Between now and the Bank of Japan there are three windows worth having a rule for, and only one of them looks like a normal event. At 12:30 UTC four American releases land in the same minute. At 14:00 UTC a data release lands on the same minute as the New York option cut, where a large USD/JPY expiry sits directly on top of spot. And at 23:30 UTC tonight Japan’s national CPI shares its minute with a speech by the Governor of the Reserve Bank of Australia, roughly three and a half hours before the Bank of Japan decides. None of that is a market call. All of it is a question about when you are allowed to be exposed.
12:30 UTC is four releases, not one
Inside the next three quarters of an hour the United States publishes weekly initial jobless claims, the Philadelphia Fed manufacturing index for September, August housing starts and August building permits. Three separate readers list that cluster; the housing starts and claims items appear on all three, permits on two.
We are not publishing consensus figures for any of them, because we could not read a calendar that gave forecast and prior in a form we trust. Forex Factory’s calendar would not open for us this run. That absence is worth saying out loud rather than filling with a number: an article that hands you a forecast it could not verify is worse than one that admits the gap.
What matters operationally does not depend on the forecasts anyway. Four prints in one minute is not four chances to be right. It is one composite surprise with four ways of being ambiguous — claims soft and Philadelphia strong is a common shape, and the dollar reaction to that combination is not predictable from either component. A news filter that opens a separate suppression window per release is doing arithmetic you do not need; a filter that suppresses the minute is doing the job. If your system trades a breakout on the 12:30 candle, be honest that it is trading a coin flip about which of four numbers the market decides to weight.
14:00 UTC is the option cut, and this time there is a print inside it
One reader schedules August pending home sales for 14:00 UTC. That is the same minute as the 10am New York option cut. It is a second-tier housing release, so on most days you would ignore it; today it is sitting inside the only window on the board where a structural effect might otherwise have been readable.
Here is why that matters. Today’s expiry board carries USD/JPY 156.00, described by the publisher as both large and close to spot, and EUR/USD strikes at 1.1430, 1.1450 and 1.1500 with 1.1500 called the standout for size. Spot has spent the session straddling 156.00 — below the figure this morning, with the Tokyo session having run 155.87 to 156.31 — and EUR/USD has been near 1.1450, close to a seven-week low. Every strike on the board is within about fifty pips of the market.
Compare that with yesterday. This desk published at 06:16 this morning that the 16 September board had been left stranded by the Federal Reserve: USD/JPY 155.00 was 114 pips under the market, EUR/USD 1.1595 was 134 pips over it, AUD/USD 0.7130 was 40 pips away. We argued then that a pin is a property of the pair and not of the strike — a second-order hedging effect that only works at short distance and is flattened by any first-order flow.
Today inverts the first condition and fails the second. The distance condition is satisfied for the first time in a week. The first-order condition is not, because a scheduled release is landing in the same minute as the cut. So the honest conclusion is the uncomfortable one: the one day the board has been worth reading is the day a data print is scheduled to sit on top of it. Keep your expiry rule gated on distance, and keep it switched off when anything scheduled shares the minute.
The board we reported as absent was eight minutes old when we looked
That same article said, at 06:16 UTC, that no 17 September expiry board existed on the publisher’s orders index at 06:10 UTC, and allowed that one might appear later. One did not appear later. It was already there.
The 17 September board carries a publication timestamp of 06:02:25 UTC on its own page. Our check of the index happened at 06:10. The article was eight minutes old and the index did not list it. We read both pages ourselves, in the same session, and this is what they say.
That is the fifth documented instance of a metadata fault on this publisher, and the shapes are now consistent enough to be a rule rather than a run of bad luck: a headline figure that contradicts its own URL slug, a Canadian core measure sitting in a headline year-on-year field, a yield headline over a stale body, an index dating an article “3 hours ago” when the article says otherwise, and now an index that does not list an article it already hosts. In every one of those cases the numbers in the body were right. The labels, timestamps and index entries were not.
The operational rule follows directly, and it is a rule about your calendar rather than about the market: do not let an index decide whether a recurring daily publication exists. If a source publishes on a fixed cadence, request the dated item and treat a missing index entry as a fault in the index, not as evidence of absence. We got this wrong six hours ago at the cost of a whole section of an article, and the reason it went wrong is that we trusted metadata over cadence.
23:30 UTC: two events, one minute, and a central bank behind them
Under twelve hours from now, Japan publishes national CPI. In the same minute, by two independent calendars, the Governor of the Reserve Bank of Australia speaks. Then, roughly fifteen hours from now — and this is an approximation rather than a bound, because the Bank of Japan does not promise a time — the Bank of Japan announces a decision that sixty-six of sixty-eight surveyed economists expect to be a hike to 1.25%.
The trap is correlation, not timing. If you hold AUD/JPY, you have one position with exposure to both events in the same minute, and they are not independent: a hot Japanese print and a hawkish Australian speech push the cross the same way, while a soft print and a dovish speech push it the other, so your realised variance in that minute is larger than the sum you would compute treating the two as separate news items. Anyone running a basket of yen crosses and a basket of Australian-dollar crosses has just discovered they are the same basket for sixty seconds.
The BOJ compounds it rather than resolving it. Sixty-six of sixty-eight makes a hike close to unpriceable in the ordinary sense; what is not priced is a hike with dovish conditioning language, or no hike at all, and both of those are yen-negative into a pair that has already broken 156. We have no useful edge on which, and we are not pretending otherwise. What we do have is a schedule, and the schedule says the exposure decision has to be made before 23:30, not after.
What this does not tell you
We do not have the notional amounts on today’s expiry board. They live inside an image on the publisher’s page and do not extract as text, which is the fifth consecutive occasion this has happened. That means the words “large” and “standout” in this article are the publisher’s characterisation, not a number we have read, and you should discount our distance argument accordingly — a strike sitting on spot with trivial size does nothing at all.
We do not have consensus forecasts or priors for any of the 12:30 releases, and the 14:00 pending home sales time rests on a single reader. If that release is not at 14:00, the second section of this article loses its sharpest point, though the option-cut argument survives on its own.
We do not have a Japanese CPI consensus. One calendar gives a prior of 1.9% year on year and displays the release without an unambiguous year, which is the sort of thing that has burned this desk before, so we are publishing the time and not the number.
And we have not verified today’s spot levels against a second publisher since the Tokyo close. The 155.87 to 156.31 range and the sub-156.00 reading come from two different readers at two different times, which brackets the figure without pinning it.
