Back above the line by nine points, with everything stacked into the next eighteen hours.

Twelve hours, two prints, and a tape built to sit still

1. The regime flipped back, barely. Tuesday’s third straight cash gain carried the index back over the flip by nine points. That restores the dampened tape — dealers selling strength, buying dips — which is exactly the state option sellers want on the day they are trying to pin the biggest print of the quarter.
2. The ceiling stepped down again. The dealer call wall came in another hundred points overnight, the second consecutive session of the ceiling moving toward the market while the floor stays put. The corridor between the walls is now two hundred points — the narrowest published band of the month — and cash sits twenty-three points under the top of it.
3. The Fed’s number arrives before the market’s number. Core PCE for July lands at 8:30 with consensus near 3.3% year over year, sharing the tape with the second estimate of Q2 GDP and durable goods. The front end carries roughly one-in-three odds of a September hike into it, eased from two-in-five a day ago. It is the last major input before the chair speaks Friday.
4. The barrel is doing the disinflating. Brent has now surrendered more than half of its two-week advance — down roughly 8.6% since Friday by Deutsche Bank’s count — and WTI printed the low 80s overnight after Tuesday’s settle at 82.36. Iran has restarted talks with Oman over a temporary Hormuz shipping corridor. The strait itself still ran three transits against a normal eighty-five on the latest count — the risk premium is deflating faster than the risk.
5. Nvidia reports tonight with the market cap of a G7 bond market in play. The Street wants roughly $92 billion of revenue and $2.08 to $2.09 of earnings, options imply a move between 5.4% and 7% — $280 to $313 billion of value — and the stock has fallen the day after each of its last four reports. Hedge funds spent the last month cutting it.
6. The rotation is the story under everything. Schwab’s Liz Ann Sonders told CNBC on Tuesday that rotation itself is the new momentum trade — and Tuesday proved it again: a third straight index gain built on roughly 180 advancers at midday. The index keeps rising while most of it does not.

The board before the doubleheader

Sentiment & flow gauges

The flow read

Nothing in the positioning data has repaired itself into tonight. Speculative accounts remain net short the S&P e-mini — the flip to net short is a week old and the next reading arrives Friday — and the prime-brokerage tape that emerged Tuesday says hedge funds spent the last stretch buying Microsoft and Amazon while cutting Nvidia, Alphabet, Meta, Broadcom and the equipment names underneath them. The discretionary book did not lighten the AI trade at the edges; it rotated inside it, away from tonight’s reporting company specifically.

Meanwhile the money that did move on Tuesday went to metal. The gold fund took in more than a billion dollars in a day, the weekly intake across gold funds is the largest in ten months, and the industry flow data shows domestic equity funds in outflow. Set that against a third straight index gain and the picture from Monday repeats: the tape is being carried by price-insensitive demand — index flows, the open corporate bid — while conviction money accumulates hedges in another asset class entirely.

Breadth said the same thing in one number: roughly 180 of five hundred advancing at midday on an up day. The index is thin, the rotation underneath it is violent, and both facts are load-bearing for how tonight resolves — a market this internally hedged does not translate a single-name shock into an index shock at par. That mechanism is graded live below.

Yesterday’s calls graded

The loudest twelve hours of the month

Scenario language describes how desks and pricing frame outcomes. It is not a recommendation.

A two-hundred-point corridor with everything crowded into the middle

Cannon Daily Levels · Pivots, Support & Resistance

Cannon Edge · Trend & 52-Week Range

Gamma levels from a public dealer-gamma (GEX) model, computed on the August 25 close.

What Cannon’s own levels say

The September contract’s daily pivot is 7,688.83, with R1 at 7,717.42, R2 at 7,742.58 and S1 at 7,663.67. Two convergences are worth sitting with this morning, because neither method knows the other exists. Today’s max pain converts to futures terms within a point and a half of that pivot — expiration arithmetic and yesterday’s high-low-close arithmetic have nominated the same number as the day’s centre of gravity. And the dealer call wall converts to within three points of R1: the nearest resistance on the pivot sheet and the nearest structure on the options map are the same price. When two levels built from unrelated inputs stack like that, the market tends to respect the pair more than it would respect either alone.

The near-dated option book adds a wrinkle the published map does not show. On today’s expiring E-mini chain, the heaviest call open interest sits at the 7,750 and 7,780 strikes — roughly 5,300 and 5,100 contracts — both above the all-expiry ceiling, and 7,750 sits within eight points of Cannon’s R2. On the put side the same chain’s largest position is at 7,600 with about 7,800 contracts, a full hundred points above the all-expiry floor, while the September monthly book’s heaviest put strike is 7,500 with over 80,000 — tying the published wall exactly. Read together: the fuller book agrees on the floor, but today’s book has drawn its own floor a hundred points higher and stacked its call bets beyond the published ceiling. Near-dated structure disagreeing with the aggregate in both directions is information, not error — the day book is braced for a wider day than the map implies, with the first real air pocket nearer than the official floor.

Trend and structure

Cannon’s trend board carries one change that deserves more attention than it will get. The S&P contract has lost its short-term uptrend arrow — it now shows only the long-term one — and the Nasdaq contract has gone a step further, flipping to an outright short-term downtrend while keeping its long-term up arrow. Both index rows going structurally soft on the same board, hours before the earnings print the whole complex is priced off, is the quiet warning in an otherwise constructive page. Elsewhere the lag is the information: crude still shows a short-term uptrend the price has spent a week dismantling, the thirty-year Treasury remains the board’s only financial row pointing down on both horizons, and natural gas — down on both horizons for weeks — has quietly shed its short-term down arrow. The model will catch up to the barrel; the question is whether the index rows catch back up before it does.

The old trader, the four questions, and the two new chairs

The roster, and where it has landed

Two seats on this roster remain vacant with no fresh call attributed to either: Christopher Harvey, formerly of Wells Fargo, and Jonathan Golub, formerly of UBS.

The verdict on the buybacks arrived in a barrel

The fight over the Treasury’s programme went from technical to personal in twenty-four hours, and the sides are now fully drawn. Against it: Druckenmiller’s op-ed (Section 05); JPMorgan’s research team, where Maia Crook writes that the interventions “belie the underlying structural challenges and do nothing to address them” and James Sullivan compares them to “paying your mortgage with your credit card”; Goldman and Wells Fargo rates research, which finds the operations “unlikely to meaningfully reset rate levels even if scaled up”; Deutsche Bank’s George Saravelos, who names the whole apparatus “soft-form financial repression”; and Mohamed El-Erian, for whom it is “financial engineering — a Band-Aid.” For it: LaVorgna (Section 05), and — halfway — Yardeni’s line that gimmicks which calm the Bond Vigilantes are still calming. Apollo’s Torsten Slok split the difference with one word: the buybacks are a “cloud” over markets — not a storm, not a rescue, a thing that obscures.

Here is what the scoreboard actually shows: the ten-year has retreated roughly ten basis points from Friday’s high and the thirty-year has stopped making two-decade highs — and almost none of that is the buybacks’ doing. The disinflation arrived from the Gulf. Brent has given back more than half of its two-week advance, down roughly 8.6% since Friday by Deutsche Bank’s morning count, after Iran restarted talks with Oman over a temporary shipping corridor through Hormuz. The strait itself remains effectively shut — three transits on the latest count against a normal day’s eighty-five — and Iran’s parliament is simultaneously weighing a bill to ban US and Israeli ships and toll everyone else. That is two-way headline risk on a price that has only been travelling one way for a week. Commonwealth Bank’s analysts sketch the downside: if Hormuz flows recover to even half of pre-war levels, WTI trades toward the bottom of a $70–100 range. Wilson’s threshold arithmetic from Section 05 belongs next to that sentence.

Slok’s second Tuesday item deserves its own paragraph, because nobody else on the tape is writing it: “China Shock 2.0 is here.” China is now exporting up the technology stack — EVs, semiconductors, advanced goods — into the exact markets American companies expected to own. The first China shock hollowed out US manufacturing employment; the second, on his framing, aims at US corporate margins. On a day when the entire market is a referendum on one American chipmaker’s pricing power, the leading private-credit economist publishing that the pricing power itself is under structural attack is the kind of dissonance this section exists to record. The metal tape agrees with the anxiety: gold sits within a session of its three-month high with the largest weekly fund inflows in ten months, and copper is parked just under its 52-week high — hedges against both inflations, the monetary kind and the geopolitical kind.

A light book walks into a heavy print

The setup. Nvidia reports tonight with the Street at roughly $92 billion of revenue — about 97% growth — and $2.08 to $2.09 of earnings. The options market prices a 5.4–7% move, $280 to $313 billion of market value, among the largest single-stock earnings swings ever offered; dealer-positioning models put the stock in positive gamma with heavy structure at the 230 strike overhead and an air pocket toward 190 if 210 fails. Two facts complicate the straddle price: the stock has fallen the day after each of its last four reports, and its average realised earnings move over the last twelve quarters — north of seven percent — is larger than tonight’s implied. The market is charging less for this event than the event has historically delivered.

The previews. Bernstein’s Stacy Rasgon — Outperform, $315 — says the numbers “should be good” and moves the weight of the event to management’s commentary on the Rubin ramp, where build plans imply upside to a data-center trajectory already pointed at half a trillion dollars next year; his standing worry is “circular financing,” the vendor effectively funding its customers’ purchases. Wedbush’s Dan Ives argues investors “still underestimate” the company — demand running as much as fifteen-to-one over supply, models “meaningfully below” the next eighteen months, a “Vegas Strip 1955” buildout. Morgan Stanley published the counterweight on Tuesday: the AI-financing push is importing credit risk into a story the market still prices as pure growth. Goldman’s preview — covered here yesterday — carries the same double message: estimates above consensus, and a warning that routine excellence is no longer enough.

The positioning. The prime-brokerage tape says hedge funds enter tonight light: buyers of Microsoft and Amazon, sellers of Nvidia, Alphabet, Meta, Broadcom and the equipment complex beneath them — hedge funds and mutual funds are now split on the AI trade itself. Domestic equity funds are in outflow, the gold fund absorbed the day’s largest inflow, and the futures crowd remains net short the S&P. Yesterday’s argument stands unchanged: the asymmetry into this print is inverted from where the consensus conversation puts it, because the sellers have mostly already sold.

Also on the tape this morning. Intuit is trading lower after its fiscal-2027 outlook disappointed — the after-hours prints ran high-single-digit percent down, and the premarket quote has narrowed the loss; treat the magnitude as unsettled, the direction as not. Dick’s Sporting Goods carries a second day of its post-earnings collapse. Oklo is up double digits again on the nuclear-AI power theme. CrowdStrike reports alongside Nvidia tonight with its own double-digit-billion implied swing. And the overnight strength was concentrated exactly where tonight’s risk lives: Kospi up 1.8% and the Nikkei up 0.9%, led by the Nvidia supply chain — Asia has already voted for the good outcome.

Two days of silence, then the speech that sets September

There are no Fed speakers today or tomorrow — the pre-symposium quiet holds, and the Kansas City Fed publishes the Jackson Hole agenda this evening, which will fix the one detail markets still lack: the chair’s exact slot on Friday morning, expected near ten o’clock. Kevin Warsh has run the Fed since May 22 without a single set-piece address on his framework. Friday ends that, three weeks before the September 15–16 meeting, in front of a market that still cannot model him — the July meeting held at 3.50–3.75% with three dissents preferring a hike, and his public record since amounts to task forces and silence.

The pricing question remains the inverted one: whether the Fed hikes in September. The strip and the prediction markets now put those odds near one-in-three, down from two-in-five a day ago, with a cut priced at zero — and the entire easing of the hike premium tracks the barrel, not the data. This morning’s core PCE is the last major input before the keynote: consensus 0.2% on the month and roughly 3.3% on the year, with the GDP second estimate expected to mark growth down toward 1.5% in the same release. A soft print into a falling oil tape closes the hike case for September almost completely; a hot one forces the market to weigh a backward-looking July number against a live August disinflation it can watch on any crude screen.

Two voices frame the range of what Friday can be. Jeremy Siegel’s weekly commentary argues Warsh’s job is to explain his reaction function, not move policy — he would be “very surprised” by a hike before the November election, the cost of waiting for December being trivially small — while noting acidly that the Treasury’s curve management makes the market’s signals harder for the Fed itself to read: he is “not fond of the Bessent twist.” Rick Rieder’s standing observation points the other way entirely: there is “more firepower in terms of how you manage the yield curve sitting at the Fed” — the question he wants answered Friday is not hike-or-hold but whether Warsh acknowledges the curve-management toolkit at all. If he does, the Treasury’s buyback fight becomes a two-agency story, and every desk position built on “too small to matter” has to be re-run with a much larger balance sheet in the equation.

The closer