Bank of Japan — scenarios before, scorecard after

Before — what we said

After the scenarios were published: release time moved 02:30 · 18/09 → 02:54 · 18/09; price-moving moment moved 05:30 · 18/09 → 06:30 · 18/09.

The BOJ prints at 02:30 UTC on 18 September, consensus 1.25% against 1.00% last — a 25bp hike. The mark that moves price, though, is 05:30 UTC, the press conference: the decision is largely in the price, the guidance on the next step is not. We are deliberately leaving gold out of the calls. Not for want of a channel — USD strength is gold's dominant lane — but because that lane forks into two branches that cancel each other (see transmission), and across 12 prints a BOJ day has given gold only 0.90 times its ordinary range. A gold call here is a coin toss. BTC and EURUSD stay in: on those two, print days genuinely differ from ordinary days.

The macro picture right now

The Fed's hike this Wednesday sits immediately in front of this meeting, and it leaves the BOJ landing on an odd base: the 2y at 4.65% (z +2.71) and 10y real yields at 2.60% (z +2.53) are both at extremes, yet broad USD sits at z -1.32. The dollar leg has stopped following US rates — and with that anchor loose, a squeeze out of Tokyo has more room to run than usual. MOVE at 83.7, up 7.569 over five sessions (z +1.19), says bond vol was already creeping higher before the print. Three positioning details matter: gold specs at 56.4% of OI (z +1.49) are a crowded long standing directly in front of a funding-cost shock; EUR specs are already short at -4.522% of OI (z -1.23), so a weaker-euro branch carries less fresh fuel than the range statistic implies; and BTC implied vol (DVOL) at 38.86, z -0.62, shows options are not priced for a carry unwind even though leverage was just flushed (funding -0.877 over five sessions).

How the last print went

We have no free series for the BOJ policy rate, so the indicator's own momentum is something we do NOT know rather than something that isn't there, and no sibling print on this channel has been declared either. The only clue is same-lane news, and it leans toward the hike: the BOJ cut its JGB holdings by a record ¥47.8tn in the year to July, so the tightening already ran ahead through the balance sheet and 25bp is continuation, not a turn; meanwhile pressure from Bessent has pushed the yen up and made the hike politically cheap. Where that clue can fail: the BOJ has historically preferred to move when the yen is weak, and a hike into a yen already strengthened by outside pressure is the kind that gets deferred. On 31 July we called EURUSD higher and it finished flat at 0.001% — 0 for 1. The lesson we carry: the euro leg on a BOJ day is second-order transmission, the range widens but we have paid for direction there before.

What it reaches, and how

Gold: the event lane is exactly gold's dominant lane (dxy r=-0.56 ±0.09), so the road is open — but it forks. A stronger yen pulling the dollar basket down pushes gold up; a carry unwind pulling risk appetite down pushes gold down, because across 60 sessions gold trades against the textbook with VIX (r=-0.38, flagged inverse). The two branches net out to that 0.90 ratio. BTC: same lane, direct, dxy r=-0.43 ±0.11. The chain: a hike with guidance left open → yen borrowing costs rise → yen-funded leverage is closed → dollar cash demand → BTC. Print-day range 3.73% against 2.73% on ordinary days, 1.37 times, over 12 prints. EURUSD: no axis clears the noise floor (us10y -0.26, vix -0.29), so the BOJ cannot reach the euro in a straight line and must detour: yen rates → the cost of holding EURJPY, the long leg of the carry → EURJPY sold → EURUSD. We cannot put a coefficient on that detour; what we can measure is that BOJ days widen the euro's range by 1.40 times, the widest of the three.

Scenarios when the number lands

Three branches, only ONE happens. Each carries its own numeric threshold and an expected direction per asset.

1. Numbered threshold

   A 25bp hike to 1.25% AND a 05:30 UTC press conference that does not push back on expectations of one more hike before year-end

   If this branch plays out:▬gold▼BTC▼EUR45%likelihood

   This is the only branch that actually opens the carry loop: yen borrowing costs first, then EURJPY and yen-funded crypto leverage closing behind it, then dollar cash demand. For how far it can run, look at DVOL at 38.86 — BTC options are not carrying this branch, which makes a surprise dearer here than in the euro, where the shorts are already in place.

2. Numbered threshold

   Still a hike to 1.25% but the press conference leans on 'gradual' and plays down a further move in 2026

   If this branch plays out:▬gold▬BTC▬EUR40%likelihood

   The decision matches consensus and is already in the price, so the entire move shifts onto the words — and words in this direction hand the yen back its gains and let the carry live another leg. BTC gets room inside its range, but the ceiling is the 76,652 area, not the 77,655 shelf that has already broken.

3. Price moved most like this branch 2 of 3 assets agreed

   Numbered threshold

   No change, the rate stays at 1.00%

   If this branch plays out:▼gold▬BTC▼EUR15%likelihood

   The only dovish shock on the table: the yen loses its support, the dollar basket gains, and for the euro that adds straight onto an already adverse rate gap. For BTC a firmer dollar presses down while yen-funded risk appetite comes back to lift — two forces cancelling, which is why we take no direction on BTC in this branch.

Where we commit — and what would prove us wrong

Unlike the branches above, these carry no “if”. They are the lines graded against real prices below, and they are never edited after the print.

• BitcoinBEARISHConviction 5/10

  Invalidated if: A four-hour close above 76,652 (a six-touch level) in the window after the press conference voids this bearish call.

• EUR/USDBEARISHConviction 5/10

  Invalidated if: A four-hour close above 1.1556 (five touches, 1.22 ATR away) voids this bearish call.

The three hours between the decision and the press conference: if the yen has already made its move by 02:30 UTC and the conference does not follow through, most of that move gets handed back the same afternoon.

Before — what we said

No forecast, no prior — what gets priced here is wording. Context: today's DXY +0.25% is almost entirely a yen story, while EURUSD is still +0.54% on flows leaving the dollar. So we put our only primary call on EURUSD, in line with the running bullish bias. Gold is deliberately skipped: the BOJ's one channel into gold is the global term premium, and that channel opens only in the two tail branches — in the base case gold sits inside its 33.2 H4 ATR and is still led by the US curve, not JGBs. Bitcoin is skipped too: only a genuine carry unwind reaches it, and a 24h change of 0.05% says this market is not looking at Tokyo at all.

Scenarios when the number lands

Three branches, only ONE happens. Each carries its own numeric threshold and an expected direction per asset.

1. Price moved most like this branch 2 of 3 assets agreed

   If this branch plays out:▲gold▼BTC▲EUR

   Hawkish branch: Ueda names a near-term hike, USDJPY falls 1.0% or more and the 10-year JGB adds at least 5 basis points. The dollar loses a leg, EURUSD takes the smaller share of that move and can test 1.1463; a firmer global term premium supports gold, while a partial carry unwind is a headwind for Bitcoin.

2. Base branch

   If this branch plays out:▬gold▬BTC▲EUR

   on script: the data-dependent line is repeated with no timeline, USDJPY stays within ±0.5% and the 10-year JGB within ±3 basis points. The hedging premium in DXY drains away, EURUSD holds 1.1434 and retests 1.1447; gold stays inside its 33.2 ATR near 4,039.0 because nothing new arrives for it.

3. If this branch plays out:▼gold▬BTC▼EUR

   Dovish branch: the BOJ pushes back on tightening, USDJPY rises 1.0% or more and DXY clears 101.50. A weaker yen carries the dollar higher, EURUSD slips through 1.1419 — this is precisely the branch that makes our call wrong; gold faces a dollar headwind and Bitcoin barely reacts.

Where we commit — and what would prove us wrong

Unlike the branches above, these carry no “if”. They are the lines graded against real prices below, and they are never edited after the print.

• EUR/USDBULLISHConviction 5/10

The most telling thing is not USDJPY but the 10-year JGB read alongside the US 30-year at pre-2008 highs: if both rise together, this stops being a yen story.