Crude is holding near $90 after renewed strikes on Gulf energy infrastructure and continued US–Iran exchanges around the Strait of Hormuz. Iran has signalled it will declare a new restricted zone in the Gulf within days, redrawing the shipping corridor. The risk premium has been in the price for over a week now, and it has stopped behaving like a headline and started behaving like a regime.
If you run automated FX and your reaction to this is “I don’t trade oil,” the exposure is already on your book. You just did not open it deliberately.
Oil stopped being a commodity story and became a rates story
The transmission is not oil into FX. It is oil into policy expectations, and policy expectations into FX. That path is currently live at three central banks simultaneously:
- ECB, Thursday 10 September at 12:15 UTC. A 25bp hike to a 2.5% deposit rate is widely expected, and the reasoning being offered is explicitly the energy shock.
- Federal Reserve, 16 September. The September hike debate came back to life on Friday’s payrolls, but the argument that keeps it alive is energy keeping the inflation path unresolved.
- Bank of Japan, 18 September. A 25bp move is largely priced, with an economy that imports essentially all of its crude.
That is three decisions in nine days, with a common input none of them controls. A single tanker headline is now a simultaneous partial repricing of EUR, USD and JPY. It does not have to arrive politely, during a session, or on a calendar.
Three positions, one trade
The practical consequence is a sizing failure that looks like diversification.
Consider a book with exposure across the oil-sensitive majors — CAD as a crude exporter, JPY as a crude importer, the commodity crosses on the demand side, EUR through the ECB’s energy-driven policy path. On a normal week these carry different drivers and a portfolio of them is genuinely diversified. This week they share a single dominant input.
The failure mode is arithmetic and it is boring: a disciplined 1% per trade becomes an undisciplined 3–4% on one Gulf headline. Not because any rule was broken, but because the rule measures the wrong unit. Risk-per-trade assumes the trades are separate events. When one input drives all of them, only risk-per-theme means anything.
Worth doing before London, in this order:
- List open positions by what would have to be true for them to lose together, not by symbol.
- Check whether your correlation window is long enough to have noticed the regime change. A 200-bar correlation on a pair that has been decoupled for six months will tell you these are independent right up until the day they are not.
- Decide the theme cap now, while nothing is happening. A cap chosen during a move is not a cap.
An unschedulable event breaks a schedule-based filter
Every news filter we have seen — ours included, in earlier versions — is built on the same assumption: events have times. CPI has a time. A central bank decision has a time. You blackout around the timestamp and you are protected.
A strike on an oil facility does not have a timestamp. Neither does the announcement of a restricted zone “in the coming days.” The entire architecture of a calendar-driven filter is inapplicable to the largest source of gap risk currently on the board.
There is no clean fix, but there are honest partial ones:
- Condition on realised volatility, not just on the clock. A filter that can only be triggered by a scheduled entry is blind by construction. One that also widens when short-horizon range expands is at least reactive.
- Size for the gap, not for the stop. Weekend and off-session gap risk is elevated while this is live. A stop is a request, not a guarantee, and a request is not honoured through a gap.
- Treat the oil-sensitive theme cap as the primary control. When you cannot time the event, the only remaining lever is how much is on when it lands.
- Assume slippage is worse than backtested. Backtests calibrated on the last two quiet years do not contain this regime. Where they filled you at the level, live will not.
What this does not tell you
It does not tell you where oil goes, and it does not tell you whether the ECB, the Fed or the BOJ actually moves. The energy-shock reasoning currently attached to all three is the market’s explanation, and market explanations are frequently the thing that gets revised first.
It does not tell you that these correlations will hold. They are elevated now; regimes end without notice, and a correlation matrix is a description of the recent past, not a constraint on the near future. The argument here is for measuring the shared exposure, not for assuming a particular sign.
And it is emphatically not a directional call on any pair. The whole point is that this is a sizing and exposure question. If you came looking for a level, that is what Signals is for, and even there it describes market response rather than predicting it.
Related
- FX Events calendar — the ECB, FOMC and BOJ windows, with times
- Signals — direction and strength on the majors, from a signal-only analyzer
- EA Track Record — verified performance, including through high-volatility stretches
