The Daily Brief

Sterling's rally has been running against its own positioning. Speculative net shorts in the pound built up ahead of the Bank of England's July meeting and have not been meaningfully unwound since, which leaves a currency drifting to three-week highs while the market that trades it is still leaning the other way. Cable closed Monday's session at 1.3501 and trades near 1.3510 this morning, its firmest since 15 July, without a single British catalyst behind the move.

The reason the short base has survived a rising price is the shape of the Bank's July communication. The vote split came in more hawkish than the market had positioned for, but the Governor's tone afterwards was cautious and gave the pound nothing to hold. A formally tighter stance paired with soft rhetoric produces exactly this: no domestic bid, no domestic seller, and a price that simply reflects whatever the dollar is doing that day.

Thursday's preliminary Q2 GDP estimate is the first thing capable of changing that, and the interesting feature is how far apart the forecasts sit. Consensus runs between 0.2% and 0.4% quarter on quarter against 0.6% in Q1, with the annual rate seen near 1.6%. Dispersion that wide on a first-tier print, with net shorts still in place, makes the reaction function lopsided: a soft number confirms a view the market already owns, while anything at or above the top of the range forces a covering move into a thin August book.

The euro cross carries the more honest signal. GBP/EUR sits near 1.1673, above a rising fifty-day average at 1.1651 but well below July's 1.1810 peak, and the unwind from that high has come as the gap between Bank Rate and the ECB deposit rate has started to narrow. That spread was the entire basis of sterling's 2026 outperformance. Its compression matters more to the pound's medium-term path than one quarter of British growth data.

At 1.3510 cable is at the ceiling of the 1.34 to 1.3558 band it has held since late July, roughly 50 pips below the July high and 110 above the floor. Sitting at the top of a range with the short base intact and the catalyst still 48 hours away is an unstable arrangement. The upside break needs only a soft US core print to trigger the covering, while the downside case requires Thursday to undershoot a consensus that is already low, which means the asymmetry currently runs against anyone waiting at these levels to buy dollars.

Wednesday's inflation print arrives with a forecast band that says more than the forecast. Headline CPI is seen at 3.4% year on year from 3.5%, but the range around that number runs from 3.2% to 4.1%, and monthly estimates span zero to 0.7% against a 0.1% consensus. A 90 basis point spread on the most-watched print in the calendar is not a forecast, it is a collective admission that nobody knows which force is winning.

Both forces are visible in the June base. That month's 0.4% monthly decline was almost entirely a gasoline effect from the brief de-escalation, and July's month has crude moving hard in the opposite direction. Core is the cleaner read: 0.2% monthly expected after a flat June, taking the annual rate to 2.5% from 2.6% and its lowest since January. Bank estimates cluster tightly at 0.20% to 0.24% on core, with one house flagging that a 0.24% core PCE equivalent at 3.3% annually would be consistent with a September hike rather than against it.

What the dollar did on Monday is the part that deserves attention. The index had fallen 0.5% to 99.4 on Friday's payrolls contraction, near a two-month low, and then rose 0.27% to close at 99.81 while the ten-year lifted from 4.651% to 4.71%. A currency and a curve moving together and upward after a labour market shock is not a growth trade. It is an energy pass-through trade, and it was reinforced by a regional Fed president saying that several further hikes may be required to bring inflation back to target.

Pricing has moved the tightening rather than cancelled it. September hike odds sit near 46%, down from around 64% a week ago, with a hold at roughly 54%. That leaves the September meeting genuinely two-sided going into a print whose forecast range straddles both outcomes. The practical bar has shifted: it now takes a firm core reading to restore a hike as the modal expectation, while a soft core has room to compress hike probability across the rest of the year, not just for September.

At 99.787 the index sits mid-range between 99.30 support and the 100 handle, 1.4% weaker over the month and around 40 basis points above Friday's low. The risk is two-sided but not evenly weighted. The top of the CPI range puts 100 and then last month's highs back in play inside a session, while the bottom takes out 99.30 and reopens the summer lows, and the width of the forecast band argues for a larger move in either direction than a 0.1% consensus would normally justify.

The sell-side has quietly changed its mind about the rand. Part of the street now argues the currency is undervalued and sees near-term strength, supported by the Reserve Bank's credibility on its 3% target, and a segment of it has moved further to suggest the next policy move is a hike rather than a cut. Three weeks ago the same currency was making a three-month low at 16.98 on the back of an unexpected July hold, so this is a genuine repricing of the carry argument rather than a restatement of it.

Monday's session gave that view no local test. With the JSE and domestic banks closed for the public holiday, the rand was marked offshore around 16.16 and has eased to near 16.19 this morning, a fifth of a percent softer against a firmer dollar. That is a small loss from the firmest level since June, but it was made without domestic price discovery, and local participation returns today into two data points.

The terms-of-trade leg is currently doing the heavy lifting. Gold has pushed through $4,400 to near $4,412, a nine-week high, platinum group metals remain firm, and precious metals together account for roughly a fifth of South African exports, so bullion strength feeds the trade account directly. The offset has not disappeared: crude is higher again on the same Hormuz stalemate, and for a net energy importer that is a rising import bill with a fuel pass-through still to land in the inflation series. The rand is holding because the metals leg is louder than the crude leg, which is a net of two large gross moves rather than a calm currency.

Today's prints set the domestic frame. Q2 unemployment lands at 10:00 BST against 32.7% in Q1, a quarter in which employment fell by 345,000 and labour force participation dropped to 59.0%, its lowest since 2022. June manufacturing production follows at midday after a 4.3% annual contraction in May, with the July purchasing managers' index at 46.8 and contracting for a second month. Neither release changes the repo rate before September, but together they build the growth case that the hawkish camp will have to answer when the committee does meet.

At 16.19 the rand sits close to the strong end of a range whose other edge is the 16.98 low of 24 July, a spread of nearly 80 cents opened in three weeks, and roughly 5% stronger than the consensus year anchor near 17.00. The asymmetry still favours the currency while gold holds its $4,400 handle and the US core print behaves, and it inverts quickly if Wednesday prints toward the top of its range with crude near $87. For rand-denominated cost bases this is the more favourable half of a wide range, with no domestic policy event scheduled to defend it before September.

Half. That is roughly the share of the yen's gains from a coordinated US and Japan intervention that the currency has already handed back. When a joint operation of that weight is unwound by half within days, the market's verdict is that the intervention corrected a price rather than a cause, and the cause is still running.

The cause is crude. Brent trades above $87 on one published series and at $84.42 on ICE October futures, up around 1% on the session, with the strait effectively closed into a sixth month of conflict and Houthi disruption complicating the Red Sea alternatives. Washington has rejected Tehran's demand for war compensation as part of any reopening deal and signalled a preference for letting economic pressure build rather than escalating militarily. Not a barrel has changed hands differently. What is being marked, repeatedly and in both directions, is the probability of an agreement.

Equities absorbed that without breaking. The S&P 500 closed at 7,753.11, down 0.06% from Friday's record 7,757.64, after trading as high as 7,773.76 and as low as 7,743.11. The Nasdaq Composite lost 0.32% to 26,605 and the Dow eased 0.11% to 53,976, with energy leading every sector and technology and property lagging. Nvidia fell on reports that major Wall Street firms are arranging a $500 billion funding package around it. An index 40 points from its 52-week high with energy leading and megacap technology selling is defensive rotation inside a flat tape, not a flat market.

Gold's behaviour is the tell. It has cleared $4,400 to trade near $4,412 this morning, a nine-week high, after rising through a Monday session in which both the dollar and the ten-year yield also rose, and it has Chinese official buying behind it after the central bank added the most bullion since October 2023 in July and first-half imports reached roughly 820 tonnes. Bullion rising alongside both the dollar and real yields is not the textbook pairing. It says the bid is inflation protection and reserve diversification rather than a rate-differential trade, which is a structurally more durable buyer than a positioning move.

Gold now sits roughly 21% below its January record near $5,600 but less than $90 from the $4,500 level the market has been openly discussing, while Brent's two published series straddle $84 to $87 and the S&P sits 40 points off its high. The cross-asset tension is clean: gold and equities are pricing a Federal Reserve that stops, while crude and the front of the curve are pricing one that cannot afford to. Wednesday's inflation print is the only scheduled event this week with the standing to resolve which of those is wrong.