AUD volatility around RBA decisions: a 48-hr playbook

The standing problem

Eight times a year, the Reserve Bank of Australia announces its cash rate decision at 2:30pm Sydney time on a Tuesday, and AUD/USD moves. Sometimes it moves a lot. Most Australian importers and exporters who use forwards know this in the abstract but treat it as a market problem rather than a treasury problem. It's both. The 48 hours before an RBA decision is the window where execution-timing decisions stop being neutral and start being directional, whether you intended them to be or not.

This article looks at what actually happens to AUD in the run-up to and through an RBA decision, what the real options are for a treasury function in those 48 hours, and what to do if you have a committed FX exposure to hedge while a meeting is approaching.

What the RBA actually does, in two paragraphs

The RBA Board meets eight times a year (February, March, May, July, August, September, November, and December) under the post-2024 reformed schedule. The decision is announced at 2:30pm Sydney, accompanied by a Statement on Monetary Policy at each meeting (also a post-2024 change), and followed by a press conference by the Governor at 3:30pm.

The market trades the decision, the statement, and the press conference as three separate events. The cash rate move itself is usually mostly priced in. The surprise lives in the statement language and the Q&A. AUD/USD volatility typically peaks during the press conference, not at 2:30pm.

What actually happens to AUD in the window

The pattern across recent meetings is reasonably consistent.

In the 24-48 hours before the decision, AUD/USD bid-offer spreads in interbank markets stay roughly normal, but liquidity at any particular price level thins out as positioning is reduced and risk desks pull back. Implied 1-week at-the-money AUD/USD vol rises 0.5-2 vol points into the meeting. That doesn't change what a forward costs, since the forward is symmetric, but it does mean the option-equivalent value of the certainty a forward provides is at a recent local high.

In the 30 minutes before 2:30pm, screen liquidity collapses. Spreads widen. The dealing screens look fine, but the size you can actually deal at the screen rate shrinks. For a sub-$1M ticket, you'll usually still get filled cleanly at most providers. For anything larger, expect to leak some pips.

At 2:30pm and again at 3:30pm, AUD/USD typically moves 0.3-1.5% on a vanilla decision, occasionally 2%+ on a hawkish-or-dovish surprise. The move is bidirectional. The market is rarely positioned cleanly one way going in. Within 60-90 minutes the new range is usually established, and liquidity normalises by the next morning.

If you've never traded an FX desk on RBA day, the most useful thing to know is that the window from 2:00pm to 4:00pm is where execution risk is concentrated, not the rest of the 48 hours. Most of the day on Tuesday is fine. So is most of Monday and most of Wednesday morning.

The temptation: trying to time it

Every CFO who has ever sat through a meeting where AUD moved 1% in their favour has had the thought: "What if we'd waited?" Every CFO who has sat through one where AUD moved 1% against them has had the opposite thought. Both thoughts are the same mistake.

The proposition that you can systematically time RBA decisions assumes you have a better read on the meeting outcome than the people quoting the swap curve. You almost certainly don't. RBA-watcher economists employed full-time at the major banks build models specifically for this, and they disagree with each other heading into most meetings. The implied probability of a 25bp cut versus a hold, extracted from the OIS curve, is the actual market consensus. It's what the people who do this for a living think the answer is.

Worse, even if you have a view, the AUD reaction depends not only on the decision but on the statement language and the press conference, and the relationship between any of those three and the spot move isn't deterministic. A "hawkish hold" can move AUD 1% higher. A "dovish hold" can move it 1% lower. A 25bp cut delivered with hawkish forward guidance can move it sideways. You'd need to be right on all three legs to actually capture the trade, and you'd need to be right consistently to call it a strategy rather than a bet.

The honest answer for a treasury function with a committed exposure is that trying to time RBA is just trading wearing a hedging hat, and it's usually a losing trade by the time you account for the wider spreads and the operational drag.

What actually works in the 48-hour window

Three rules cover most situations.

First, if you have a committed exposure that's due to be hedged on a calendar schedule, hedge it on the calendar schedule. Don't push the layer to "after the meeting" because you have a hunch. The framework you're using exists to remove discretionary decisions from execution. A 48-hour deferral on a ratio basis is fine. A 48-hour deferral because you're trying to time the meeting is the entry point to running directional views by accident.

Second, if you have flexibility on when to execute, prefer Monday or Tuesday morning to Tuesday afternoon or Wednesday morning. Spreads are tighter, dealing rates are cleaner, and you're not trying to deal into the period where every desk is hedging their own book. The difference is usually a few pips on a $1M ticket. Small but real, and it compounds across a year of execution.

Third, if your dealing process can support it, instruct the desk to step aside the 30-minute window from 2:00pm to 2:30pm and the 30-minute window from 3:30pm to 4:00pm Sydney. Resume normal execution from 4:00pm. This isn't market timing. It's avoiding the windows where execution quality predictably degrades.

The combined effect of these three rules is that RBA day stops being a treasury event. You execute around it rather than into it, and you don't try to capture the move.

A concrete 48-hour framework

For a treasury function running a layered hedging programme with monthly execution, a meeting-week protocol works like this.

The week before the meeting, the treasurer reviews the upcoming month's required execution. If the calendar execution date falls on RBA Tuesday or the Wednesday morning following, the execution is brought forward to the previous Friday or to the Monday before. Not delayed. Brought forward. The bias is to execute early rather than late, because deferral is the path that introduces directional risk by the back door.

If the execution is for a flexible window forward, where the exact pricing day matters less, the trade is normally booked on Monday morning. If it's a fixed-date forward, the same.

If a payment-driven execution falls on the meeting day itself, say a supplier payment is due Wednesday and you're spot-converting on Tuesday, the conversion is brought forward to Tuesday morning before noon Sydney, or pushed to Wednesday after liquidity normalises.

The outstanding hedge book is reviewed for any positions maturing on or close to the meeting day, where roll execution would land in the volatile window. Those are rolled in the week before, on a normal trading day.

Triggers, if you use them, are reviewed. A trigger to add a layer "when AUD/USD touches 0.6650" left live through the meeting is asking for a fill at a level that isn't representative of the post-decision range. Triggers are typically suspended from Tuesday morning to Wednesday morning, then reactivated.

The whole protocol is documented in the hedging policy as a one-paragraph rule: execution scheduled on RBA decision days is brought forward to the previous trading day; the two 30-minute windows around 2:30pm and 3:30pm are excluded; triggers are paused from market open Tuesday to market open Wednesday.

That's it. No view-taking, no timing, no leaning into or against the meeting. Just an avoidance protocol, applied consistently.

A worked example

Consider a treasurer running a $20M annual import programme on a 75/50/30/15 layered framework. The September RBA meeting falls on a Tuesday. The monthly layer execution is scheduled for the first Tuesday of the month, which is the meeting day.

Without a meeting-week protocol, the trades go to market on Tuesday morning, perhaps run through 2:30pm if the broker doesn't have a stand-aside policy, and the average effective rate on the layer's $1.5M USD notional reflects whatever AUD did across the day. That's fine on a low-volatility meeting and meaningfully bad on a surprise meeting.

With the protocol, the trades are executed on Monday morning instead. The mid-rate at the time of execution is whatever it is, but the spread is normal, the desk has full liquidity, and Tuesday afternoon's volatility doesn't bear on the result. The trade is logged on Monday afternoon, the audit trail is clean, and the treasurer isn't fielding a Tuesday-night phone call from the CFO asking why AUD just dropped 1.2%.

Now overlay a hawkish surprise. RBA holds when the OIS curve was pricing a 60% chance of a cut. AUD/USD jumps 1.4% from 0.6520 to 0.6611 across the press conference. The Monday execution at 0.6520 looks 1.4% worse than what could have been achieved Wednesday morning. But the comparison is the wrong frame. Treasury didn't capture the move because treasury wasn't trying to capture the move. The trade did exactly what it was supposed to do: add a layer of certainty against a forecast import payment, at a known cost, without exposing the operational team to two-hour windows of unpredictable execution quality.

Now flip it to a dovish surprise. RBA cuts 25bp where the curve was pricing a hold. AUD/USD drops 1.1% from 0.6520 to 0.6448. The Monday execution at 0.6520 now looks 1.1% better than Wednesday morning. The treasurer didn't earn that either. The protocol just happened to fall on the right side of the move this time. Across enough meetings the directional outcomes wash out. The protocol's actual contribution is the avoided execution-quality cost, which is consistent rather than directional.

Common mistakes

A few patterns turn up repeatedly.

The first is anchoring on the most recent meeting outcome. A surprise hawkish meeting last cycle leads the treasurer to defer this cycle's execution, expecting another move. Markets don't reliably repeat patterns at this level. The recency bias is expensive when it's wrong.

The second is dealing in the screen window between 2:30pm and 4:00pm because the screen rate looks attractive. The screen rate is real, but the size behind it isn't. Filling a $3M USD ticket at 3:15pm Sydney on a meeting day is the kind of trade that looks fine until post-trade analysis shows the actual fill 8 pips wide of the screen mid.

The third is forgetting overseas branches and subsidiaries. A treasury team running domestic execution discipline can have an overseas finance manager with their own dealing line, blissfully transacting through the volatile window with no policy guardrails. The protocol needs to apply to every dealing point in the group, not just the head office.

The fourth is failing to update the protocol for the post-2024 schedule. The RBA used to meet 11 times a year on the first Tuesday of every month except January. Now it meets 8 times. Hedging policies written before the schedule reform sometimes still reference monthly meetings, leading to protocol activation on dates that aren't actually meeting days.

The fifth is treating every meeting as equivalent. Meetings where the OIS curve is pricing 80%+ probability of a hold and the language is widely expected to be unchanged tend to produce small moves. Meetings where the curve is split 50/50 between hold and cut, or where consensus is divided on the language, produce the larger moves. The protocol can be applied uniformly, which is its strength, but the treasurer should still be aware which type of meeting is approaching, because the size of any execution-quality cost in the volatile window scales with the expected move.

Closing

RBA decisions aren't a treasury opportunity. They're a treasury timing problem with a small number of practical answers.

The answer isn't to forecast the meeting, lean into or against the consensus, or pull or push hedges based on a view of the cash rate path. The answer is to protect execution quality by avoiding the volatile windows, bringing forward calendar-driven executions that would otherwise fall on the meeting day, and pausing any trigger logic across the meeting itself.

A finance team that runs this protocol consistently across eight meetings a year removes a real source of avoidable execution cost without taking any directional view. The protocol is two sentences in the hedging policy. The savings are small per meeting but real, and they compound across the programme. The treasurers who do this well decided early that meeting day isn't a chance to be clever. It's a window to be careful around, and once the rule is set, it never has to be debated again.