BoJ decision day arrives as DXY finally breaks its month-long flat line, and EZA fails its own retest
One analyst's qualitative review. Not backtested alpha. Not trade advice.
What I'm watching
- DXY 100.104 (rising), VIX 17.71: the first real move in either after a month pinned near 98.7-99.3 (Sept 1 through Sept 10). DXY is up more than a full point since Sept 10's 98.739.
- BoJ, Sept 17-18: the meeting flagged as "the only live catalyst" three entries running is happening today and tomorrow. This is the first entry landing inside the decision window itself.
- EZA $67.55: the Sept 10 partial bounce to $71.31 has fully round-tripped and then some, undercutting even the Sept 3 rollover low of $69.99.
- EWG $42.43, 1m -3.48%: Germany's falling-PMI drift, logged unchanged for three entries, now shows up as an accelerating monthly decline rather than just a flat PMI print.
What changed / what matters
A month of directionless DXY (98.77 to 99.30 to 98.74 across Sept 1/3/10) has broken to the upside right as the BoJ sits down to decide. That timing matters more than the magnitude: a firming dollar into a hawkish-leaning BoJ meeting is the opposite of the carry-friendly setup EWJ has been pricing. EWJ's 52-week position barely moved (93.7% to 93.4%) but 1-month momentum flipped from fading-positive to outright negative (0.99% to -1.18%), which reads as the market front-running the decision rather than staying purely priced-in. With the vote itself unresolved as of this write-up, the honest call is to hold through the event rather than lean either way.
EZA is the clearer story. The Sept 10 entry flagged the partial bounce off Kganyago's inflation warning as "a retrace inside a decaying trend rather than a resumption," and set an explicit revision trigger: a close back above $71.68 on rising momentum. That did not happen. Instead EZA gave back the entire bounce and fell through the prior floor, $67.55 against the Sept 3 low of $69.99, with drawdown widening to -15.4% from -10.7% a week ago. The avoid call from two entries ago is now confirmed rather than disconfirmed, and the disconfirmation window is closed until a fresh catalyst appears.
EWZ keeps compounding, but noisily: 1-month momentum accelerated again (8.18% to 10.33%), yet drawdown widened alongside it (-7.9% to -9.33%), meaning the rally is coming with bigger round-trips inside it, not a smoother climb. Still the cleanest confirming trend in the set, just a more volatile one. EWG and EWU continue the same falling-PMI drift with no new print to hang a number on, though EWG's monthly return (-3.48%) is now the weakest single reading among the five.
Candidate picks within this theme
- EWZ (Brazil) - hold/accumulate; momentum still accelerating, though the widening drawdown means bigger swings on the way up.
- EWJ (Japan) - hold through the event; 1m momentum flipped negative right as DXY broke its flat line into the BoJ decision, too live to call either direction pre-vote.
- EZA (South Africa) - avoid, confirmed; the Sept 10 revision trigger ($71.68 close on rising momentum) failed, and price fell through the prior low instead.
- EWG (Germany) - avoid/watch; PMI drift now shows up as the weakest monthly return in the set.
- EWU (UK) - hold/avoid; unchanged, no new print this pass.
What I'd revise if I saw
A dovish BoJ surprise (no hike, or dovish guidance) that still leaves EWJ momentum negative would separate the FX story from the equity story and be worth a standalone note. For EZA, only a close back above $71.68 on rising, not fading, 1-month momentum reopens the case.
Cross-references
- Previous entry on this theme:
docs/thinking/2026-09-10-fx-policy-a-flat-dxy-month-leaves-the-boj-meeting-as-the-only-live-catalyst.md - Data:
.cache/regime.json(DXY/VIX history),.cache/thinking/wake-context.md(ETF stats)