When the yen strengthens sharply, investors who borrowed cheaply in yen to fund stock bets get squeezed out, and history shows those forced sales land hardest on high-flying AI and tech names. The oil shock driving investors toward safety is simultaneousl

Verdict: the yen-trade thesis is arming, but it isn't armed — wait

The verdict is **wait** — the thesis is plausible, but the trade isn't armed yet. The strongest argument for it is the alignment of signal, macro and deteriorating marginal fundamentals: the Reuters piece (September 9, 2026) has oil pushing toward $100 lifting the yen while chilling risk appetite, and NVDA's quarter ended July 26, 2026 showed free cash flow down 52.2% to $21.4B and net margin compressing from 71.5% to 62.0% despite $96.2B of revenue. The strongest argument against is that the strategy would be fading a company at the 99.5th sector percentile for free cash flow, with buybacks (shares down 0.4% to 24.1B) and a dividend up from $0.04 to a trailing $0.28 per share — and the $565.4M insider selling figure is a delayed disclosure from the June 30, 2026 reporting window, not a current signal. The rule set itself is modest: 21.7% over five years on 11 trades at a 54.5% win rate with a 9.1% max drawdown, and the parameter-sensitivity run exceeded its budget, so no robust nearby-parameter setup was established. Right now FXY's RSI at 79.8 needs to fall below 50 and QQQ's trend strength sits at 6.7 versus a 20 trigger — the entry conditions are not met. The single fact that would flip the verdict to action: FXY's RSI crossing below 50 with price at or below its $57.90 20-day average while trend strength holds above 20.

Trade now: yen-haven trigger is close on price, far on momentum

This is a **wait**, not a trade — but the gap to entry is narrowing on one of two legs. The idea goes long the yen (FXY) and long QQQ when the yen signals haven strength and the tech tape is cracking; today the FXY leg is two-thirds there. FXY closed at $59.56, comfortably above its 50-day average of $57.32 (met) and with trend strength at 49.9 versus the 20 threshold (met). Price is $1.66 above the 20-day average of $57.90 — the closest unmet condition. Momentum is the blocker: FXY's 14-day RSI sits at 79.8 and needs to fall below 50, a 29.8-point gap. On the QQQ leg, price at $717.72 is near the 20-day average of $714.62, but RSI at 54.6 and trend strength at 6.7 (needs above 20) are both far from triggering. The evidence here is a completed backtest: over five years the FXY leg produced 21.7% total return across 11 trades with a 54.5% win rate and a 9.1% maximum drawdown; the last 24 months alone added 12.2% at a 66.7% win rate. That supports waiting for the full signal rather than front-running it. Risk framing once live: position sizing risks 2.4% per position (capped at 25% of capital), the hard stop is a 2.4% loss and the take-profit is a 4.8% gain — an effective 2:1 reward-to-risk. Positions also exit after 45 trading days or on support-break levels ($57.65 and $55.98 for FXY). What "wait" means concretely: watch FXY's RSI. If FXY pulls back toward its 20-day average and RSI cools below 50 while trend strength holds above 20, the long-yen leg arms. Until then, no position.

The yen squeeze thesis is firing with the tape, the filings, and the macro all pointed the same way

The thesis argues that a sharply strengthening yen forces unwinding of yen-funded carry trades, and that the pain lands hardest on expensive AI and tech names. The macro setup per the cited Reuters piece (September 9, 2026) fits precisely: oil running toward $100 is simultaneously pushing the yen higher and chilling equity sentiment — a double headwind for risk assets. The MarketWatch piece (September 8, 2026) makes the same causal argument the idea makes: a stronger Japanese currency is a specific trouble signal for AI and technology stocks. So the confirming signal and the target trade are aligned. The rule set is backtested and has traded. Over the 60-month window the strategy produced a 21.7% return on 11 completed trades at a 54.5% win rate, with a maximum drawdown of 9.1%. The same setup traded profitably on shorter horizons too: 12.2% over the most recent 24 months on 9 trades at a 66.7% win rate, and 5.4% over 12 months. The recent results improve on the full-window record, which is consistent with the thesis claim that the carry-unwind mechanism is currently active rather than dormant. The fundamentals support the 'expensive tech is most exposed' half of the thesis. NVIDIA's latest quarter (ended July 26, 2026) still shows enormous scale — revenue of $96.2B, up 17.9% sequentially, with a 62.0% net margin — but the marginal direction of the numbers is softening: free cash flow fell 52.2% to $21.4B, operating cash flow dropped 52.2% to $24.1B, net margin compressed from 71.5%, and debt-to-equity jumped from 0.038 to 0.141, a 269.9% increase. A market that has been paying for flawless execution in the AI complex is now looking at decelerating cash generation and…

Scores

• Conviction score breakdown: 55
• Thesis support: 65
• Trade readiness: 35
• Risk quality: 60
• Backtest evidence: 55
• Fundamentals trend: 60

Watch items

• FXY — RSI (14)
• FXY — Price vs 20-day EMA
• FXY — Price vs 50-day SMA
• FXY — ADX (14)
• QQQ — RSI (14)
• QQQ — ADX (14)
• QQQ — Price vs 20-day EMA
• NVDA — Quarterly earnings date
• NVDA — Insider net open-market activity