We asked Claude's new frontier model to build an institutional research stack on a retail subscription. Live data, comps, sensitivity grids, a probability-weighted thesis tracker. One hour, start to finish. Then it built the most sophisticated sheet in the workbook unprompted. A full risk module, correlation matrix, portfolio volatility. That sheet shipped broken. Its own verification corrupted the formulas, and its defensive error guards converted the failure into calm zeros. Every automated check passed. A full recalculation reported zero error cells. Zero error cells is not the same claim as correct. The only thing that caught it was a human who knew a risk sheet reporting zero volatility is lying. Every prompt is in the letter. The finished workbook is attached, free. So is the broken one. Run the liar through your own verification stack and see if anything you trust catches it. If nothing does, that is the most valuable thing you will learn this week, and it costs nothing. Full free article:

Everyone has a reason gold is finished this week. The war is ending. Warsh is a hawk. The chart looks broken. So let me tell you the one thing that would actually make me wrong, because it is none of those. The trade has a single kill switch, and I have written it down in public. The official sector, the central banks doing most of the buying, steps back for two straight quarters. That is it. That is the thing that ends this. Not a ceasefire, not a press conference, not two ugly weeks on the tape. Now ask whether a Hormuz deal produces it. Central banks are not buying gold because of a war in the Gulf. They are buying it because they watched three hundred billion dollars of another country's reserves get frozen, and they want an asset that is nobody's to freeze. A ceasefire in the Strait does not give them their counterparty risk back. Warsh sounding tough on Tuesday does not unfreeze Russia. So when the bears hand me peace and a hawkish chair as the reason to sell, I check them against the only falsifier that matters, and neither one is on the list. The official sector is still buying. Until that number turns, for two quarters, the rest is noise with a chart attached. I will tell you the day my kill switch trips. This is not that day.

Forget the terms for a second and watch the behaviour. Trump announced a signing for today. Within hours, Iran went on the record to say no, not today. Think about what that takes. A party desperate for a deal does not publicly correct the President of the United States on the way to the altar. It signs, it smiles, it takes the win. Iran is doing the opposite. It is content to let Trump look like he jumped the gun, content to drag the clock, content to make Washington wait. That is the tell, and it is worth more than the leaked text. A counterparty that behaves like it has time is a counterparty that thinks it holds the leverage, and a counterparty that holds the leverage does not sign on your terms. So whatever eventually gets inked costs the US more than its officials are admitting, because the side supposedly being squeezed is the one setting the pace. People keep asking who blinks first. Watch who is relaxed about the calendar. It sure as hell is not Washington. The regime that is supposedly cornered is the one making the superpower wait. That should bother you more than any clause in the document.

Gold is getting hit again, and the reason on every desk is the same. Peace is here, the war premium comes out, sell the metal. So let me point at the wire those same desks are quoting. Trump says the deal gets signed Sunday. Iran says it will not be signed Sunday. The resolution you are selling gold for has not happened, and one of the two parties just said it is not happening this weekend. So price in the relief if you want, but be honest about what you are doing. You are handing back your hedge on a ceasefire that has been approved, then imminent, then Sunday, then denied, all in four days. You sold protection against an event that keeps not occurring, against a counterparty that keeps not signing. I am not selling gold into an unconfirmed peace. If the deal is real, the metal does not need it. If the deal is fake, you just dumped your insurance on a headline that has been wrong all week. Sell the hedge when the thing it hedges is actually gone. Not when one side tweets it might be, on a Saturday, about a Sunday the other side says isn't bloody happening.

Let me give you the sequence, because it is funnier than any take I could write. First the deal was approved by all parties. Then it was twenty four hours from signing. Then it was getting signed Sunday. Now Iran says it will not be signed Sunday. So the deal has been imminent for four straight days, and one of the two parties to it just publicly corrected the other on what day it happens. Here is the only thing a moving date tells you. It is not a date, it is a negotiating position. A deal that is genuinely done gets a time and a place, not a fresh countdown every morning from whichever side wants the headline. Trump keeps announcing a signing. Iran keeps declining to confirm it. That is not paperwork being finalised, that is one man selling a close the counterparty has not agreed to. I am not putting a single dollar around Sunday, because Sunday was already approved, then twenty four hours, then Sunday, and Iran says it is not even fucking Sunday. When the date keeps moving, the deal is not late. It is not done.

Two things crossed the wire this week and almost nobody put them next to each other. American officials swearing not one dollar moves to Iran until it performs, the Vice President posting that no funds are released for simply signing. And, separately, the UAE preparing to unlock billions of dollars for Iran. Put them together and you have the whole trick. The gate Washington is selling its public, no money without performance, is real, and it is also completely porous, because the UAE is not bound by Washington's gate. Iran can be paid on signature while the United States releases nothing, because the cash carries an Emirati signature, not an American one. Both sentences are true at the same time. That is not an accident, it is the design. So when a US official stands at a podium next week and says not one American dollar was released, believe him. It will be true and it will be fucking meaningless, because that was never where the money was coming from. The deal does not need American money to fund Iran. It needs American permission and a neighbour with a chequebook. Watch the Gulf accounts, not the Treasury statement.

To everyone who sold gold at the bottom: You sold on one rule. Real rates up, gold down. It held for a decade. Then it broke. Real yields near a fifteen year high. The old model says gold near $1,000. It trades around $4,200, and the gap has not started to close. A sovereign that cannot fund itself at positive real rates has two exits. Default, or print to cap yields. Both are bullish for the one asset that is nobody's liability. Rising real rates here are the buy signal, not the sell. The thing scaring you out is the thing setting up the next leg. The thesis is free. June 17 is the tell:

Every time I lay out the trapped Fed, the same objection comes back. The Fed took rates over five percent in 2022 and 2023, the system creaked, nothing broke, so it can clearly do it again. It is the best objection there is, so here it is head on. The Fed got away with it once for one reason. The debt was still cheap. The federal stock had been financed near zero, the corporate maturity walls were years out, and a higher funds rate did not yet feed the interest bill or the refinancing calendar in real time. The hikes bit slowly, and the system had a cushion. That cushion is gone, and it is gone because those exact hikes spent it. The debt rolled into them. The federal interest bill roughly doubled toward a trillion dollars, close to ten trillion comes due again inside a year, and the leveraged credit struck at zero in 2020 was termed out for years, so in 2022 it was not due and now it is. So 2022 is not proof the Fed can hike into leverage. It is the story of how the trap got built. You do not get to point at the cushion as evidence while standing in the hole it left. The market has priced roughly two thirds of a hike by year end. I think that premium is wrong, and the reason it is wrong is the same reason the bears think it is safe.

Gold fell 10% and everyone blamed the war. @GrantsPub told @MebFaber the frame is backwards, three days before Warsh's first FOMC. He's right, and the selloff is the proof. The Fed is trapped. It can't deliver the hikes the market is pricing without breaking what's underneath. Buying the most hated trade into June 17. $GLD Read Below.

Iran has announced a state funeral for Ali Khamenei. Funeral July 4, burial July 9. The office that sits above the president and holds the final word on every strategic decision Iran makes is empty, and a peace deal is supposed to be signed inside the next twenty four hours. Stop and ask the only question that matters for pricing this. Who is bound by that signature. A deal is worth exactly what the counterparty can deliver, and Iran's ability to deliver anything now runs through a succession that has not happened yet. The contest for that office has been the central unanswered question in Iranian politics for a decade. Whoever inherits it inherits the deal, and a successor consolidating power has every incentive to reopen terms his dead predecessor agreed to under fire. I do not know who wins that fight, and neither does anyone selling you certainty on this ceasefire this weekend. That is the point. The signing is being rushed precisely because the structure underneath it just became unstable. A ceasefire signed over an open succession is not an ending. It is the first item on the new man's desk.

The crowd is selling gold this weekend because the war is ending. Fine. They were always trading the wrong thing. Here is the trade, and it has nothing to do with Hormuz. A government that cannot fund itself at positive real rates has two exits, default or debasement, and both are bullish for the one asset that is nobody's liability. That is one wall of the box, the sovereign one, the interest bill. I have written that wall for two months. Jim Grant handed me the second wall this week, and it is the one I had underweighted. The cheap money of 2020 and 2021 loaded private equity with six and eight turns of debt at rates that are never coming back. That debt rolls now. A Fed that hikes hard into it is not fighting inflation, it is detonating the leverage its own policy created. So the sovereign cannot afford the interest and the private credit complex cannot survive the refinancing, and the Fed is boxed by both walls at once. Now the part that should stop you. About two trillion dollars of that private credit sits inside life insurers, floating rate, levered, understood about as well as a CDO squared was understood in 2005. That same two trillion is the drain that pulls on gold every time the system delevers, because when the margin call lands you sell the liquid thing, not the thing that broke. The leverage that traps the Fed and the selling that has hit the metal are not two stories. They are one machine. Warsh can still hand us a flush on Tuesday, and if he does it is the entry, not the exit. But the machine does not give a damn what he says at the podium.

$GLD On Monday I told you my base case was one more flush into the 17th, fifteen to twenty percent through the complex, and that it would be the entry. If this deal is real, I may not get it. Good. The letter had two paths in, written down in advance precisely so I would not need the market's permission this week. The flush was one entry. The resolution is the other. The seller's reason disappears, and in these episodes sentiment bottoms on relief, not on a crescendo. So the reserve sleeve goes to work on the turn instead of the capitulation. Same ladder, same order. What I lose is a better price. What I get is the driver gone. And if the deal wobbles, which with two sides describing two different contracts it absolutely might, the flush path reopens and the other half of the plan is still sitting there. That is the whole point of writing both down. Wrong about the path, right about the destination is a trade I will take every week of my life. The only plan that fails here is the one still waiting for a bottom that just got negotiated away.

The photo on the wire today says more than the statements. Tankers anchored in the Strait of Hormuz, shot from the Omani side, waiting. Here is the thing about this deal. The ink is instant. The water is not. Mines take thirty days to clear by the deal's own terms. The reopening is gradual across sixty. War premiums come off hulls when hulls stop being at risk, not when a ceremony airs. So the reopening you can actually trade is not the signature. It is that anchorage emptying, escorted convoys turning back into boring traffic, insurance quotes deflating. Everyone will watch the ceremony. Almost nobody will watch the queue. I will. I am going out there, to talk to the people who move these ships and the people who insure them, and to look at the water myself. The ceremony is on television. The reopening is visible from a headland.

The consensus take on gold miners this week runs, peace means gold down, and anyway oil at 150 was about to eat their margins. Both halves wrong, and the second half was always wrong. Run the diesel math. Roughly six dollars an ounce for every ten percent move in oil, after hedges. Against cash margins north of two and a half thousand dollars an ounce, the cost line was never the risk. It was a rounding error wearing a scary mask. The real risk was gold beta, the forced selling that dragged the whole complex down with the metal. And the deal is what retires the forced seller. So the miners come out of this week with the fake risk collapsing and the real risk fading at the same time, still priced for a gold number a third below where gold actually trades. The market spent a month hedging the rounding error and selling the franchise.

Read the structure of this deal, not the headline on it. The ceasefire gets extended sixty days. The strait reopens gradually. Iran clears its mines inside the first thirty days. The money only moves when Iran performs. The nuclear talks have not even started, the deal just builds the room they will happen in. Nothing in that list ends the standoff. Every line schedules it. A thirty day mine deadline is a fresh chance to fail in thirty days. Performance gates on 24 billion dollars are an argument with a due date, every tranche. A sixty day extension is a countdown clock with a signing ceremony in front of it. The market will trade the word peace. The document is a sequence of conditional steps where the war is the penalty clause, and you get to re trade the whole thing every month. The mine clause is the first checkable fact this war has produced. I intend to check it.

If you bought gold in February because missiles were flying, this is your exit. Take it. I will happily be on the other side of your trade. Because the case was written before the first strike and it reads the same after the last one. A government that cannot fund itself at positive real rates has two exits, default or debasement, and both end in the same place for the metal. No ceasefire touches that arithmetic. No signing ceremony shrinks the debt, slows the rollovers, or talks the official sector out of a buying streak that started years before this war and ran straight through it. The war was a chapter. It was never the book. So let the tourists hand back their war premium this week. The sorting is healthy. And the calendar has not moved either. Wednesday is still Wednesday.

Buried in a Houston energy event, the most important number of the war. The US Energy Secretary says the military is helping move seven million barrels a day out of the Persian Gulf. Right now. Through the strait everyone calls closed. A waterway that moves seven million barrels a day is not fucking closed. It is taxed. The closure was never a wall, it was a toll, paid in naval escorts, war insurance, and delay. Normal flow was close to a fifth of the world's seaborne oil, so the toll is real and most of the flow is still missing. But the strait has been a convoy lane for months, not a wall. Which reframes what this deal actually does. It does not open the strait. It removes the tax. And the difference between a convoy lane and ordinary commerce is the spread the whole reopening trade lives in. Convoys, helpfully, are countable. I am going to Oman to count them myself. An anchorage headcount does not have a communications staff.

There is a number that explains why the company that built America's best nuclear reactor went bankrupt building it, why the UK's flagship plant nearly tripled in cost without a single technical failure, and why China builds the exact same designs on time while the West cannot. It is not the cost of concrete. It is not regulation. It is not the reactor. Western reactors work. Every one of these projects failed for the same reason, and almost nobody talks about it. I wrote a full breakdown of how nuclear power actually works, from the physics up: why a well designed reactor's default state is off, why Chernobyl was a design pathology that cannot happen in a Western plant, and the one mechanism that quietly decides whether the nuclear buildout happens at all. If you are going to have an opinion on nuclear, this is the foundation.

Oil is at two month lows on the deal news, and the first instinct everywhere is that peace is the bearish gold event. War premium out, sell the metal. Walk through who was actually selling first. For three months the marginal seller of gold was the energy importer, bill inflated by the closure, dumping the most liquid asset in the reserve account to raise dollars. The war was never gold's bid. The war was gold's seller. So as the strait actually reopens and the bills shrink, the scramble ends and the seller retires. And the buyer underneath, the official sector that bought right through the war, never needed the war in the first place and never left. Peace does not fire the buyer. It fires the seller. Read that twice before you sell what might be the bottom.