Week In Review: The Headliner Bombed. The Opening Act Went Up 177%.

Last week the light turned green and nobody moved. This week somebody finally floored it, just not the car everybody was watching.

The Dow being the best index of the four is the market's way of telling you that a basket of insurers, drugmakers and industrial conglomerates just beat the future. Somewhere a very expensive quant is explaining that this was in the model all along.

Had $1,000 riding the Nasdaq on Monday morning? You have $975.90 now. Had $1,000 in gold miners instead? You have $1,142.90. Nobody puts that trade in a retirement brochure.

On Wednesday, Moderna and Merck said their personalized mRNA cancer vaccine succeeded in a late-stage trial. Moderna's stock opened at $116.02 against Tuesday's close of $62.96, finished the day at $174.38, and posted a 176.97% single-session gain on 199 million shares, roughly 46 times the volume it traded the day before. Merck, a company many times Moderna's size, rose 12.60% the same afternoon. Merck does not move 12% in a day. Merck barely moves 12% in a quarter.

Here is what they found, in plain terms. The shot is called intismeran autogene, and it is built to order: doctors sequence the mutations inside one patient's tumor and make a vaccine that teaches that patient's immune system to hunt those exact cells. It was given alongside Merck's Keytruda to more than 1,000 people whose melanoma, a serious skin cancer, had already been cut out by surgeons. It met its main goal, recurrence-free survival, meaning the cancer came back less often, and a secondary goal, distant metastasis-free survival, meaning that when it did come back, it spread less. No new safety problems turned up. Researchers have chased this idea for decades and built a long, expensive graveyard of near-misses, so "it worked" is doing a lot of load-bearing work in that sentence.

For the week, Moderna finished up 129.20%, having dropped 23.55% on Thursday and gained 8.86% on Friday, because a stock that nearly triples on Wednesday attracts a specific kind of visitor Thursday morning. A $1,000 lottery ticket looks like this: $1,000 in Moderna on Monday was worth $2,292 by Friday, if you had the stomach to sit through Thursday, which almost nobody does.

That one trial is most of why health care was the best sector on the board. It is not, please note, because scared money rotated into safe boring pharma. UnitedHealth, the biggest health insurer in America, fell 2.89% on the week in the middle of this supposed health care rally. This was not a sector. This was a science result with a ticker attached.

Investors spent two years pricing machines that write emails. The best week of the year came from a machine that reads tumors.

Nvidia closed lower every single session this week. Monday, Tuesday, Wednesday, Thursday, Friday. It ended down 4.64%, and at no point did anything you would call a panic occur. That is the interesting part. It was not sold off a cliff; it was set down gently, one day at a time, by people who had simply stopped wanting more of it. We flagged the same pattern in Micron on Tuesday, where a 7% drop happened on below-average volume.

The rest of the neighbourhood took it worse. AMD fell 8.00%. Broadcom fell 6.24%. Meta fell 6.77%. Technology was the worst of the eleven sectors at -3.53%, and the Nasdaq's -2.41% against the Dow's -0.85% is the entire story in two numbers. When one index loses three times what another loses in the same five days, that is not a market having a bad week. That is a market changing its mind about one specific idea while everything else carries on.

The idea it is changing its mind about is how all this gets paid for. On Thursday, Bloomberg reported that Broadcom is in talks with lenders to raise more than $60 billion in debt for an AI chip deal benefiting Anthropic, following a partnership it struck in June with Apollo and Blackstone to finance a $35 billion expansion of that company's computing capacity. Note the tense: talks, reported, not signed and not funded. But the direction is unmistakable. Two years ago, the AI buildout was paid for with petty cash by companies with more money than they knew what to do with. Now it is being paid for with borrowed money, and borrowed money comes with a lender who wants a schedule, a covenant and a fairly specific answer about when this thing earns anything. Utilities, the quiet backdoor way to own the AI power buildout, fell 3.47% this week, which suggests the market is repricing the whole chain, not just the chips.

Nothing kills a story faster than someone asking to see the repayment plan.

Retail week went eight for eight. Home Depot beat. Target beat. Lowe's beat. TJ Maxx beat. Ross beat. BJ's beat. Deere beat. And Walmart beat. Walmart earned $0.81 a share against $0.73 expected, grew revenue 5.9% to $187.9 billion, and raised its full-year sales outlook to 4%-5% from 3.5%-4.5%. Then it closed down 9.15% on 3.73 times its average daily volume. Beat, raised, and taken out behind the barn. We wrote it up the next morning.

The number that did it was comparable sales, which is the only retail figure that matters and the one nobody explains. It measures sales at stores that have been open at least a year, so it strips out the growth you get from simply opening more stores. Walmart's US comparable sales, excluding fuel, grew 2.6%, against a Street estimate of 3.67%. That is a miss of about one percentage point, and it took 9% off a company the size of a small country, because the stock was trading at a 37.6x forward earnings multiple. Pay a growth-stock price for a grocery store and the grocery store has to grow.

The contagion was picky, which is the useful detail. On Thursday alone, Costco fell 2.45% and Kohl's fell 8.01%, while the retail sector ETF ended the week down 1.39%. Target, which had reported the day before and beaten, closed Thursday down just 0.47%. So this was not "investors gave up on shopping." This was investors deciding that the single best operator in American retail, the one that wins when times are hard because everybody trades down to it, could only squeeze 2.6% more out of the American customer. That is a statement about the customer, not the company.

When the cheap store cannot grow, the problem is not the store.

On Thursday morning, the Philadelphia Fed's manufacturing survey printed 47.4 against an estimate of 25, its second straight month of humiliating the forecasters. On Friday, S&P Global's flash composite reading for the whole economy came in at 56.0, up from 54.5 in July and the expected 53.2, which S&P Global says is the strongest since April 2022. Any number over 50 in these surveys means more companies said things got better than said things got worse, so 56 is loud. S&P Global reckons it points to growth near a 3% annual rate this quarter.

Now the other file. Consumer sentiment came in at 51.0 for August. Retail sales fell 0.4% in July when economists expected a gain. Walmart's shoppers gave 2.6%. And inside that booming Friday survey, the factory output component actually fell to 51.9 from 53.9, a 13-month low, while services jumped to 56.8. So the "booming" economy is booming in offices and software and consulting, and it is not booming in the places where things get built, which happen to be the places where a lot of readers of this newsletter work.

Both files are real. Businesses are confident, spending on equipment and hiring, with the Philly Fed's employment gauge nearly tripling to 27.9. Households are miserable. That is less a contradiction than a description of who the last three years went well for. The bond market did not agonize over it: on Thursday the ten-year Treasury yield rose to 4.706% and the thirty-year to 5.251%, both up about six basis points, which is six hundredths of one percent and the polite way markets say rates are staying up.

The economy is booming. It is simply booming somewhere else.

Gold rose 5.45%. Gold mining stocks rose 14.29%. Bitcoin rose 18.15% to $74,405.88. Oil rose 6.35%. Every one of those had a better week than every one of the eleven stock sectors. If you had told a room full of investors in January that the winning trades of an August week would be a metal, the companies that dig up the metal, an internet coin and a barrel of crude, they would have asked what went wrong. Fair question.

Something did. On Wednesday morning the Treasury announced it is at least doubling the size of its long-end buyback operations, from a $2 billion maximum per operation to at least $4 billion, covering longer-dated government bonds. A buyback here just means the government going into the market and purchasing back its own outstanding debt to keep trading in it orderly. Two things about it matter, and both get lost in the coverage. First, nothing has been bought yet. The larger operations begin on September 9. Second, this arrived two weeks after Treasury had said buybacks would stay the same size, so it is a reversal, and it landed the morning after the thirty-year yield touched 5.339%. We covered that collision the next morning. Gold rose 3.84% that day, and bitcoin rose 7.09%. Draw your own conclusions about what people think it means when the government announces it is going to be buying more of its own paper.

The same afternoon, the minutes of the Fed's late-July meeting said policy tightening "would likely be necessary if inflation did not decline." So on a single Wednesday, one arm of the government sounded worried about inflation, and another arm announced it would buy more bonds, and the assets that go up when people distrust paper money all went up together. Bitcoin got its own extra push from a White House crypto event where the President pressed Congress to pass the CLARITY Act, a crypto rulebook that has passed the House, has not passed the Senate, and faces its first real Senate procedural vote around September 15. Announced is not passed. Passed is not law.

Pulled live Friday from Polymarket's economy board, ranked by money actually wagered.

Look at the volume column and then look away. Ninety-five million dollars is riding on what the Fed does and twenty-two thousand on whether the economy grows. The people betting real money have decided the referee is more interesting than the game, and after this week they have a point.

That 86% is the line worth keeping. The market thinks there will be no rate cuts at all this year. Not fewer. None. And with three Fed officials having voted in July to raise rates, the surprise nobody is positioned for is a hike, not a cut.

Gold rose 5.45% on the week and is up 36.31% over the past year. Not the miners, not a leveraged product, just the metal, up more than a third in twelve months while doing absolutely nothing productive with its time. It pays no dividends, reports no earnings, employs no one, and has now beaten most professionals who have jobs, dividends, and earnings.

The miners did better still, up 14.29%, which is how mining works: the cost of digging a bar out of the ground barely moves when the price of the bar jumps, so the profit on each one balloons and the shares run several times faster than the metal. It cuts exactly as hard the other way, which is the part the brochures skip.

Grandma is winning again, and she is not being gracious about it. She bought the coins because a man on the radio frightened her in 2011, held them through a decade of being laughed at over Thanksgiving dinner, and now they are up more than a third in a year while her grandson explains what a GPU is.

The thirty-year mortgage rate came in at 6.65% on Thursday, down from 6.67% the week before. Two hundredths of one percent. On a $350,000 loan, that is about $5 a month, which will not change anyone's life but is at least pointing downhill.

The data underneath went two ways at once. Housing starts fell 12.4% in July to an annual pace of 1.239 million, badly missing, while building permits rose 5.0% to 1.443 million and beat. Starts are shovels in the ground today; permits are paperwork for shovels later. Builders filed more plans and broke less ground, which is what people do when they want to be ready to build and do not want to be caught holding the lumber. Pending home sales fell 2.3%, and builder sentiment sits at 35, where under 50 means most builders call conditions poor. It has been under 50 a long time.

For an actual human, the week changed nothing. Rates are near 6.65%, houses cost what they cost, and the market is still shut to anyone who does not already own one. The only real good news is that somebody filed a lot of permits, which means they still believe buyers are coming.

Here is the strangest reading of the week. News sentiment on Nvidia remained strongly positive every day, scoring between 0.77 and 0.85 on a zero-to-one scale, with 136 to 271 articles per day. The coverage was upbeat Monday through Friday. The stock went down Monday through Friday. Somewhere between the headlines and the order book, several billion dollars quietly disagreed with the press.

Walmart ran the opposite way. Sentiment sat at 0.93 on Monday, based on 25 articles, then collapsed to 0.43 on Thursday, based on 101 articles. Four times the coverage, half the warmth. That is the sound of a story getting big and getting worse at the same time.

And Moderna, on the day it rose 177%, scored 0.29, the lowest reading of anything we tracked all week. The machine that reads the news is not built to understand that "melanoma," "tumor," and "recurrence" appearing in every headline can be the best possible outcome. It saw a wall of cancer words and marked the stock down. The humans, who could read, bought it with both hands. Chalk one up for the humans.

The week arrived as a tasting menu at a restaurant that had lost its head chef and not told anyone.

The amuse-bouche was Monday, three tiny beige things on a wide plate, technically food. Tuesday brought the appetizer: a semiconductor selloff plated as a warning, which the room ate anyway. Wednesday was the main course, and it was chaos, because two dishes came out at once: a hawkish set of Fed minutes and a Treasury bond announcement that flatly disagreed with it, while at table nine a biotech nobody recognised was set on fire tableside and the entire dining room stood up to watch. Thursday's course was Walmart, which arrived beautifully presented, beat every expectation on the menu description, and was returned to the kitchen nine percent lighter.

Dessert on Friday was oddly pleasant, a warm survey saying business is booming, served to a room full of people who had just read their own grocery bill.

The bill came in gold, and it went up 5.45%.

Strip out the noise and this week did one thing: it separated the crowded trade from the good trade, and they turned out not to be the same trade.

The crowded one is the AI complex, and it did not blow up. It just drifted lower on ordinary volume, five sessions running, with nobody screaming. That is a much more serious signal than a crash, because a crash is a group of people panicking and a slow bleed is a group of people rethinking. The question underneath it changed this year from "how big can this get" to "who is lending the money and when do they want it back," and Broadcom reportedly hunting more than $60 billion of debt is what that new question sounds like out loud.

The good trade was everything nobody wanted to talk about. A metal. The companies that dig up the metal. A coin. A barrel of oil. And one small drug company that spent this decade as a punchline and then published a trial result that may genuinely matter to people who have never owned a share of anything. All of it beat every sector of the stock market in the same five days.

The one question that decides next week is whether the headliner can still play. Nvidia reports Wednesday after the close, the Fed's preferred inflation gauge lands that same morning, and the new Fed Chair speaks Friday from Jackson Hole. Three chances to find out whether this was a pause or a verdict.

The opening act won't be able to top itself. It does not have to. It already got the standing ovation, and the headliner has to walk back out there on Wednesday and explain himself in front of everyone.

Follow us on X and Instagram for daily market updates

Markets don't wait, and neither do we. Follow Tracking the Trade for fast takes, the numbers that matter, and the occasional meme that hits a little too close to your portfolio.