The Wrong End of the Curve.

Executive Summary

Between the 7 and 21 August closes, the market split the Treasury curve into two stories and priced only one of them. At the front, three backward looking prints, a July payrolls loss of 23,000 with a further 103,000 of downward revisions, a July CPI of 0.1 per cent on the month and 3.4 per cent on the year, and a flat July PPI, cut the CME FedWatch implied probability of a September hike from roughly 70 per cent at the start of August to roughly 35 per cent by the 21 August reading. At the back, every price that discounts the future moved the other way. The US diesel crack spread settled in triple digits for the first time in its history on 17 August, touching $102.20 per barrel against a pre-crisis norm of $15 to $25. Retail diesel reached $5.40 per gallon, a record for the date and 46 per cent above a year ago. The 30-year Treasury yield rose to 5.31 per cent on 17 August, its highest since 2007, in a selloff echoed in Germany, France, Japan and the United Kingdom, and on 19 August the US Treasury announced it would at least double its buybacks of 10 to 30 year debt, days after the national debt crossed $40 trillion. Spot gold closed the week near $4,607 per ounce, roughly 15 per cent above its mid-July low and up more than 11 per cent in August alone. Brent finished at $94.39 per barrel, a second consecutive weekly gain above 5 per cent, as Washington pivoted from blockade enforcement to what the Treasury Secretary called the toughest sanctions in history. In our view the front of the curve is trading the last print while the diesel market, the gold market and the long end are trading the next one. We expect the energy pass-through to reach the August and September inflation releases, we continue to expect the Fed to tighten before it eases, and, as Section 3 sets out, the gold position is not a bet on the rate path in either direction. We hold our positioning: gold as the largest allocation with the July add now well onside, selective non-US equity funded from crowded US exposure, and no US duration.

Since last month

The 17 July note made five checkable calls, and this month the scorecard is unusually clean in both directions. The gold call has been emphatic: we added in the $3,950 to $4,050 zone flagged in that note, and spot closed 21 August near $4,607 per ounce, roughly 15 per cent above the mid-July low near $4,000. The duration call has been equally clear: we stayed out of US duration through a selloff that took the 30-year yield to its highest level since 2007. The energy pass-through call remains right on mechanism and is not yet visible in the data: the July CPI and PPI energy lines fell again, because the July collection window closed before the August repricing, while the physical market moved decisively our way, with Brent up from $84.63 on 16 July to $94.39 on 21 August and the diesel crack spread at an all-time high. The tightening call is open and under pressure: the front end has cut the September hike probability to roughly 35 per cent, although three FOMC members dissented in favour of a hike at the 29 July meeting, the first time three policymakers aligned on one directional dissent since 2016. We hold the view that the Fed tightens before it eases, and we note plainly that the market currently prices against us on timing. The volatility call has again not been earned, with the VIX closing the week at 15.13; we do not raise it. The non-US equity book has been mixed: the Swiss Market Index set an all-time high of 14,669.51 on 11 August and runs near 21 per cent year to date, while the Ibovespa endured an eleven-session losing streak before recovering to 171,032, up 6.1 per cent year to date. The Kharg seizure we flagged as the July tail risk did not materialise; Washington instead pivoted to economic warfare, which changes the instrument and not the direction of the pressure on delivered energy.

1. Three prints took forty points off the hike

The repricing of the front end this month rests on three releases, each real, each backward looking. July payrolls, released 7 August, showed a loss of 23,000 jobs against a consensus near 85,000, the first decline since February, with May and June revised down by a combined 103,000. The unemployment rate fell to 4.1 per cent for the wrong reason, fewer people seeking work, and the employment-to-population ratio slipped to 58.9 per cent, its lowest since May 2014. July CPI, released 12

August, rose 0.1 per cent on the month and 3.4 per cent on the year, with core at 0.2 per cent and 2.5 per cent; shelter accounted for roughly two thirds of the monthly increase and the energy index fell. July PPI, released 13 August, was unchanged on the month, with final demand goods down 0.7 per cent and services up 0.2 per cent, taking the annual rate to 4.7 per cent from 5.5 per cent.

The market response was immediate. On CME FedWatch, the implied probability of a September hike stood near 70 per cent at the start of August, after the 29 July FOMC held rates at 3.50 to 3.75 per cent on a 9 to 3 vote with all three dissents in favour of a hike. The jobs report took the hold probability to roughly 60 per cent; the CPI print pushed the hike probability toward 42 per cent; by the 21 August reading it sat near 35 per cent. The two-year Treasury yield closed the week at 4.18 per cent, essentially where it stood in mid-July. This is a front end that has looked at three July data points and concluded the inflation episode is fading.

The difficulty is the collection window. The July price data was gathered while crude eased through late July and the first days of August on reports that a Hormuz transit deal was imminent. No deal materialised. Brent has since risen more than 11 per cent from its 16 July level, retail fuel set records in the very week the CPI was released, and Chair Warsh speaks at Jackson Hole on 28 August with his committee's three dissenters on the record. The August CPI, collected across the record-setting weeks, is released on 11 September, four days before the 15 and 16 September FOMC. The front end has priced the conclusion before the evidence arrives.

2. The margin, not the barrel

The consensus watches crude, and crude tells a reassuring story: Brent at $94.39 is a third below its April peak of $138. The shortage has moved somewhere the crude price cannot show, into refining, and the clearest evidence is a number almost no generalist reader tracks. On 17 August the US diesel crack spread, the premium of ultra-low sulphur diesel futures over WTI crude, touched $102.20 per barrel intraday and settled in triple digits for the first time in the history of the contract, per Bloomberg reporting of CME settlements. The pre-crisis norm for this spread was $15 to $25. The previous record, set in October 2022 in the early phase of the Russia-Ukraine war, sat in the high $80s to low $90s. A refiner is now being paid roughly four to six times the normal margin to turn crude into diesel, which is the market's way of saying that refined product, not crude, is what the world is short of.

Sources: Bloomberg reporting of CME ULSD/WTI settlements; AAA retail fuel data; EIA Weekly Petroleum Status Report; CNBC. Data as of the dates shown.

The mechanism from this table to the inflation data is short and physical. Diesel is the freight fuel: nearly every good in the CPI basket is delivered by truck, rail or ship burning it. Retail diesel at $5.40 per gallon against $3.70 a year ago is a 46 per cent input-cost increase to the entire distribution layer of the economy, and it passes into goods prices with a lag of one to two print cycles. The July CPI could not contain it because the July collection window closed before the records were set. The August print will be the first to sample it. Supply is not positioned to relieve the pressure: US distillate inventories sit 13 per cent below their five-year average even as refineries run hard, because exports are absorbing the output at a record 1.884 million barrels per day, and press reports indicate a substantial share of Russian refining capacity remains offline, a figure we cannot verify precisely and therefore do not stake the argument on. What we can verify is the price, and the price is the highest ever recorded.

The geopolitical setup points the risk one way. Roughly 20 million barrels per day of oil and products transited the Strait of Hormuz before the war; the US military said this week that escorted convoys have moved more than 660 million barrels since early May, implying transit in excess of 7 million barrels per day in recent weeks, and RBC Capital Markets estimates the war is still removing about 8 million barrels per day from supply. On 19 August the President threatened the most crushing economic operation ever taken against any country, on 20 August the Treasury Secretary promised the toughest sanctions in history and

said the objective is regime collapse, and on 21 August the President declared the Strait of Hormuz American territory at a campaign rally. Whatever one's view of the diplomacy, none of it reads as a fast track to normalised refining margins.

3. The buyer of last resort arrived early

The rates market spent the week telling a different story from the front end, and the fiscal authority blinked first. On 17 August the 30-year Treasury yield rose to 5.31 per cent, its highest since 2007, and pushed toward 5.34 per cent intraday the following day; the 10-year touched 4.74 per cent. The move was global: German 30-year yields reached their highest since 2011, French long yields their highest since 2008, and Japan's 10-year yield its highest in three decades. In the same week, the US national debt crossed $40 trillion, five months after crossing $39 trillion. Then, on 19 August, the Treasury announced it would at least double the maximum size of its liquidity support buybacks of 10 to 20 year and 20 to 30 year debt, from $2 billion to at least $4 billion per operation, effective 9 September through 4 November. Yields fell on the announcement, the 30-year by 9 basis points to 5.196 per cent, and gave most of it back by Friday's close at 5.24 per cent.

The instructive market response was not in bonds; it was in gold. Gold jumped more than 4 per cent on the announcement day, December futures reached $4,569.40 on Friday, their highest since mid-May and a fifth consecutive weekly gain, and spot closed the week near $4,607. Here is the reading the gold market applied, and in our view it is the correct one. A Treasury that buys its own long bonds while running large deficits is not reducing debt; it is removing duration from the market and funding the removal with new issuance, which is the operational signature of quantitative easing, executed by the fiscal authority, in the same month three members of the monetary authority argued for a hike. The two arms of US policy are now pulling the curve in opposite directions, and the buyback window runs through 4 November, which is also the final stretch of the midterm campaign.

This resolves the puzzle in gold's five-week, $600 rally, which the consensus narrates as an expected-real-rate trade: soft CPI, lower expected real rates, gold up. The channel is real, but the August sequence does not support it as the driver. If falling expected real rates were doing the work, gold should have risen on the soft CPI of 12 August and fallen on 17 and 18 August, when nominal long yields backed up to two-decade highs with no matching inflation print, which is a rise in real long rates. Instead gold rose on all three kinds of day, and jumped hardest on the buyback announcement of 19 August. It rallied on days when expected real rates fell and on days when they rose; the constant across all three is the fiscal signal, a statement about the ceiling on the real return bondholders will be permitted to earn. Nor, in our view, does a hike undo this: a 25 basis point move made in response to a 46 per cent rise in the price of diesel does not raise expected real rates, it barely defends them. The configuration that hurts gold is a central bank tightening into falling inflation; that was the June and July configuration, it took the metal to $4,000, and it is where we added. It is not the configuration on the screen today. Beneath it, the strategic bid we documented last month became a record: per the World Gold Council's Gold Demand Trends release of 30 July, central banks purchased a net 288.9 tonnes in the second quarter, up 62 per cent on a year earlier and the strongest second quarter in the data series, with the People's Bank of China adding 33 tonnes, its largest quarterly purchase since late 2023 and its twenty-first consecutive month of buying. The buyers who accumulated through the drawdown at $4,000 are the same buyers at $4,600, because their mandate is reserve composition, not entry price.

4. Portfolio positioning

Positioning is unchanged in direction from the July note. This month we hold rather than add: the July additions are in place and marked well onside, and we see no case for chasing strength.

Gold remains our largest single allocation. The add executed in the $3,950 to $4,050 zone flagged in July now sits roughly 15 per cent onside with spot near $4,607. We hold the enlarged position and do not chase; the record central-bank bid and the fiscal signal described in Section 3 are the reasons the weight stays where it is. The position is deliberately not a bet on the rate path: it rallied this month on falling and rising yields alike, and a hawkish repricing that knocks the financial bid would, as in July, be treated as an entry rather than an exit.

Equities. We remain underweight the crowded US market, which closed at 7,674.37 on the S&P 500, up 12.1 per cent year to date on a price basis after a week of technology-led losses. The Swiss Market Index closed at 14,456.98, up roughly 21 per cent year to date, having set an all-time high of 14,669.51 on 11 August; the overweight is retained. Brazil is retained as the decoupling trade with eyes open: the Ibovespa closed at 171,032, up 6.1 per cent year to date, after an eleven-session losing streak driven by election positioning, and rallied 1.85 per cent on Friday as US yields eased; the Selic stands at 14.00 per cent

with the Copom easing into disinflation while the developed world debates hikes. China remains structurally underweight; we do not add.

Fixed income. We continue to avoid US duration entirely. The buyback is a liquidity backstop, not a change in net supply, and a 30-year at 5.24 per cent with a 10-year at 4.69 per cent does not compensate for record refining margins passing into the price data, a Fed whose dissenters want a hike, and a Treasury whose response to the long end is to buy it. The two-year at 4.18 per cent prices out the hike we still expect, and offers no cushion if the August data forces the question back open.

Volatility. We continue to expect higher realised volatility than the market prices, and we say plainly that this call has not paid for two cycles running. A VIX of 15.13 discounts a live conflict, an all-time-record fuel margin, a three-dissent Fed five days from Jackson Hole, and a long end that required an emergency fiscal backstop this week. We retain the position and do not raise it.

Risk management. The scenario the equity market is not priced for is the one our own thesis implies. If the three dissents harden into an actual hike at the September or October meeting, delivered against an implied probability of roughly 35 per cent, we see the US equity market as exposed to a correction of at least 10 per cent, in our estimation, from a tape trading at a VIX of 15.13 and 12.1 per cent year to date on the assumption of a hold. We therefore either carry index hedges or sell strength rather than chase it, and we hold the proceeds to redeploy into our reflation-driven trades, the energy-linked and non-US exposures described above, on the pullback rather than at current levels. The discipline is the same one that governed the July gold add: the correction our thesis predicts is the entry it also provides.

5. Conclusion

The controlling contradiction of this note is a front end pricing away the Fed hike on July data while every forward-looking price, the diesel crack, the pump price, the long yield, and gold, set records in the same ten days. For the market's current pricing to be right, three things must hold together. The sanctions campaign must produce a rapid Iranian capitulation and a full Hormuz reopening fast enough to collapse refining margins before they reach the August and September CPI prints. The Fed must read through record pump prices on the strength of a softening labour market, against the recorded objection of three of its own members. And the Treasury must succeed in capping the long end at no inflationary cost, with a buyback funded by new issuance. Each is individually possible. In our view their joint probability is materially lower than a VIX of 15.13 and a 35 per cent September hike probability imply. We are positioned for the alternative: gold as the largest allocation, selective non-US equity funded from crowded US exposure, no US duration, and patience on the volatility book.

Important Information

This document is a research note prepared by the Neovision Group for institutional and professional clients. It is not investment advice, an offer, or a solicitation to buy or sell any security or financial instrument. All figures cited are sourced from public information.

The Neovision Group comprises Neovision Wealth Management Limited (regulated by the Financial Services Regulatory Authority in the Abu Dhabi Global Market), Neovision Investment Fund Management (NIFM, regulated by the Capital Markets Authority in the onshore UAE), and Héritage Riviera SA (regulated by FINMA in Switzerland).

Sources referenced in this month's note include: US Bureau of Labor Statistics (Employment Situation for July 2026, released 7 August; July 2026 CPI release of 12 August; July 2026 PPI release of 13 August); US Department of the Treasury (press release of 19 August 2026 on increased sizes of nominal long-end liquidity support buybacks); Federal Reserve (FOMC statement of 29 July 2026 and minutes released 19 August 2026); CME Group FedWatch Tool; US Energy Information Administration (Weekly Petroleum Status Report and Short-Term Energy Outlook of 11 August 2026); AAA retail fuel data; World Gold Council (Gold Demand Trends, Q2 2026, released 30 July 2026); Bloomberg; CNBC; Reuters; Axios; CNN; CBS News; NBC News; NPR; Quartz; TradingEconomics; Kitco; S&P Dow Jones Indices; Yahoo Finance; SIX Swiss Exchange; B3; RBC Capital Markets commentary as reported by CNBC; Banco Central do Brasil.

Market data as of close, 21 August 2026, unless otherwise stated. Forecasts and views expressed represent the opinion of the Neovision Group at the date of publication and are not guarantees of future outcomes. Past performance is not indicative of future results.