is the opposite direction. A yen-strength episode is typically associated with global risk-off, which is positive for long-duration Treasuries through the safe-haven channel. The 30-year Treasury rallied 22bp during the August 2024 unwind in three sessions. If the October BoJ delivers a clean hike and the carry trade unwinds, expect Treasuries to rally aggressively into year-end, the dollar to weaken broadly, and gold to outperform. The complication is that all of those moves are already partially priced — the yen short positioning is meaningfully smaller than it was in June 2024, and the duration trade has already absorbed some of the rally room.
What to watch next
Three things. First, the BoJ’s 14 June Tankan survey of corporate sentiment — a strong large-manufacturer reading above +14 gives Ueda the cover to telegraph a September hike rather than October, which would be the cleanest near-term yen-positive catalyst. Second, the MOF’s monthly FX reserve disclosure on 7 June — any unexplained $10bn+ drop indicates stealth intervention, which historically precedes formal intervention by 5-15 sessions. Third, the cross-currency basis on 3-month JPY funding, which has been widening as banks position for tighter yen funding. A move to 70bp+ in the cross-currency basis is the cleanest signal that the structural hedging programmes are accelerating.
For risk-asset positioning, the cleanest expression is long yen calls struck at 152 USDJPY with September expiry, paired with a partial short Nikkei futures position to capture the systematic risk-off. The cost is modest because the implied volatility on yen options is currently 8.4 vol points, below the 12-vol points that typically prevails during intervention episodes. The asymmetry favours owning the optionality here, even if the central case is that USDJPY drifts toward the 160 line slowly rather than gapping there. Track BoJ communications, MOF reserve data, and the Nikkei tape on our events calendar and the live yen and curve data in the market hub.
The honest read: the August 2024 unwind was the appetiser. The full carry-trade rebalancing has not happened, and it will not happen in a single session — it will happen across the BoJ rate normalisation cycle that extends into 2027. Each step of that path is a yen-positive catalyst, and each catalyst has the potential to compress risk-asset prices globally. Position with that asymmetry in mind. The yen tells you when. The Nikkei tells you it has started.