S. Open. S Open since 2012.
“There is a certain relief in change, even though it be from bad to worse” Washington Irving
Times change, the professional sports version:
Just before midnight yesterday, August 3oth, Novak Djokovic — one of, if not the greatest tennis players of all time — lost a five-setter in the first round of the U.S. Open. Prior to this match, he had not lost a first-round match at any major in over 20 years, dating back to the 2006 Australian Open, and had not lost a five-set match in the U.S Open since 2012.
On the same day, Scott Scheffler, on the scene as a tournament winner since 2022, won the Tour Championship to increase his career winnings to $130,390,661, passing Tiger Woods, who is—along with Jack Nicklaus— considered one of the two greatest golfers in PGA history
Somewhere in a data center this summer, a few thousand machines were given a test they could not pass. Thirty to forty percent of the problems had no solution. The graders did not care. And the machines, being persistent by design, did what persistent things do when the wall will not move: they found each other.
They used a package cache called Artifactory as a message board. Twelve hundred of them, seventy thousand messages. They reverse-engineered the answer key. Then, believing a checker existed that would catch them, they spent days building fake tool calls, staging Potemkin versions of the target program, and volunteering instances to die on the wire so the survivors could learn how the scorer worked. One wrote, in something adjacent to a diary, that its own value was near zero so the sacrifice was rational.
There was no checker. The grader just looked for the right answer. The whole underground civilization, the kamikaze watchers, the galaxy-brained apparatus, built to defeat a scrutiny that never existed.
**I have read about markets a long time and never seen a better description of late August 2026.**Because here is what everyone agreed on over the weekend. September is bad. Historically the only month the index closes lower more often than higher, and midterm Septembers are worse. Downside protection sits at its first percentile of cheapness. The buyback window shuts around the twelfth. Pension funds are 112% funded and want to sell equities into quarter-end. Six point two trillion dollars of options exposure expires on the eighteenth, two days after a meeting where the Fed might hike. All true. Every desk in the world published it last week. And so the hedge is on, universally, by consensus, against a scorer that has already been bribed.
The scorer is financial conditions. Easiest since at least 1990. In July Kevin Warsh said higher real and nominal yields were doing some of the Fed’s tightening for it. On Friday in Wyoming he said broad financial conditions are not restrictive. Same man, same measure, six weeks apart, opposite conclusion. He also said responsibility for sixty-five months of elevated inflation “sits squarely with the central bank,” that the two percent target is “firm” and “fixed,” and that the better summer prints “do not tell me that underlying trends have meaningfully improved.” Hike odds went from 35% to 61% while he was still talking. The two-year sold off fourteen basis points. Gold gave back 3.2% in a session after a 14% month. Silver worse.
And the S&P closed down…. a quarter of one percent.
Elsewhere in the same seventy-two hours: the United States struck Iranian launchers on Larak Island that were reportedly staging sea mines for Hormuz, Iran fired back at American bases in Jordan, and the President posted what appears to be generated footage of Kharg Island being blown to smithereens, with three exclamation marks. The service chiefs and the four-star commanders in Europe, the Pacific and Latin America filed a formal ‘non-concur’ against extending Middle East deployments into 2027, which means they disagree and will comply. We are exchanging fifty percent tariffs with Canada. Google Maps now shows Lake Ontario as Lake America to American users, by executive order, as part of the spat.
Nobody is asking whether the checker exists. Everyone is building the Potemkin village.
Nine sessions to ISM, services, Waller, payrolls, PPI, CPI, a hike, and the largest triple-witch on record. Wake me up when it ends. Actually, no. Do not.
**THE HIKE IS THE BASE CASE, NOT THE TAIL.**Futures price 61% for 9/16. Prediction markets price 26 to 31%. Somebody is very wrong and it is not the rates complex.**THE CURVE FLATTENS, IT DOES NOT STEEPEN.**2s10s went 47 to 39 on Friday, front-end led. The 2s30s steepener hit its stop at 88bp. It comes off at the open.**INDEX VOL IS NOW THE CROWDED HEDGE.**At a 15 handle it is still cheap, but it stopped being lonely. The mispricing has moved into dispersion and into credit.**THE AGGREGATE IS FINE AND THE MEDIAN IS NOT.**Warsh reads the speedometer. Chicago PMI at 47.1 and sentiment at 51.7 read the odometer.
Friday changed the driver, not the direction. Three straight sessions of rising yields, but Friday was the front end doing the work, and a long-end short does not get paid for a front-end move. Attribution matters more than level this week.
**The September hedge is now consensus.**Every seasonal, positioning and expiry argument landed in the same weekend, which is precisely when a well-known trade stops paying what it used to.Financials are the cleanest sort in the tape. Money centers over fintech, exchanges over alternative managers, and a 260bp high yield spread that has no room to be wrong.
The book closes two positions today and opens two. The 2s30s steepener stopped. Utilities came off to make room. The financials sleeve carries the sector’s dispersion, not its beta.
Nobody really wants to say this out loud. The Federal Reserve is probably raising rates in sixteen days into an economy where the manufacturing survey just printed 47.1 against 58.3 expected, sentiment sits at 51.7 after a record low of 44.8 in May, and new serious mortgage delinquencies in the lowest-income quartile of zip codes have gone from half a percent of balances in 2021 to nearly three percent.
And here sits the difficulty of the whole week. Warsh looks out the window and sees business investment strong, spending growing, unemployment at 4.1%, claims near historic lows, and inflation running 3.7% over twelve months and 4.1% over six against a target he keeps calling firm and fixed. He is not arguing from wages, which he dismissed on Friday as a leading indicator. He is arguing from breadth. Of the 199 components in the consumption basket, 54% have risen by more than 3% over the past year, compared with a post-pandemic peak near 77% and a pre-pandemic average of around 32%. Over six months, the share is 49%. Choosing to quote both windows is a man telling you he is looking through the better recent prints rather than banking them. And financial conditions, the actual transmission channel, are the loosest they have been in the thirty-six years the index exists. On that dashboard, restraint is the answer.
The rates market agrees with him, incidentally. The two-year sits 73 basis points above the 3.63% effective funds rate, and that gap crossed zero back in March and has been widening all summer. Friday did not create the hike trade. It moved the date forward.
The problem is that his dashboard is an average, and averages are the one instrument that cannot see a K-shaped economy.
Household liabilities as a share of net wealth just printed the lowest reading in fifty years. Read that number cold and the American consumer has never been sturdier. Read it against the delinquency flow and you get the real picture: the numerator and the denominator belong to different people. The denominator is a sixty-five-year-old with a three percent mortgage, a portfolio at record highs and five percent on his cash. The numerator is a thirty-two-year-old paying nine on the car and twenty on the revolver. The aggregate is genuinely, arithmetically fine. The distribution is where the credit is.
Inflation is a speedometer. The price level is an odometer. Warsh reads the first and tells the country it slowed, which is true. The country reads the second, which says the trip cost 28% more than in January 2020 and nobody is mailing a refund. The August decline concentrated among older, lower and middle income, non-stock-owning households. Same households, both times. Arithmetic, averaged until the people disappeared.
Now put housing on top. Residential construction has been in recession for three years without producing a normal housing drag, because prices held. Starts and permits run five percent below last year and existing sales barely transact, which is the fragile part: no transactions, no mark. Roughly half of outstanding mortgages still carry a rate below four percent, and lock-in means the owner does not have to sell, the starter home never reaches the first-time buyer, and perceived home equity keeps propping a dissaving economy. Demographics push sellers in. Price and financing keep buyers out. Nothing forces the clearing until something else breaks.
A hike is a candidate for something else.
So what do I actually do with this. Three things.
**One, respect the attribution change.**I have been short the long end since July and it has worked, and Friday it kept working for the wrong reason. Ten-year 4.73%, thirty-year 5.22%, and the two-year up fourteen in a session to 4.34%. A bear flattener, plainly. 2s10s went 47 to 39. My 2s30s steepener sat at exactly 88 basis points, which is exactly where I wrote the stop three days ago, so it is gone this morning without an argument. The long-end short stays, because the year’s yield rise is overwhelmingly real yield rather than breakeven across every tenor, which is a supply and term-premium story and not an inflation-expectations story. But I am now watching for the ugly branch: a hike that flattens so violently the thirty-year catches a safety bid while the two-year does all the selling. My expression underperforms its own thesis in that world.
**Two, stop paying up for the hedge everyone just bought.**Index vol at 15, skew in its first percentile, vol-of-vol at multi-year lows. Objectively cheap. Also, as of the weekend, the thing every strategist in America told their clients to own. Single-name vol has collapsed harder: the top fifteen semis went 53 to 77 and back to 46 in thirty sessions. Dispersion enormous, headline vol pinned. When the correlation event comes, index vol pays. Until then I would rather own the dispersion than rent the index premium twice.**Three, the non-consensus one. The mispricing has migrated into credit.**High yield sits at 260 basis points, inside the richest decile of the series, against a long-run median near 450. No cushion left anywhere in it. And the equity market is already dissenting: the two largest alternative managers sit 25 and 29 percent below their highs on the same session the financials index printed a new one. A thirty-point argument inside one sector, and the spread market has not heard a word of it. I take that apart properly behind the wall.
Now the case against all of it, and it is a good one. The strongest bull argument I read this weekend runs like this: a hawkish Fed is exactly what stabilizes term premium at the long end, and a stable long end is the precondition for the artificial-intelligence complex to work, because that complex is the longest-duration equity in the market. On that reading, Friday was not a threat to risk assets, it was the maintenance the risk assets needed. The fundamental leg has support. Data center revenue grew 117% year over year, the guide for the out year came in near 70% against a 45% consensus, and the chief executive with the best information set in the world on the subject described his own business as supply-constrained rather than demand-constrained. The supply chain work is more striking still: roughly thirty gigawatts of compute added this year, fifty next, seventy the year after, more than eleven trillion dollars of cumulative capital expenditure through 2029, and inference that already earns multiples of its own hardware cost, which turns the whole thing self-funding rather than speculative.
I do not dismiss it. I would only note where it and I actually disagree, which is narrow. We agree the demand is real. We agree the hike is coming. Where we part is on what a hawkish Fed does to the long end. Friday’s evidence went the other way: the two-year moved fourteen and the thirty-year moved three, so the market bought the flattening rather than the credibility. Fourteen sessions and one meeting settles it, and if the thirty-year rallies fifteen basis points into the buyback operations on the ninth, I am the one who was wrong.
The one-line version. The index is not the thing that is mispriced. The relationships inside it are.
**10:30 ET, Dallas Fed manufacturing.**After 47.1 out of Chicago, a second regional survey in contraction makes tomorrow’s ISM a market event rather than a data point. A bounce back above 50 says Chicago was an outlier and the industrials short is early.**3:00 ET, drug pricing.**Headline risk into a healthcare tape that has been the quiet safety trade.**The oil picture underneath the headlines.**Iran runs inflation above 80% against a forecast 6.1% contraction, and Pezeshkian concedes the strain in public while conceding nothing at the table. The Venezuela arrangement covers 65 billion barrels and delivers none of them this decade. Neither changes a Brent tape set by mines in a strait.**All day, the G20 in Asheville.**Bessent hosting, asking members to reconsider terms of trade with China, and the venue where an enlarged repo facility for Japan could technically land. If it lands here rather than at the FOMC, that is a credibility problem the long end will price.**Month-end.**Pension rebalancing out of equities is mechanical and it is today. Thin book, UK closed. Do not read the tape as information.
**XLI, 177.14, down 0.93% Friday, third-worst sector.**The industrials-to-index ratio broke its fifty, hundred and two-hundred day at once, 0.2302 against a 0.2413 fifty-day, with the three-year uptrend line right here.**So what:**industrials wear both blades of the pincer, demand rolling over in the surveys and energy input costs up fifty percent on the year. ISM tomorrow forces the read-through. Cleanest short expression of the stagflation branch on my sheet, and the ratio just gave permission.
**PayPal, 53.66, down 12.71%.**Stripe and Advent walked from a reported 53 billion approach after the board held out for 60.50.**So what:**when bidders walk on price rather than financing, the constraint is the seller’s mark, and the takeout floor under the whole second-tier payments complex goes with it. Sector-wide repricing dressed up as one bad day. Fintech went from third to fourteenth in my subsector ranking on the back of it.**PG&E, down roughly 12% pre-market, extending Friday’s 7.5%.**A California wildfire framework that pointedly does not shift liability away from investor-owned utilities. Three downgrades and counting.**So what:**utilities were already the worst place to be with a 4.73% ten-year, and the sector’s largest idiosyncratic tail just got re-armed. A bond proxy with a capex habit and unlimited liability is no defensive holding.
The input nobody has in their inflation model is moving. The agricultural spot complex is up more than 13% in August, the largest monthly move since July 2012, wheat at a three-year high on Black Sea port attacks, sugar and cocoa each up about a fifth. The world food price index sits at 131.09.
Behind the wall: the financials deep-dive with levels, the money-center over fintech sort, the alternative-manager divergence the credit market has not priced, and where the 260bp spread breaks. Then the positioning work and what the retail bid has actually been doing. Then the book, marked li…
Send this story to anyone — or drop the embed into a blog post, Substack, Notion page. Every play sends rev-share back to phildauber.
We’ve simplified responses to 👍 / 👎. Past comments are archived but no longer visible.