The yen carry trade was the most consensus position in global macro for fifteen years. Borrow at 0% in Japan, lend at 5% somewhere else, pocket the spread, sleep well. Until August 5, 2024, when the trade unwound in 36 hours and Japanese equities had their worst single day since 1987. By the close that Monday, USD/JPY had moved from 161 to 142. Carry traders who had been comfortable for a decade were getting margin-called by their brokers in real time.
I want to think through where the trade actually stands in May 2026, because the consensus narrative has rotated through three different "this time is different" framings since then, and most of them are wrong.
The Bank of Japan started rate normalization on March 19, 2024 with the first hike since 2007. Policy rate moved from -0.1% to +0.1%. By July 2024, BOJ had hiked again to 0.25%. By December 2024, to 0.5%. Through 2025, BOJ paused, citing wage data. In March 2026, BOJ resumed with a 25bp hike to 0.75%. The current expectation embedded in OIS markets: BOJ reaches 1.25% by end of 2026, possibly 1.5% by mid-2027.
That's the technical state. The economic state is different.
Japan's CPI is running 2.6% as of Q1 2026. Real wages have been mildly positive for six consecutive quarters — the longest stretch since the 1990s. The BOJ's 2% inflation target has been sustainably achieved, by their own definition. The political pressure to normalize fully has shifted from "go slow because deflation might return" to "go fast because the yen is too weak." That's a different policy environment than even six months ago.
What the August 2024 Unwind Actually Taught
The August 5 carry unwind wasn't caused by a single bad data print. It was caused by a positioning extreme meeting a coincidence of triggers: BOJ's surprise hike on July 31, weak US payrolls on August 2, geopolitical concerns over the weekend, and a margin call cascade as USD/JPY broke 150.
CFTC speculative positioning data showed JPY shorts at multi-year extremes going into the unwind. Estimated total positioning, accounting for unreported retail and corporate hedging: $1.5 to $2 trillion notional in carry-trade-equivalent positions across various structures. When forced unwinding hit, there was no natural counterparty to absorb it because everyone was on the same side.
The lesson institutional desks took: position sizing matters more than rate differential. A 5% carry yields nothing when you take a 12% mark-to-market loss in 36 hours. The lesson retail hasn't fully internalized: leverage on carry trades amplifies the tail risk in ways that look fine for years and catastrophic in days.
Where the Trade Stands in 2026
The current rate differential between US and Japan is approximately 4.0%. The Fed funds rate is 4.75% (after 2025 cuts), BOJ is 0.75%. That's still a meaningful carry — about half what it was at the August 2024 peak — but enough to attract structured products, FX-linked yield enhancements, and the more disciplined macro hedge fund flow.
What's different about the trade in 2026:
First, BOJ is no longer pinned at zero. Every hike compresses the carry. The marginal trader needs to model not just the current differential but the path differential.
Second, JPY positioning has normalized. CFTC speculative JPY shorts dropped from 184,000 contracts in July 2024 to a net long of 28,000 contracts by November 2024, then drifted back to a small short of 67,000 contracts by early 2026. There's no extreme positioning to unwind violently anymore. The trade is more crowded than 2014 but less crowded than 2024.
Third, the implied volatility regime has reset. USD/JPY 1-month implied volatility ran 7-9% through most of 2018-2023. It spiked to 19% during the August 2024 unwind. It's currently running 11.4% — elevated relative to the pre-2024 era but not crisis-level. That higher vol means option-implied tail risk is higher. Option-replicated carry positions are more expensive.
What Actually Works in 2026
The institutional carry trade in 2026 isn't borrow-yen-buy-treasuries. That trade is dead because the differential doesn't justify the volatility risk. The new structure is more nuanced: short JPY against higher-yielding emerging market currencies (MXN, BRL, ZAR) where the differential is 8-12% and the correlation to broad risk-off events is mostly idiosyncratic.
The cleanest 2026 carry equivalent: long MXN/JPY through a tier-1 broker, sized so that a 8% adverse move (the 2024 stress level for MXN/JPY) wouldn't hurt you. Daily carry currently around 4 basis points. Annual carry approximately 10-11% gross. Realistic net after spread and overnight adjustments: 7-8%.
For pure USD/JPY directional traders, the key signal is the 2-year US-Japan rate spread. When US 2y yield rises faster than Japan 2y yield, USD/JPY tends to follow within 5-10 sessions. When Japan 2y rises faster (rare but happening more often as BOJ normalizes), USD/JPY weakens. Track both yields daily; the differential change is the trade.
What Retail Should Avoid
Avoid leveraged USD/JPY carry positions sized larger than 2x your monthly income. The 2024 unwind taught a lesson that traders who weren't trading then haven't learned: carry trades pay 8% per year and lose 18% per crisis. The annualized return is positive only if you survive the crisis.
Avoid yen-denominated leveraged products structured by Japanese retail brokers. Daily-resetting leverage on JPY pairs amplifies the wrong end of the volatility distribution.
Avoid the assumption that BOJ will be slow. Every Japanese central banker since Ueda has signaled a faster normalization path than the previous one. Position for the more hawkish scenario, not the consensus one.
The yen carry trade is alive in 2026, just smaller and more sophisticated than it was. The retail version of it — borrow free yen, buy anything — is gone, and isn't coming back as long as Japanese inflation stays sustainably above 2%. Trade the trade that exists, not the one that used to exist.