Gold (XAU/USD) Slips to $3,983.86 as $80 Crude Turns Safe-Haven Demand Into a Rate-Hike Trade — 2026 Low at $3,941 Is the Line

Spot gold trades $3,983.86, up 0.19% on the session and sitting below the $4,000 handle it lost on Thursday. COMEX August futures are $4,000.80, higher by $8.70 or 0.22%. The metal is tracking a weekly loss of more than 3%, its worst five-day stretch in six weeks. Over the past month gold has dropped 5.37%. It remains 18.88% above where it traded a year ago and 28.89% below the all-time high of $5,602.23 printed on January 29.

Now hold that against the news flow. U.S. Central Command has completed a sixth consecutive night of strikes against Iran, hitting five bridges in Hormozgan province, the Chabahar maritime control tower, the Bandar Khamir overpass, the Gariveh Bridge, and a railway terminal near Bandar Abbas. Iran has retaliated against U.S. bases in Kuwait and Jordan. September Brent is $85.01 and August WTI is $79.74, both up more than 11% this week. President Trump has warned that U.S. forces will target Iranian infrastructure next week without a diplomatic breakthrough. Tehran has told the Houthis to stand ready to close Bab el-Mandeb.

That is a live shooting war inside the chokepoint carrying 20% of the world's seaborne crude, escalating for six straight nights, and gold is down 3% on the week.

The thesis is simple and it is the only one that explains the tape: gold has stopped trading as a safe haven and started trading as a pure real-rate instrument. Every input that should be bullish is arriving through the wrong transmission channel. Higher crude raises transport and production costs, keeps inflation above target longer, forces the Fed to stay tight, lifts real yields, and raises the opportunity cost of holding an asset that pays nothing. The war is not a gold catalyst. The war is a rates catalyst, and rates are what set gold's price in 2026.

The market has roughly 73% odds priced on another Federal Reserve rate increase before December. That number is the entire forecast. Everything else — the strikes, the tanker traffic, the Bab el-Mandeb threat — is noise routed through it.

The buyer that matters is no longer in the ETF complex. It is sovereign, it is price-insensitive, and it is why $3,941 is the line rather than $3,500.

The Week: Largest Loss in Six Weeks With a War Escalating

The five-day tape is the cleanest indictment of the safe-haven thesis this year. Gold entered the week of July 13 near $4,070 and is exiting at $3,983.86. That is $86.14, or 2.12%, on spot, with the intraweek damage running over 3% at the lows. The ceasefire between Washington and Tehran is dead — both sides have confirmed the memorandum of understanding is no longer in force after five straight days of exchanged attacks, now six.

The stalemate is structural and nobody expects it to break. Washington will not cede control of the Strait of Hormuz. Tehran will not surrender its nuclear program. Both sides say they are willing to negotiate. Neither has anything to negotiate over. That is a conflict with no natural end point, and gold has spent the week selling into it.

Thursday was the session that broke the level. The dollar reversed a two-day losing streak and traded firmly higher against nearly every major, and XAU/USD pierced $4,000 to establish a fresh weekly low of $3,974. That is a $26.14 break below the handle on a day when the U.S. was bombing Iranian bridges. Friday's $3,983.86 is a $9.86 recovery, or 0.25%, off that low — stabilization, not a reversal.

The comparison that matters is what gold did the last time this conflict escalated. In early 2026, the war drove Brent above $114 and gold ripped to $5,602.23 by January 29. Same war, same chokepoint, opposite reaction. What changed between January and July is not the geopolitics. It is the Fed.

Through 2025, Western ETF buyers were the dominant marginal flow, and they drove gold from $3,865 in October 2025 to $5,595 in January 2026 — a 45% move in roughly four months, built entirely on an expected easing cycle. When the conflict flipped the inflation math in March, Chair Kevin Warsh executed a hawkish recalibration the market had not priced. The dollar surged to a 13-month high. Hike probability for September went from 29% to 68% inside a single week. Those same ETF buyers reversed, and the flow that built the rally started dismantling it.

Gold is not failing as a hedge. It is being repriced by a different buyer.

$80 Crude Is a Gold Bearish Input, Not a Bullish One

The inversion at the heart of this market deserves its own accounting, because it is counterintuitive and it is the whole trade. Higher oil is currently bearish gold.

The mechanism runs in four steps. Crude at $79.74 raises transportation and production costs across the entire goods complex. Those costs feed into headline inflation with a two-to-three-week lag at the pump and a one-to-two-month lag in the CPI series. Persistent inflation above target removes the Fed's room to cut and creates room to hike. Higher nominal rates against sticky inflation lift real yields, and real yields are the single variable that has explained gold's price for two years. An asset that yields nothing loses to a Treasury that yields 4.525% every time the real return on that Treasury climbs.

The energy complex is delivering exactly that. Brent has risen 7.64% over the past month and 24.07% year over year. WTI settled Thursday at its highest since June 15 and both benchmarks are on track for their best week since late April. Brent averaged $85 per barrel in June, $22 below May, then cratered below $70 on July 1 as the ceasefire held. It is now back at $85.01. That round trip — $114 in March, $70 on July 1, $85 today — is being imported directly into the inflation series.

The June CPI relief was manufactured entirely out of the July 1 low. The energy index fell 5.7% in June, its largest one-month decline since April 2020. Gasoline dropped 9.7%. Electricity fell 1%. Strip those out and core was flat at 0.0%. Every basis point of the disinflation that gold rallied on was an energy artifact with a two-week shelf life.

July CPI lands August 12 and carries the reversal. Twelve-month energy is still up 15.7% and gasoline up 26.7%. The market has already worked this out — traders are pricing that June's data did not reflect the recent surge in crude and that further tightening is now more likely, not less.

That is why bombing bridges in Hormozgan sells gold. The bombs raise the price of oil, the oil raises the price of money, and the price of money is what gold competes against.

The Fed Is the Whole Trade: 73% Odds of a Hike Before December

Strip everything else away and gold is one number: the market-implied probability that Kevin Warsh raises rates again this year. As of July 17 that sits at roughly 73%.

The near-term calendar is less dramatic. The July 29 decision carries a 66.3% probability that the Fed holds the target range at 3.50% to 3.75%, with other pricing putting the hold near 70%. What matters is the composition of the remaining third. It does not contain a cut. It contains a hike. There is no scenario on this calendar where the Fed rescues a non-yielding asset, and that asymmetry is why gold cannot hold $4,000 while a war escalates.

Warsh testified before Congress on July 14, the same morning the June CPI landed, and told lawmakers the Fed has no tolerance for persistently high inflation. On a 2.6% core print he did not have to demonstrate it. On an August print carrying $80 crude through the energy component, he will. The hike probability that collapsed from 40% to 15% for the near-term meeting on the June data can travel that distance in reverse just as fast, and the 73%-before-December number is the market saying it already expects exactly that.

This is the regime change nobody wants to underwrite. Through 2025, ETF investors bought gold as a leveraged expression of an easing thesis, and they were rewarded through January 2026. The thesis broke in March. It has not been rebuilt. Every forecast now working off the current rate path lands in the same place: one analyst set has gold trading $3,365 to $4,236 across July with an end-of-month range of $3,542 to $3,887, and a bearish year-end target of $2,875 to $2,994. A major Asian bank expects decline through the end of 2026 on rising Treasury yields, a stronger dollar, and weaker precious-metal demand.

The University of Michigan inflation expectations reading prints at 10:00 a.m. ET today and is the session's only scheduled catalyst. June's final long-run expectation fell to 3.3% from 3.9% in May. The one-year measure eased to 4.6% from 4.8% and remains far above the 3.4% recorded in February before the conflict began. A move higher in either measure on $80 gasoline is what tips the 73% toward 85%.

The Dollar at 100.75 Is Doing the Damage

The Dollar Index sits at 100.75, having pushed to 100.79 on the overnight session, and it is the proximate cause of every tick gold has lost this week. The relationship is arithmetic — gold is priced in dollars, so a stronger dollar mechanically compresses the price for every buyer outside the United States before any fundamental argument gets made.

The dollar's strength is not accidental. It is the direct output of the same chain that is hurting gold from the rates side. Six nights of strikes drove crude up 11% on the week. Crude drove inflation expectations up. Inflation expectations drove hike odds to 73%. Hike odds drove the dollar. Every geopolitical escalation this week has been a dollar catalyst first and a gold catalyst never, which is the opposite of how this trade worked for two decades.

Friday added a second dollar leg that had nothing to do with Iran. President Trump released declassified intelligence alleging China interfered in the 2020 U.S. election and obtained records on 220 million American voters. The Chinese Embassy denied it in full. The Australian dollar — the cleanest liquid proxy for Chinese growth exposure — weakened immediately, and DXY pushed to 100.79 on the headline. A U.S.-China rupture is a dollar bid, and a dollar bid is a gold offer.

The rates complex confirms the setup rather than contradicting it. The 10-year Treasury yield is more than 4 basis points lower at 4.525%. The 2-year shed 3 basis points to 4.124%. The 30-year is more than 3 basis points lower at 5.061%. Every point on the curve is bid, which is a classic risk-off configuration — and gold is not participating in it.

That is the tell. When capital moves into duration and the dollar simultaneously, and gold does not catch a bid, the market is telling you it has reclassified the metal. Treasuries are the haven. Gold is the inflation-hedge trade that stopped working when the Fed proved it would defend the target. The DXY reached a 13-month high earlier this year on precisely this dynamic, and it is rebuilding toward it.

Gold at $3,983.86 with DXY at 100.75 and the 30-year at 5.061% is not a contradiction. It is the correct price.

The Chart: $3,941 Is the Line

The technical structure is clean, which is unusual and useful. Spot at $3,983.86 sits above exactly one meaningful level and below everything else.

The 2026 bottom is $3,941. That is $42.86, or 1.08%, below spot. Thursday's low of $3,974 is the first intraday reference and it is $9.86 under current price. Below $3,941 the next stop is $3,890 to $3,900, a zone with no structural memory behind it, and one forecast set has $3,940 and $3,890 as the explicit downside targets from here. The weekly framework puts nearest support at $3,945, within four dollars of the yearly low, which means the two most-watched support definitions are effectively the same line.

Overhead, $4,002 is the immediate resistance — the reclaim level, $18.14 or 0.46% above spot. The consolidation band runs $4,007.83 to $4,114.01 and price is currently below it. On the intraday charts the 20-period SMA sits near $4,031, the 100-period at $4,070, and the 200-period at $4,174 as the stronger barrier if a deeper corrective bounce develops. The broader zone that gold needs to reclaim to signal any shift back toward bullish momentum runs $4,090 to $4,125.

Add it up and the risk-reward at $3,983.86 is unattractive in both directions. Downside to the yearly low is 1.08%. Upside to the first real resistance cluster is 2.16%. Gold is trapped in a 3.24% box between a floor it has already tested and a ceiling it has failed at four times.

The momentum reads confirm the trap. MACD remains in negative territory, though the decline is slowing. The Stochastic Oscillator has turned down after exiting overbought. Prices hover in the lower half of the Bollinger band. The one-week technical rating is a sell. The one-month rating is a buy. That split — bearish short-term, constructive medium-term — is the numerical expression of the entire divergence in this market between Western flow and sovereign flow.

The constructive read is a falling wedge that has been building since the correction, and wedges resolve higher more often than not. Reclaiming $4,002 and holding it is the confirmation. Losing $3,941 kills it.

Every Moving Average Is Overhead

The moving-average structure is the most damning thing on gold's daily chart, and it is why bounces keep dying.

The 20-day SMA sits at $4,081.06. That is $97.20, or 2.44%, above spot, and it has acted as the first cap on every rebound attempt for weeks. The 200-day SMA sits near $4,495.72, a full $511.86 or 12.85% overhead. The 100-day SMA is even further out at $4,547.88, $564.02 or 14.16% above. On the weekly framework, price remains below the 50-period moving average with key resistance defined at $4,388 — $404.14, or 10.14%, above where gold trades right now.

Read that stack. Gold is below every meaningful moving average across every meaningful timeframe, and the nearest one is 2.44% away. That is not a market in a pullback. That is a market in a confirmed downtrend where the first technical objective — reclaiming a 20-day line — requires a 2.44% rally that the metal has not managed in six weeks.

The 100-day and 200-day sitting at $4,547.88 and $4,495.72 tells you where gold spent the bulk of the last two quarters. Spot at $3,983.86 is a $511 to $564 dislocation from that base. Either the last two quarters mispriced the metal or the current tape has overshot, and the moving averages will not resolve that question. Real yields will.

While price trades below $4,388 on the weekly, sellers have the upper hand and the base case is consolidation with a retest of $3,945 in the absence of a new upside driver. That framework was published before this week's data and it has held perfectly — gold consolidated, then broke.

The one thing that would change the structure fastest is the 20-day at $4,081.06 flattening rather than falling. It has not. Momentum indicators are extending their slopes into negative territory, which means the moving averages are still rolling over rather than converging with price. A market grinding sideways under a falling 20-day is distributing, not basing.

Gold needs a daily close above $4,002 to start the conversation, above $4,031 to get the intraday structure back, and above $4,081.06 to make anyone care.

The June CPI Print Bought Gold Two Sessions

Gold got the inflation data it needed and it lasted less than 48 hours. That is the second-most important fact in this market.

Headline CPI fell 0.4% in June against a 0.2% expected decline, the largest single-month drop since April 2020, pulling the annual rate to 3.5% from 4.2% in May against a 3.8% consensus. Core was flat on the month versus a 0.2% expected rise, taking the twelve-month core rate to 2.6% from 2.9% — three-tenths below forecast and the lowest since February. The producer price report missed as well. The dollar fell 0.6% on the day. Near-term Fed hike probability collapsed to 15% from roughly 40%.

Gold ripped on it. There was a short squeeze around the key liquidity zone directly following the Tuesday release. Then the market regained its bearish momentum and resumed selling in line with the prevailing trend. Two sessions of relief, then straight back down through $4,000 and into a fresh weekly low at $3,974.

The bulls lost steam because the market read the print correctly. June's data did not reflect the surge in crude that started July 8 and accelerated through this week. Traders are now positioned for the possibility that U.S. rates stay elevated and that a non-yielding asset is the wrong place to be while that is true. The 73%-before-December hike probability was set on July 17, three days after the softest CPI in five months. That is the market saying it does not believe the print.

This is the definitive test of a thesis. If an asset cannot rally on the exact catalyst its bull case requires, the bull case is not what is setting the price. Gold got a downside CPI surprise, a downside PPI surprise, a 0.6% dollar decline, and a 25-percentage-point collapse in near-term hike odds — inside one session — and it is now $86 lower than where it started the week.

The next scheduled input is the University of Michigan inflation expectations reading at 10:00 a.m. ET. It will capture six nights of strikes, an 11% weekly crude move, and pump prices that have not yet fully reset. Gold does not want that number.

Western ETFs Are the Marginal Seller

The flow picture explains everything the chart cannot. Gold-backed exchange-traded funds recorded net outflows of 16 metric tonnes in May 2026 and kept bleeding into June. North American funds lost $5.5 billion in June alone, taking regional first-half outflows to $7.7 billion — the weakest opening six months since 2013. Roughly 298 tonnes of ETF gold is currently underwater.

The mechanism is the same one that built the rally in reverse. Western ETF flow was the dominant marginal buyer through 2025, and it drove gold from $3,865 in October 2025 to $5,595 in January 2026 — 45% in four months — as a leveraged bet on Fed easing. When Warsh's hawkish recalibration landed in March, that same capital reversed. Record inflows became record outflows. Rising real yields and a strengthening dollar pushed up the opportunity cost of holding gold, and the funds sold.

Asia is a partial offset that has stopped offsetting. Global first-half ETF inflows were driven entirely by Asian funds, but bullion fund investors across the region including China sold roughly $3.6 billion in late May and June. The counter-signal is that Chinese local gold price premiums averaged 1.0% in June, their highest since April 2025 — a level that historically precedes a rebound. The last time premiums ran this hot, Asian gold fund inflows surged $14.7 billion across the back half of 2025.

The physical trade in China supports that read. Non-monetary gold imports into the mainland surged in the second quarter — 160 tonnes in April, up 25% year over year, and 163 tonnes in May, up 63% — on the back of a 120% year-over-year increase in March during the early stage of the conflict. That is physical accumulation into a falling price, and it is the argument for bullion trading $4,500 to $5,000 later in the second half if Western fund flows stabilize.

That conditional is doing enormous work. Western flows have not stabilized. They are the marginal seller, they are rate-driven, and they will not turn until the 73% hike probability turns. Chinese premiums and import tonnage are a floor, not a bid.

GLD's $14.4 Billion Exodus Is Partly a Fee Story

The headline number on gold ETF redemptions is worse than the underlying reality, and separating the two matters for anyone reading flow as a sentiment gauge.

SPDR Gold Shares has bled $14.4 billion since March 1. The run started with a record $2.91 billion single-day outflow on March 4 — the largest withdrawal from the fund in over a decade — followed by a $4.2 billion weekly outflow in the week ending March 5, also a record. GLD trades $370.07 with $177 billion in assets under management as of July 15.

Part of that is a genuine exit from gold exposure. Part of it is a fee migration. GLD carries a 0.40% expense ratio, and at spot prices in the $4,000 zone that drag has become material enough that holders are rotating into iShares Gold Trust and SPDR Gold MiniShares, both of which charge less for the same underlying exposure. The bullion does not leave the market when that happens. It changes custodian.

That distinction cuts the bearish signal from the flow data meaningfully. A $14.4 billion outflow from the largest, oldest, most expensive product in the category over four and a half months, during a period when the metal fell 28.89% from its high and the Fed pivoted hawkish, with a demonstrable cheaper alternative sitting on the same shelf, is not a referendum on gold. It is a referendum on 40 basis points.

The pieces that are a referendum on gold are the regional totals. North American funds shedding $5.5 billion in June and $7.7 billion across the first half is aggregate, product-agnostic, and cannot be explained by fee migration. That is capital leaving the asset class. The weakest first half since 2013 is the correct frame.

The structural competition is also real. Institutions are choosing between gold at zero yield, Treasuries at 4.525% on the 10-year and 5.061% on the 30-year, and a digital alternative that has spent 2026 marketing itself into the same allocation slot. At current real yields, gold loses that comparison on arithmetic before anyone opens a chart.

The rotation resolves when the rate path resolves. Not before.

Central Banks Bought 244 Tonnes and Never Stopped

The other side of this market is buying with a conviction the ETF complex has not shown in a decade, and it does not check the price.

Central banks bought a net 244 tonnes of gold in the first quarter of 2026, exceeding both the prior quarter and the five-year average. Poland added 14 tonnes in April alone, taking its year-to-date total to 45 tonnes. The People's Bank of China extended a buying streak that has now run 18 consecutive months. The Czech National Bank added 2 tonnes. All of that purchasing continued while gold sat 28% below its January peak.

Over the past four years, central banks have accumulated an average of 1,000 tonnes annually, up from the 500-tonne average across the preceding decade. That is a doubling of the structural sovereign bid, sustained across a period covering both the rally and the correction.

The forward survey data is more emphatic than the historical data. In the latest official-sector poll, 89% of respondents expected global central bank gold reserves to keep rising. A record 45% said they expected their own reserves to increase over the next 12 months. And 74% expected lower U.S. dollar holdings within global reserves over the next five years. Those two lines are not independent variables.

The behavioral asymmetry is what makes this bid different from every other buyer in the market. A central bank targeting a specific tonnage allocation has a mathematical incentive to buy more when prices fall — each ounce costs less and moves the institution closer to its strategic target. ETF holders operate on quarterly rebalancing windows and mark-to-market performance. Sovereign reserve managers operate on decade-long mandates. They are playing different games measured in different units of time, and only one of them is forced to sell into a drawdown.

That is why $3,941 is the floor rather than $3,500. When Western ETF flow reversed in March, sovereign buyers absorbed part of the selling. They are still absorbing it. The 28.89% drawdown from the January high has not produced a single tonne of official-sector liquidation on the tape.

Gold's price is being set by the seller. Gold's floor is being set by the buyer.

Gold Passed Treasuries as the World's Largest Reserve Asset

The single most consequential fact about gold in 2026 is buried in a European central bank report and almost nobody trading the metal today is pricing it.

Gold has overtaken U.S. Treasuries as the single largest asset class held across global central bank reserves. Gold now accounts for 27% of global central bank holdings against 22% for Treasuries. That crossover was confirmed in the June 2026 assessment of the euro's international role, and it is the first time in modern financial history it has happened.

That is a structural transition that took decades to build and will not unwind on a short-term timeframe. It is also completely orthogonal to whether the Fed hikes in September. The 73% hike probability that is currently pinning gold at $3,983.86 is a 2026 variable. The reserve-composition shift is a generational one, and the 74% of surveyed central banks expecting lower dollar holdings over the next five years is the forward extension of it.

The tension between those two horizons is the entire gold market right now. On a three-month view, gold is a real-rate instrument getting repriced by a hawkish Fed and $80 crude, and every technical read confirms the downtrend — below the 20-day at $4,081.06, below the 200-day at $4,495.72, below the weekly 50-period, with a one-week sell rating. On a five-year view, gold is the reserve asset that just displaced the Treasury, bought by an official sector that doubled its annual accumulation rate and told surveyors it plans to keep going.

Both of those can be true simultaneously. They are being expressed by different buyers on different clocks, and right now the short clock owns the price.

For traders, that means the drawdown has a floor that most drawdowns do not. The 298 tonnes of ETF gold currently underwater will get sold if rates keep climbing. The sovereign tonnage will not. When the marginal seller exhausts, there is no equivalent overhang on the other side, because the entity holding 27% of global reserves in this metal is not a seller at any price the Fed can produce.

That is why the one-month technical rating is a buy while the one-week rating is a sell. The chart is describing two different markets.

Miners Are Down 15% and Carrying the Leverage

The equity complex has taken the beating that the metal's 5.37% monthly decline understates, and that is the operating leverage working in reverse.

The VanEck Gold Miners ETF has closed at its lowest level since last November, with many constituents trading to new year-to-date lows. Inside the index, Newmont shed 14.9% on a 10.6% weight, dragging the fund 1.58%. Barrick fell 13.7% on an 8.0% weight for a 1.09% contribution. Kinross dropped 21.7% on a 4.4% weight, costing 0.97%. Wheaton Precious Metals fell 15.3% on a 5.7% weight. AngloGold Ashanti dropped 16.5% on a 4.9% weight. GDX carries $22.83 billion across 69 holdings.

The math is the point. Gold fell 5.37% over the month. The senior producers fell three times that. Mining costs are relatively fixed, so when the metal drops, margins compress non-linearly and equity values follow. The same leverage that let GDX post a 43% one-year return while bullion did a fraction of that is now amplifying the downside.

The technical damage confirms the regime shift. GDX's 50-day moving average crossed below its 200-day on June 26 — a long-term bearish signal — and its momentum indicator dropped below zero on July 10.

The valuation case is where it gets interesting. Newmont closed the first quarter with $8.8 billion of cash and $3.2 billion of net cash, which materially reduces financial risk if gold weakens further and gives management genuine flexibility to repurchase shares into the drawdown. It trades at 6 times trailing adjusted EBITDA and roughly 9.5 times 2026 consensus earnings. That slots it between Agnico Eagle at 7.2 times trailing EBITDA and Barrick at 5.5 to 5.6 times. Average all-in sustaining costs across the sector have run near $1,424 per ounce, which at $3,983.86 spot leaves a $2,559.86 gross margin per ounce.

Miners producing at $1,424 and selling at $3,983.86 are printing money, trading at 6 times EBITDA, and sitting at eight-month lows. The earnings season landing in the final week of July is where operational execution either offsets the commodity move or does not, and management commentary on all-in sustaining costs in the wake of the energy swing is the number that decides third-quarter margins.

The Trade: $3,941 Floor, $4,081 Ceiling, $4,388 Ceiling Above That

The levels are tight and the asymmetry is poor. Spot gold is $3,983.86. The 2026 low is $3,941, 1.08% below. The immediate reclaim level is $4,002, 0.46% above. The 20-day SMA at $4,081.06 is 2.44% overhead and has capped every bounce for six weeks. Weekly key resistance sits at $4,388, 10.14% above. Downside targets below the yearly low run $3,940 and $3,890.

The base case is consolidation inside a 3.24% box between $3,941 and $4,114.01 while the Fed question stays open. A daily close above $4,002 starts the repair. Above $4,031 restores the intraday structure. Above $4,081.06 makes it a trade. Below $3,941 opens $3,890 with nothing structural underneath.

The bear case needs one thing and it is already happening. Crude at $79.74 and Brent at $85.01, both up 11% on the week, feed the August 12 CPI print. If July inflation carries the energy reversal, the 73% probability of a hike before December goes to 85%, the dollar clears its 13-month high, and the 30-year pushes past 5.061%. Gold does not hold $3,941 in that scenario. One published year-end forecast puts the metal at $2,875 to $2,994 on exactly that path.

The bull case needs the Fed to blink, and the Fed has 66.3% odds of doing nothing on July 29 and no scheduled reason to cut. Failing that, it needs Western ETF flow to stabilize — North American funds have shed $7.7 billion in the first half, the worst since 2013 — at which point Chinese physical demand running 163 tonnes of monthly imports and 1.0% local premiums does the rest. That path is the one that produces $4,500 to $5,000 later in the second half, and it is entirely conditional on a flow reversal that has not started.

Watch $3,941 and watch WTI. The official sector bought 244 tonnes in the first quarter and has not stopped, which is why the floor exists at all. But sovereign accumulation on a decade mandate does not move a price that is set daily by real yields, and real yields are going the wrong way.

That's TradingNEWS