Japan Stock Positioning Hits All-Time Highs as AI Narrative Becomes Top Tail Risk
Alina Collins
Goldman strategist Bruce Kirk warns that crowding in Japanese equities now exceeds pre-August 2024 crash levels across the board, but the next trigger may not be the yen — it may be a collapse of the global AI narrative.
How crowded is the positioning, exactly?
Foreign investors have poured in roughly ¥14.8 trillion since April 2025. Net positioning is more than 20% above where it stood before the July 2024 crash.
Retail margin balances are 35% higher than July 2024, near a five-year peak.
Hedge-fund gross and net allocation to Japan sits at the 99th and 98th percentile of the past five years, respectively. This means → it is almost impossible to find a more crowded moment.
TOPIX and the Nikkei 225 are 37% and 53% above their July 11, 2024 levels — but the rally is concentrated in banks, steel/non-ferrous metals, electronics, and AI-linked exporters. A large share of index constituents still trade below their 200-day moving average.
How similar is the structure to the pre-2024-crash setup?
The Nikkei/TOPIX ratio — a gauge of large-cap concentration — widened in June to an 18× all-time high.
The median valuation of AI-related stocks in TOPIX is now nearly twice that of non-AI stocks. This means → the market's pricing of AI has far outrun its pricing of everything else.
In plain terms = portfolios carry a heavy implied tilt — long exporters and financials, short domestic defensives — a structure that closely mirrors the setup before the July 2024 crash.
What actually happened in the 2024 crash?
Phase one (July 11 to month-end): U.S. CPI surprised to the downside while yen intervention hit. Exporter sectors fell first, but bank stocks barely moved — they even rallied 5% on the day the BOJ hiked.
The real carnage came in phase two (July 31 to August 5): the BOJ's hike was more hawkish than expected; 48 hours later, U.S. payrolls collapsed. Two independent negative shocks converged. Banks plunged 27% from their hike-day peak to August 5.
In plain terms = most multi-strategy hedge funds set drawdown limits around -2.5% of deployed capital. A market-neutral book with a five-percentage-point sector skew could see a peak-to-trough loss of roughly -5% — enough to trip stop-losses. Stop-loss → forced unwind → long-only funds sell in sympathy → risk-parity and trend-following funds sense momentum reversal and pile in. A full negative-feedback loop.
After the August 5 plunge, TOPIX rebounded 23% from the low to September 3. This reflects a liquidity crisis, not a fundamental repricing.
Could the yen be the trigger again this time?
Goldman has raised its 3-month / 6-month / 12-month USD/JPY forecasts to 162 / 163 / 165. The case: higher-for-longer U.S. rates + low recession risk + Japan fiscal concerns + an extremely slow BOJ hiking path.
The driver of yen weakness has shifted. After the LDP lost the upper-house election in the second half of 2025 and the Takaichi government took power, markets began pricing in doubts about Japan's fiscal sustainability. The 10-year JGB yield is now approaching 3%, but this rise mostly reflects a widening of Japan's term premium — the extra yield investors demand for holding long-dated bonds — rather than a narrowing of the U.S.–Japan rate differential.
CFTC data show speculative yen shorts near July 2024 levels. But the key difference → this time the market has already priced in yen weakness. The 2024 crash happened precisely because markets had not priced in a sudden yen rally.
One-month USD/JPY implied volatility sits relatively low. This means → the market's guard against surprises is thin; if a surprise does come, the impact will be outsized.
Where is the real tail risk?
Kirk's view: the probability that the yen itself triggers a flash crash is lower than in 2024.
But any event that shakes the global AI growth narrative — something akin to the DeepSeek-driven selloff in Q1 2025 — or a geopolitical shock severe enough to undermine the "solid U.S.-led global growth" story could put today's crowded AI positions in the same predicament as the exporter positions of 2024.
In plain terms = the 2024 crash was later framed as a "Japan-specific problem." But right now, Japan's equity market is one of the most concentrated global expressions of the AI theme. If that narrative reverses, the fallout may not stay contained to Japan.
Content is for reference only, not financial advice.