Intervention against a fundamental trend buys days, not a reversal: reserves are finite, the market’s flow isn’t, and everyone knows which side runs out first. “How to burn $10bn in a week” — the 08/2026 USDJPY tick chart: the joint US/Japan yen defense knocked USDJPY from ~158 to 155.23 on 08/02; by 08/06 it printed 157.92, the entire estimated $60–80B (dgt10011 says $10B; sizes vary by teller) round-tripped in four sessions. Repetition makes it worse, not better: each intervention-driven yen rally hands carry traders a better re-entry level, so the position the intervention fights comes back bigger (Yen carry trade — Macrobysunil’s paradox, 08/2026).
The deeper problem, via jameslavish’s chart: the US–Japan 10y spread has narrowed from ~4.2% (2022–24) to ~1.85%, yet USDJPY kept climbing to ~163. When the rate differential no longer explains the weakness, the driver is confidence/flow (carry positioning, distrust of JGBs) — precisely what intervention can’t fix. Classic precedents from memory: sterilized BoJ interventions 2022 and 1990s faded in days-to-weeks; interventions only stick when they coincide with a fundamentals turn (rate convergence), which is why the durable tool is the FIMA repo — borrowing against Treasuries instead of selling them + policy shift, not spot sales.