The 30-Year Real Yield Hits 3.06% (17 August 2026): Why Gold Fell in 2026 — and Why a 2007-High Long Bond Did Not Stop It

On Monday 17 August the long end of the US Treasury curve broke to its highest level in nearly two decades, and gold went up. The 30-year nominal yield closed at 5.31% against 5.25% on Friday — the highest since 2007, per Bloomberg — while the 30-year inflation-protected real yield rose the same six basis points to 3.06%, the highest reading of 2026. Gold rose about 0.45% to roughly $4,395.38, with December futures opening at $4,440. On the conventional account this is impossible: the real yield is gold's opportunity cost, and gold's opportunity cost just hit a multi-decade high. The resolution is in the shape of the curve. The front end barely moved — the one-month bill unchanged at 3.79%, the two-year up two basis points to 4.19%, the five-year real yield up one to 2.13% — and the front end is where the expected path of policy lives. Bullion fell roughly 28% from January's record through an active Middle East war because that war's energy shock lifted US inflation, pushed the Fed from expected cuts to expected hikes, and drove the real yield gold competes with to a seventeen-year auction high of 2.438%. Every leg of the recovery since has come from that expectation being priced back out — and on 17 August, a long-end shock that left the expectation untouched left gold untouched with it.

This is the seventh test of the same channel, and it isolates the mechanism more cleanly than any of the six before it. The 14 August session had already shown gold rising on a day its 10-year real yield rose — but that day came with an escape hatch, because the 10-year breakeven widened from 2.24% to 2.27% at the same time. A market that cuts near-term hike odds while demanding more compensation for long-run inflation has done two separate favours to a zero-coupon asset, and the widening breakeven did some of the explanatory work.

On 17 August there was no such hatch. Subtract the real yield from the nominal at the same maturity and the thirty-year breakeven was unchanged at 2.25% — 5.25% minus 3.00% on Friday, 5.31% minus 3.06% on Monday, both legs up six basis points in lockstep. At ten years the breakeven moved a single basis point, to 2.28% from 2.27%. So the long-end selloff was almost purely real: not an inflation-expectations event, but a repricing of what investors demand to lend for thirty years. That is a term-premium and fiscal question, and gold declined to treat it as a monetary one.

The wider scoreboard is still a drawdown. Spot near $4,395 on 17 August sits about 21.6% below the record $5,608.35 set in January, and a 10-year real yield of 2.44% remains near the stiffest competition bullion has faced since 2008. Nothing here says the 2026 regime is over — a roughly 31% September hike probability is not a cut probability. It says the regime is legible, and now legible with a distinction the earlier tests could not draw: seven tests, in both directions, from war headlines, payrolls, oil, CPI, PPI, consumer spending and now a long-end fiscal shock — and every one of them answered by the front of the curve rather than by the back of it, or by the war.

What actually happened on 17 August: the long end broke, and gold did not care

The 14 August version of this post argued that gold trades the expected path of real rates rather than today's level. Monday 17 August delivered the experiment that separates those two claims properly, because it moved the level hard while leaving the path alone.

Start with the primary data. The US Treasury's daily nominal yield curve closed the 30-year at 5.31%, up from 5.25% on Friday 14 August, and the 20-year at 5.30% from 5.25%. Bloomberg reported the long bond's level as the highest since 2007. On the real yield curve — the one that matters for a metal that pays nothing — the 30-year inflation-protected yield went from 3.00% to 3.06%, and the 20-year from 2.78% to 2.84%. That 3.06% is the highest reading of 2026 in the series, above the 3.03% set on 31 July and well above February's low of 2.43%.

Then look at what gold did. Spot rose toward $4,400, trading around $4,395.38, up about 0.45% on the day, with CNBC quoting spot at $4,387.86 at 9:00 a.m. Eastern and December futures opening at $4,440, up 0.1% on Friday's close. Gold's opportunity cost, measured at thirty years, reached its highest level of the year, and gold appreciated.

There is a second discipline to apply before calling this a real-yield event at all, and it is the one that makes 17 August a stronger test than 14 August. Nominal yield minus real yield at the same maturity gives the breakeven inflation rate — the compensation the market wants for expected inflation over that horizon. Run it at both ends of the move: at thirty years, 5.25% − 3.00% = 2.25% on Friday and 5.31% − 3.06% = 2.25% on Monday. Unchanged. At ten years, 4.68% − 2.41% = 2.27% became 4.72% − 2.44% = 2.28%. One basis point.

That arithmetic matters because it closes off the explanation the previous session leaned on. On 14 August the real yield rose two basis points and gold rose, and the available resolution was that the breakeven widened at the same time — the market cut near-term hike odds and demanded more inflation compensation, two separate favours to a zero-coupon asset. On 17 August the second favour was absent. Inflation compensation was flat, the long-dated real yield rose six basis points, and gold rose regardless. Strip out the inflation leg and the policy-path leg is all that is left, which is the argument this post has been making since July, now with the alternative explanation removed rather than merely outweighed.

The oil leg, which is the bridge this post has argued the Middle East conflict travels across to reach bullion, was firm but not dramatic on the day. CNBC had Brent futures up 55 cents at $89.07 and WTI up 17 cents at $82.57 in Monday morning trade, after both contracts gained more than 5% the previous week following attacks on tankers operated by the Abu Dhabi National Oil Company in the Strait of Hormuz and on a Saudi Aramco refinery. Iran's foreign ministry spokesperson Esmaeil Baghaei said talks with Oman were continuing and were taking a long time given the complexity of the subject, while foreign minister Abbas Araqchi said over the weekend that Iran had not decided to resume talks with the United States. SEB Research's Bjarne Schieldrop told CNBC that prices were unlikely to move substantively higher without a halt in the current night-time flow of crude out through Hormuz or a closure of the Bab el-Mandeb Strait. Crude near $89 keeps the inflation impulse alive without adding to it, which is consistent with a September hike probability that has stopped falling near 31%. The shipping and OPEC+ side of this is tracked in the Hormuz and OPEC post and the Hormuz storyline.

So the refinement 17 August forces on this post is worth stating precisely, because it sharpens a phrase that has been doing loose work. "The real yield gold competes with" is not a single number. There is a short real yield, which tracks the expected policy path and is what a zero-coupon asset is genuinely discounted against, and there is a long real yield, which additionally carries term premium and fiscal risk. For most of 2026 the two moved together — the energy shock lifted real yields across every tenor, which is why the distinction never had to be drawn. On 17 August they came apart, and gold followed the short one. Rates are one of the five factors the meter scores across the eight majors, and the same curve repricing shows up on the USD page.

What actually happened on 14 August: a spending miss below the whole forecast range

The 13 August version of this post named the next scheduled test explicitly: July retail sales, 8:30 a.m. Eastern on Friday 14 August. It arrived, and it was the largest single input to gold's hurdle rate in the entire fortnight — larger than either inflation print.

The US Census Bureau put advance retail and food services sales at $763.6 billion for July, down 0.6% on the month and up 5.0% on the year, with June unrevised at a 0.2% increase. That was the first monthly decline in nine months. The scale of the surprise is what made it a rate event: economists polled by Reuters had forecast a 0.1% increase, with individual estimates spanning a 0.5% drop to a 0.7% rise. The print came in below the weakest call in the poll. Forty minutes later the University of Michigan's preliminary August sentiment index landed at 51.0, down from 55.2, with expectations at 50.6.

Gold's response was small, immediate and entirely legible. Reuters had quoted spot little changed at $4,351.45 an ounce at 11:13 GMT, before the data, with December futures at $4,407.70 and traders taking profit on the mid-week rally. After the release spot traded around $4,373.50, up 0.53%, with silver at $64.53, up 0.32%, the dollar index softer and the 10-year Treasury yield near 4.7%. On the rate market, CME's FedWatch tool had a September increase near 31% on Friday against roughly 44% a week earlier, and the probability slipped further through the session after the retail figure.

Before treating this as a verdict on the American consumer, read the composition — because the mechanism matters more than the headline, and here two of the three moving parts are calendar and price rather than demand. Reuters attributed much of the decline to payback after Amazon pulled its Prime Day event forward into June, with other retailers running competing promotions in the same window, and to a retreat in gasoline prices that mechanically cut receipts at service stations. Retail sales are reported in dollars and are not adjusted for price changes, so cheaper fuel shows up as weaker sales even when the same volume moves. Sales were still up 5.0% year on year. What the print reliably tells you is what the rate market did with it; what it tells you about household demand is a narrower claim than the headline supports. The same distinction between aggregate spending and category spending runs through the Home Depot preview for 18 August, where building materials grew 6.7% year on year in the very month the aggregate fell.

There is a second reason this test is worth separating from the five before it. The earlier legs all ran through the inflation channel — an oil price, a CPI print, a PPI core — which is the bridge this post has argued the Middle East conflict travels across to reach bullion. A retail sales miss does not use that bridge at all. It reaches the Fed through demand: weaker spending lowers the expected path of policy directly, without any claim about energy or the war. That gold responded to it in the same direction, and by roughly the size the probability shift implied, is the strongest available evidence that the channel is genuinely about rates rather than about the conflict having a gold-shaped side effect. Rates are one of the five factors the meter scores across the eight majors, and the same repricing shows up on the USD page.

What actually happened on 12–13 August: consensus CPI, a firm PPI core, and 12 points off the hurdle rate

The 11 August version of this post ended on a specific conditional: the July CPI print, not the Hormuz negotiation, is what sets the level of the hurdle gold trades against into September. Two prints later, that conditional has an answer — and it arrived in two halves that pulled in opposite directions.

The first half was clean. July CPI landed at 8:30 a.m. Eastern on Wednesday 12 August and did exactly what it was supposed to on every line markets watch: headline up 0.1% on the month and 3.4% over twelve months, down from June's 3.5%; core up 0.2% and 2.5%, down from 2.6%. That was the Dow Jones consensus on all four. Gold's response was immediate and, for once, uncomplicated — spot rose 1% to $4,409.35 an ounce, cleared its 100-day moving average at $4,387.28 and reached its highest level since 5 June, while US gold futures rose 0.6% to $4,466.80 and spot silver settled near $65.14, up 0.89%.

The rate market is where to read the cause. CME's FedWatch tool cut the probability of a September increase to about 40%, from 46% immediately before the release; the 10-year Treasury yield settled near 4.68% and the two-year slipped toward 4.20%. "The CPI data has been encouraging. It was higher than last month, but it was in line with estimates, along with a weaker dollar and technicals which have all helped gold piggyback on it," Marex analyst Edward Meir told Reuters. Note the ranking in that sentence: a print with no aggregate surprise still moved the metal, because what it removed was a probability, not a level.

The second half, on Thursday 13 August, is why this is a stall rather than a breakout. Final-demand producer prices were unchanged in July against a +0.2% consensus — a soft headline on its face — but the measure that strips out food, energy and trade services rose 0.4% on the month and is running at 5.09% annualised over three months, against 2.21% for the measure that includes distributor margins. Spot gold slipped to about $4,393.90, down 0.31%, silver to $65.07, and September hike pricing stopped falling around 40%. The full anatomy of that release — including why the annual rate fell almost nine-tenths of a point on a flat month, and why these were 14 July prices collected before the new Section 301 duties took effect — is in the PPI breakdown.

The labour-market leg did not help either, and that is worth saying plainly because it cuts against the metal. Initial jobless claims for the week ending 8 August rose 9,000 to 209,000, above a 202,000 forecast — but continuing claims fell 22,000 to 1.777 million, below the 1.8 million expected. A labour market that is loosening at the margin while people who lose work still find it quickly is not the deterioration that produced gold's 7% week on 7 August. Two of the three inputs gold needs pointed its way this week; the third did not move.

The dollar deserves a note, because it is gold's other input and it behaved differently across the two days. On 12 August the US dollar index edged lower, making dollar-priced bullion cheaper for overseas buyers and adding to the metal's gain; on 13 August it held steady after PPI, which is part of why the gold move stalled rather than extended. Risk appetite was firm through both: the S&P 500 closed 12 August up 20.30 points, or 0.3%, at 7,748.50, the Nasdaq Composite up 143.04 at 26,588.49, with the Dow slightly lower at 53,770.27. Gold reaching a two-month high in the same week that equities pressed record territory is not a contradiction — both trade off a lower expected policy path, and this week they got one.

What happened on 11 August: a two-month high, and the fade that explains it

The 9 August version of this post ended on a conditional: if the Strait of Hormuz stays shut and crude round-trips higher, the case for a September hike rebuilds. Four sessions later, that is exactly what happened, and the metal wrote the answer out in a single trading day.

The rally came first. Gold ran to its highest level in almost two months on Tuesday 11 August as traders weighed signs of movement toward a deal that would reopen the Strait of Hormuz. Comex December gold touched $4,495.00 an ounce, up 1.7% and its highest since mid-June; spot traded around $4,407.79, up about 0.4%, after Monday's close near $4,390 — a 1.11% gain and the highest daily finish in roughly ten weeks. Note what the bullish catalyst was. It was not the war getting worse. It was the prospect of the war's supply consequence being lifted, because a reopened strait means cheaper crude, a smaller energy contribution to US inflation, a lower expected policy path and therefore a lower real yield for gold to clear. A war-hedge asset does not rally on peace talk. A rates asset does.

Then the same channel ran the other way. The negotiation clouded over: oil held a four-day gain, with Brent near $88 a barrel after advancing 5% in the previous session and WTI around $82, after President Trump made fresh demands on Iran — including compensation for those killed in the conflict — while Tehran repeated its own reparations request and has said the US must lift its blockade before it will fully reopen Hormuz. Bullion faded with it, Comex December settling back to $4,440.00, a gain of 0.5%, by midday in New York, and spot slipping under $4,400 as investors took profits. (Reporting: Bloomberg; the oil leg of this storyline is tracked in the Hormuz and OPEC post.)

The dollar, gold's other input, was not the story on the day — the ICE dollar index sat near 99.8, broadly unchanged. That matters for attribution: when the currency of quotation is flat, whatever moved the metal moved it in real terms.

That 52% reading is the level the CPI print a day later removed. Consensus into it was 3.4% headline against 3.5% in June and 2.5% core against 2.6%, with the Cleveland Fed's nowcast at 3.42% and 2.52%; the release matched, and hike pricing fell to about 40%. The full mechanism of that report, including why pump prices rose 8.1% through July while the CPI gasoline average fell, is in the CPI breakdown.

What actually happened on 7 August: a negative payroll print and gold's best week since January

The July employment report landed at 8:30 a.m. Eastern on Friday 7 August and it was worse than a miss — it was a contraction. Nonfarm payrolls fell by 23,000 on the month, after a June figure revised down to a 20,000 increase, against a Reuters poll forecast of an 80,000 gain. May and June together were cut by a combined 103,000. The unemployment rate did not confirm the weakness, holding near 4.1%, because the household survey improved on a shrinking labour force rather than on hiring. (The full labour-market read, factor by factor, is in the July payrolls post.)

Gold's reaction was immediate and, on the war-hedge story, inexplicable. Spot jumped about 2.3% to $4,336.11 an ounce, having traded more than 3% higher intraday to its highest level since 17 June, and finished the week up over 7% — the largest weekly rise since 19 January. US gold futures climbed 2.3% to $4,396.9. Silver went with it, up 3.19% to $63.35 an ounce at a six-week high. There was no new conflict headline driving any of this. If anything the geopolitical tape was pointing the other way: the same day, President Trump told reporters he believed the war with Iran would be over soon.

What did move was the rate market. The benchmark 10-year Treasury yield eased to about 4.66%, down roughly 8.6 basis points on the week, and futures repriced the September FOMC meeting hard — a 43.9% chance of tightening against 57% immediately before the release, per LSEG data. "The weaker-than-expected jobs data presents a scenario where the Fed is going to be less likely to raise interest rates at its next meeting," said David Meger, director of metals trading at High Ridge Futures, adding that falling energy prices and a less likely rate increase together point to a weaker dollar.

This is now the third clean test in the same direction, and the scoreboard is consistent. Escalation: gold flat to lower through four months of war. De-escalation on 3 August: gold slightly higher. A labour-market shock with no geopolitical content at all on 7 August: gold up 7% on the week. The metal is not ignoring the war — it is pricing the war's consequence for interest rates, and reacting far more to anything that hits rates directly.

What happened on 3 August: the war premium left oil and gold went up

This storyline has spent five months offering escalation tests, and gold has failed all of them as a headline asset. On 3 August it finally got the mirror-image test — a large, sudden de-escalation — and failed that one too, in the direction that proves the point.

The trigger was documented and dated. On Sunday 2 August President Trump said he had called off a planned strike on Iran after a request from Tehran and other governments in the region, writing that the US had "just been asked by Iran, and other Middle Eastern Countries, to hold off any attack in that the perimeters of a deal has been agreed to." Crude gapped lower at the Asian open and never recovered: by Monday's settlement WTI was about 5% lower at $80.34 a barrel and Brent had lost 4.7% to $83.77, after a July in which the same benchmark had settled 7.9% higher in a single session at $90.74. Risk assets took the cue — the S&P 500 closed up 1.5%, the Nasdaq Composite 2.1% and the Dow Jones Industrial Average 1.3% at a record — while the 10-year Treasury yield fell to 4.68% from 4.75% late on the Friday.

Note what did not happen. The strait itself did not reopen. Iran's foreign ministry described its discussions as being with Oman about routing vessels rather than as negotiations with Washington, and foreign minister Abbas Araghchi said those talks were in their final stages. No incremental barrel moved on 3 August; what moved was the probability the market attached to barrels moving later. (Same-day detail on the shipping and OPEC+ side is in the Hormuz and OPEC+ post.)

And gold rose. Futures ended up about 0.56% at $4,113.40, having opened at $4,135.20, with spot near $4,051 an ounce at 10 a.m. Eastern — roughly 0.3% above where it finished July. A day that removed a meaningful slice of geopolitical risk premium from the most war-sensitive asset on the board left the supposed war hedge slightly higher.

The arc: four months down, then a shrug of a gain

The arc of gold in 2026 is easy to state and easy to misread. Bullion rose more than 65% over 2025 and peaked at $5,608.35 in January 2026. It then fell in each of the four months from March through June, with June the steepest leg — spot gold dropped 11.7% that month, testing $4,000 an ounce in fits and starts, while silver fell 22.2%, bitcoin 20.4% and spot commodities 9.2%. On a risk-adjusted basis gold actually outperformed all three, which is worth holding onto: the metal did its diversifying job during the selloff, just from a much lower starting point than January's record implied.

July interrupted the sequence, barely. Gold traded at $4,057.12 on 31 July, on course for a monthly gain of roughly 0.64% — its first in five months — after slipping back below $4,100 on the final session. Year-on-year, the metal is still up about 20.6%. The tension between those two numbers is why a price-only read of gold is close to useless right now: from the peak it looks like a rout, from a year ago a strong bull market, and neither framing identifies what is setting the price.

The hurdle rate: 2.438% is the highest a 10-year TIPS has auctioned since 2008

Gold has one structural disadvantage against every government bond ever issued: it pays nothing. No coupon, no dividend, no rent. So the fair way to think about the cost of owning it is the real yield — the return on a bond after expected inflation is stripped out — because that is the income an investor forgoes by holding metal instead.

When real yields sit near zero, that forgone income is trivial and gold competes easily. When they rise, every ounce acquires a visible annual carrying cost, and the marginal buyer starts asking harder questions. That is precisely what 2026 has done. At the US Treasury's auction of a new 10-year TIPS on 23 July 2026 — CUSIP 91282CRE3, maturing 15 January 2036 — the security priced at a real yield to maturity of 2.438%, the highest at auction for that term since October 2008, carrying a 2.375% coupon, the highest for the maturity since July 2007. Demand was unenthusiastic, with a bid-to-cover ratio of 2.30 and a result above the 2.41% when-issued level. Across the curve, the 10-year TIPS real yield has recently run around 2.1%, against a 10-year average near 0.9%. (Auction data: TreasuryDirect; series history: FRED.)

Why the war pushed gold down, not up

Here is the part that looks like a paradox and is really just two channels with opposite signs, one of them much bigger than the other.

The conventional channel is genuine: a shooting war creates a bid for safe assets, and gold gets some of it. The Middle East conflict that erupted in late February 2026 produced exactly that reflex repeatedly, in bursts. But the same conflict also shut the Strait of Hormuz for long stretches and repriced crude violently — Brent settled 7.9% higher at $90.74 on 29 July after the US launched a heavy wave of strikes on dozens of Islamic Revolutionary Guard Corps targets, then eased about 2% to $89.03 the next day when Saudi Arabia proposed a naval coalition to protect shipping. An energy shock of that size is an inflation shock, and an inflation shock in 2026 does not produce Fed easing. It produces the opposite.

That is the transmission that mattered. State Street's gold strategy team documented the repricing directly: the US overnight index swap curve moved to pricing roughly 1.5 hikes for 2026, against expectations of two to three cuts as recently as February, lifting real yields across tenors and pushing US money market fund assets to a record $7.9 trillion. Cash and inflation-protected bonds both got more attractive at the same moment, for the same reason — and both of them pay something.

So the war reached gold twice: a modest, episodic haven bid pushing up, and a large, persistent rates repricing pushing down. The net is measurable. Over the March-to-June war period, gold underperformed the US dollar, relative to the rest of G10 FX, by about 2.6 percentage points. A metal whose entire reputation rests on crisis performance lost ground to a currency during a war, because the crisis arrived through the one channel that hurts it most.

Read the middle column alone and 2026 looks like a market that has stopped making sense. Read the right-hand column and it is one variable — the real return available on cash and bonds — being revised over and over, with the war acting on gold mostly through that variable rather than around it.

Who sold, who bought: $18.7bn out of the West, 289 tonnes into central banks

If gold demand had genuinely broken, the volume data would show it. It does not — the composition changed instead, and the split is close to geographical.

The World Gold Council's second-quarter figures put total gold demand at 1,269 tonnes, flat year-on-year, taking the first half to 2,522 tonnes, up 2%, at a record value of US$380 billion. Underneath that flat headline: central banks bought a net 289 tonnes in the quarter, up 62% year-on-year, while gold-backed ETFs shed 45 tonnes, leaving first-half ETF demand only modestly positive at 18 tonnes. Jewellery volumes fell 17% year-on-year on high prices even as first-half jewellery value rose 22% to US$86 billion, and bar-and-coin investment ran 262 tonnes, down 3%. (Primary data: World Gold Council.)

The fund-flow picture sharpens it. North American gold funds took record seasonal inflows of $11.5 billion in January and February, then liquidated $18.7 billion over the following four months — including roughly $5.3 billion of redemptions from US-listed gold ETFs in June alone. Chinese funds ran $5.9 billion of inflows year-to-date, and Asia including China totalled about $12.6 billion in the first half. Physical flows corroborate the regional divide: Chinese non-monetary gold imports hit 160 tonnes in April (up 25% year-on-year) and 163 tonnes in May (up 63%), after a 120% year-on-year jump in March, with local Chinese premiums averaging 1.0% in June — their highest since April 2025, a sign of tight onshore supply against firm demand.

The official sector, meanwhile, is telling you what it intends to do. In the WGC's 2026 Central Bank Gold Reserves Survey of 76 institutions, 89% expected global central bank gold reserves to rise over the next twelve months, a record 45% expected their own institution's holdings to rise, and 1% expected a decline; 84% expected gold to hold a higher share of total reserves in five years, while 74% expected lower US dollar holdings. Set against a record global debt load of $353 trillion in the first half of 2026, with the government share approaching a third, that is a structural bid that a rate cycle can outvote in the short run without removing.

What changed in July: the dollar, not the war

The July turn is the cleanest natural experiment in the whole sequence, because the war got louder and gold barely moved, while the dollar got weaker and gold finally rose.

Two things happened to the dollar. First, the Federal Reserve met on 28–29 July and left the target range at 3-1/2 to 3-3/4 percent for a fifth consecutive meeting. The vote was 9-3, with Beth M. Hammack, Neel Kashkari and Lorie K. Logan each preferring a quarter-point increase, and the statement described activity as expanding solidly despite uncertainty owing in part to the Middle East conflict, with inflation still elevated above the 2% objective partly because of supply disruptions affecting energy and other sectors. (Primary source: FOMC statement, 29 July 2026.) Second, the dollar simply sold off into month-end — the index slid for three straight sessions to near 100.36, down about 1.03% for July, a roughly six-week low, with the greenback falling as much as 3.3% against the yen after suspected Japanese support operations, and US Treasury Secretary Scott Bessent describing the yen as "very undervalued" while arguing against excessive currency volatility.

Gold is priced in dollars, so a softer dollar mechanically lifts the number without anything changing about the metal. That is most of July's 0.64%. What capped the rest is the same hurdle rate: into month-end, markets were pricing roughly a 63% chance of a hike when the FOMC next meets on 15–16 September. An asset that pays no income cannot rally freely into a live tightening probability, however bad the news elsewhere. For how the dollar leg of this works in both directions, see the dollar-gold correlation and the live read on the USD currency page.

What would change the picture

Nothing here is a forecast — but the conditions are specific enough to watch, and they are almost all rate conditions rather than war conditions.

There is now a fifth condition, and 17 August put it on the list: whether the long end keeps repricing on its own. A thirty-year real yield at 3.06% with an unchanged breakeven, following an auction that cleared at the highest yield since 2001 with a 2.39 bid-to-cover, is a term-premium and fiscal signal rather than a policy signal — and the evidence of that single session is that gold is close to indifferent to it. What would make it matter is the front end joining in: if the same fiscal pressure ever began lifting the short real yield, or if a steepening of this kind started arriving alongside a widening breakeven rather than a flat one, the mechanism in this post would reach gold again. Watch the two-year and the five-year real yield for that, not the thirty. The minutes of the July meeting on 19 August are the next scheduled read on how the committee is thinking about exactly that split (preview here).

The first is what the September meeting does to the hurdle rate — and the market's own answer has now moved five times in a fortnight while the mechanism has not moved once. About 31% is not a cut probability; it is a tightening bet that has been steadily defunded, and it sits inside a range that has been 63% and 43.9% and 52% since the end of July. What would take it lower is more of what 12 and 14 August delivered: prints that show the aggregate inflation impulse fading, or demand cooling enough to do the Fed's work for it. What would take it higher is what 13 August hinted at — a core measure that stays near 5% annualised, which is a domestic services story the Fed cannot describe as an energy shock — or a retail sales revision that reverses much of the July drop, which is a live possibility given how much of it Reuters traced to the Prime Day calendar shift and cheaper fuel. The next scheduled tests are the minutes of the 28–29 July meeting on 19 August, which will show how a 9-3 hold was argued internally by a committee three of whose members already preferred a quarter-point increase, and then the decision itself on 15–16 September, which carries a fresh Summary of Economic Projections (FOMC calendar). The second is whether the energy leg of the inflation impulse persists, because that is the bridge from the conflict to the Fed — the war reaches gold through crude and CPI, not through the headline itself. That bridge has now been tested three times in a fortnight and given three different answers: 3 August took crude back to $80.34 on the expectation of a Hormuz reopening; by 11 August Brent was near $88 with the strait still shut; by 13 August WTI was near $82 and Brent near $87.70, lower on demand concerns, with US officials describing the strait as open while regional authorities said traffic remains restricted and Iran's conditions for a full reopening were unmet. Whether relief survives contact with actual shipping data, rather than with negotiating positions, is the question that decides whether a September hike stays priced — and a renewed crude spike would rebuild the inflation impulse and the hike probability with it, in that order. The third is the dollar, which sets the units gold is quoted in: it edged lower into the CPI print and held steady after PPI, and that difference is visible in the two days' gold prices. The fourth is flows: whether Western ETF liquidation stabilises, and whether central banks sustain a Q2 pace of 289 tonnes. Note that central bank buying has been resilient but not uniform — Russia's and Turkey's central banks sold gold in March to address domestic funding and currency pressures, which is a reminder that reserve assets get mobilised precisely when they are needed.

The takeaway

Gold's 2026 is not a mystery and it is not a failure of the asset. It is the most legible demonstration in years of what actually prices bullion: the real return available on the alternatives. A record in January when markets expected two to three cuts. A 28% slide as an energy shock turned those cuts into roughly 1.5 hikes and drove a 10-year TIPS to auction at 2.438%, the highest since October 2008. A first monthly gain in five when the dollar finally fell 1% — worth all of 0.64%. A rise of about 0.56% on 3 August, on the single most de-escalatory headline of the entire conflict, because the same headline took 5% out of crude. And then, on 7 August, a 2.3% day and a 7% week to $4,336.11 on a labour-market print — the largest move of the entire sequence, delivered by the one release with nothing to do with the war at all. And then, on 11 August, the tidiest demonstration of the lot, compressed into a single session: a two-month high at $4,495.00 in futures on the hope of a Hormuz reopening, and a fade back to $4,440.00 when firmer crude pushed the September hike probability from 43.9% to 52%. The metal rose on peace talk and fell on an oil price. Neither of those is what a haven asset is supposed to do, and both are exactly what a zero-coupon asset priced off the expected path of real rates does.

The three prints that followed closed the argument. On 12 August a CPI report that surprised nobody — consensus on all four headline lines — took the September hike probability from 46% to about 40% and lifted spot gold 1% to $4,409.35, its highest since 5 June, on a day the 10-year real yield moved a single basis point. On 13 August a PPI report with a soft headline and a 5.09%-annualised core stopped the move at $4,393.90 without reversing it. Then on 14 August a sharper instance still: retail sales fell 0.6%, below every forecast in the Reuters poll, the September hike probability went to about 31% from roughly 44% a week earlier, and gold rose to about $4,373.50 — on a day the 10-year real yield rose two basis points. When the level of the hurdle rate goes up and the metal goes up with it, the level was never the thing being traded. The expected path was, and the widening breakeven alongside it says the market repriced both legs at once. There was war news in that window too: the US and Yemen's Iran-aligned Houthis reported separate attacks on shipping on Tuesday 11 August, and prospects for ending the conflict were described as dimming. It priced almost nothing. The data priced everything.

And then 17 August took the argument one step further than any of the six sessions before it, by removing the last alternative explanation. The 30-year yield closed at its highest since 2007 and the 30-year real yield at its highest of 2026, six basis points higher on the day — the largest single-session rise in gold's nominal opportunity cost in this entire sequence — and gold rose 0.45% anyway. It could not be attributed to widening inflation compensation, because the 30-year breakeven did not move at all. It could not be attributed to a dovish repricing, because the one-month bill was unchanged and the two-year moved two basis points. What was left was a bear steepener: a market demanding more to fund thirty years of US government borrowing, on the day a 5.216% auction — the priciest since 2001 — settled. Gold is not discounted against that. It is discounted against the expected path of short real rates, and the distinction had never been visible before because the 2026 energy shock had lifted every tenor at once. One session pulled the tenors apart, and the metal told you which one it had been watching.

The war is in that story everywhere, but as a cause of the rate path rather than as a haven trade. The two channels the conflict opened were never the same size, and for four consecutive months the larger one won. Meanwhile the buyers who never priced gold off headlines in the first place — central banks adding 289 tonnes in a quarter, Chinese importers paying a 1.0% local premium — kept accumulating through the entire drawdown, because their reason for holding it was never this year's real yield.

That is the whole discipline in one asset. Identify the channel a story travels through, size it against the competing channel, and the price stops looking irrational. Gold did not ignore the war, and it did not ignore the truce talk either. It priced the consequence of both — the path of real interest rates — more accurately than it priced the headlines themselves.

To learn how Pip Theory builds its fundamental currency-strength scores, see the methodology overview.

Educational macro context only — not investment advice.