The morning brief (The Early Bird Curd) called a strong but narrow, catalyst-driven bid and wrote down the one thing that would prove it hollow: the equal-weight tape turning negative while the Nasdaq proxy held green. Three hours into the session, that kill condition has triggered. The equal-weight S&P proxy is -0.23% while the Nasdaq proxy is +0.22% and the cap-weighted proxy +0.23%. The average large-cap company is down on a day both headline indices are up, and the small-cap proxy is -0.87%.

That the morning named its own failure condition in advance, and that the condition then fired, is the most useful fact on the board — it converts a vague worry about narrowness into a dated, falsified premise. Eight of eleven sector funds are red. The advance is two companies wide.

The morning also flagged four questions for this read. All four now have answers, and three of them cut against the premarket framing.

Interim status of the nine morning leans. Seven are working, one has been killed outright, and one is working on price while its stated kill condition has fired underneath it. Final scoring belongs to the after-close report; these are interim reads only.

A calibration note the morning brief itself asked for. It flagged that the gold-short, dollar-long and materials-short tickets share one engine, and warned that nine tickets were closer to five independent bets. Today sharpens that: all three worked, but the dollar leg contributed almost nothing (+0.18%) while gold, silver and the crypto proxy fell between 1.6% and 3.1%. The common factor was not the dollar. It was real yields — which means three tickets that will score as three independent wins were, mechanically, one rates bet that happened to be right for a reason other than the one written down.

This session is a two-legged reversal that has since gone quiet, and the two indices are telling materially different stories about it.

The Nasdaq proxy is the dramatic one. It opened 692.11, drove immediately to 695.77 within the first ten minutes — clearing the zone that had capped it twice this week — and then failed. From 09:40 it fell almost without interruption to 680.05 by 10:20, a decline of 15.7 points, or 2.26%, from the high. The reversal bar carried 1.74 million shares, the second-heaviest of the session, so the failure was made on volume rather than drift. Since 10:20 it has done nothing but balance between roughly 680.5 and 686.5, and it has tested the underside of its volume-weighted average price at 686.14 at least four times without closing above it. It trades 685.07, still 1.07 below that average.

The cap-weighted proxy traced the same shape and then diverged. Its high was the very first bar (746.301), its low the same 10:20 moment (737.68). But where the Nasdaq proxy stalled, this one has ground steadily higher for two hours, posting a near-unbroken series of higher lows — 737.68, 739.28, 740.13, 740.45, 741.44, 742.55, 742.92 — reclaiming its 741.80 average price around 11:00 and holding above it since. It trades 743.40, 1.60 above that average and back in the upper third of the range.

The small-cap proxy never participated in the recovery at all: it opened 293.56, fell to 287.83, and sits 290.05, still -0.87% and below its own average price.

The day-character test asks whether price held one side of its volume-weighted average price and made progressively higher highs — a trend day, favouring continuation — or round-tripped through it, favouring mean reversion. Today it returns a split answer for the second session running, but with the roles reversed from yesterday. The Nasdaq proxy round-tripped violently and is now balancing below its average price: a range day. The cap-weighted proxy round-tripped and then reclaimed, making higher lows for two hours: a slow trend day off the low. Neither is the clean directional session the opening ten minutes advertised.

Dispersion is the one measure that has moved decisively and unambiguously. The standard deviation across the eleven sector funds has widened from 0.891 premarket to 1.231 now, and the top-to-bottom spread from 3.38 to 5.30 percentage points. The morning classified the day neutral mixed at low confidence because no archetype dominated; three hours of trading has resolved that ambiguity in the direction of more rotation, not less.

An honesty note on the regime label itself. The rotation classifier is unusually sensitive to one uncertain input today, and readers should know it. Fed the volatility gauge as up 6.43%, it returns risk-off defensive; fed the same gauge as up 1.11% — the figure implied by the confirmed prior close — it returns reflation cyclical. Both come back labelled high confidence. The sector evidence breaks the tie: a genuine risk-off day requires defensive sectors to lead, and today they do not — utilities are fifth, staples seventh and health care ninth of eleven. Nothing defensive is leading. So reflation cyclical is carried as the better-supported label, with the caveat stated plainly rather than buried: the classification rests on a volatility reading whose own source is internally inconsistent (Section 9).

What is robust across both runs, and therefore worth acting on, is narrower and more useful than the label: dispersion is high, and the artificial-intelligence, semiconductor, quantum and crypto clusters score zero favourability in both classifications. Those cluster readings do not depend on the disputed input at all.

One model output deserves an explicit warning. Both runs assign precious metals a 0.70 favourability, because the archetype assumes a defensive or reflationary day bids gold. Today it is doing the exact opposite — gold is down 1.59% and silver 2.99%. This is a rates-driven de-risking, not a fear-driven one, and gold behaves inversely in the two cases. Treat that particular model reading as contradicted by the tape.

Posture into the close. Respect the relative-strength axis, not the index direction. Path-to-close invalidation: the Nasdaq proxy closing back above 686.14 and holding would negate the failure structure that defines this read; on the other side, the cap-weighted proxy losing 741.80 would remove the one genuinely constructive element of the session.

This is the section that explains the whole session. The reversal at 09:40 was not an equity event; it was a bond event that equities responded to.

Treasury yields spiked to eighteen-month highs during the first hour. The ten-year rose more than six basis points to 4.738%, its highest since January 2025, and the thirty-year added five and a half to 5.26%, the highest since 2007. The long-bond fund is -0.85%. That repricing is what erased an overnight gain that had the Nasdaq future up more than a percent.

The confirmation is in what fell alongside it, and this is the number most likely to be misread today. Gold -1.59%, silver -2.99%, the spot-bitcoin proxy -3.11%, the materials fund -2.30% — and the dollar fund up only +0.18%. A dollar-driven move would show a strong dollar as the mirror image of weak metals. A real-yield-driven move shows exactly this: every long-duration and non-yielding asset marked down together while the currency barely moves, because the discount rate rose rather than the dollar.

The morning theme was that the market has stopped punishing capital spending and started punishing unquantified capital spending. At the single-stock level that thesis has been spectacularly confirmed: Amazon is +14.9% on cloud revenue accelerating 37% to $42.2 billion, and Alphabet is +5.8% alongside it. Microsoft holds +2.5% the day after its best session since 2008.

But the morning brief made a second observation — that the macro data cut the other way and was being comprehensively ignored. That is the part which changed at 09:40 this morning. The macro is no longer being ignored. Three data points landed hot in the space of ninety minutes (Section 7), yields went to eighteen-month highs, and the entire long-duration complex repriced within the hour.

So the corrected theme at midday is a two-sided market: the earnings channel is delivering genuine, quantified upside in a handful of mega-cap names, and the rates channel is simultaneously raising the discount rate applied to everything else. Those two forces are not cancelling out evenly — they are producing a market where roughly two companies carry the index while five hundred and the entire hard-asset complex go the other way.

The tape has been driven by two threads since the bell, one macro and one single-stock.

The rates thread is the session driver. Markets opened strongly, with the Nasdaq future up more than one percent on the overnight cloud results, and then Treasury yields spiked back to eighteen-month highs on renewed inflation concern, erasing the entire overnight gain within roughly an hour. Every subsequent move in metals, crypto and small caps traces back to this.

Apple is the single-stock thread, and it is severe. The stock is -9.6%, having gapped down to open at 304.81 from a 333.43 prior close and then continued lower to 301.46. The proximate driver cited is weaker-than-expected Services and Greater China revenue, offsetting strong iPhone and Mac sales. Notably, this arrives on top of the quality-of-earnings problem the morning brief had already identified independently — that roughly two percentage points of gross margin and $0.11 of earnings came from tariff refunds, reducing a headline beat to about two cents. Two separate problems in one report.

Apple alone accounts for most of the technology fund being -0.81% and tenth of eleven sectors on a day when Microsoft, Alphabet and Amazon are all firmly higher — a rare configuration where the sector aggregate actively misrepresents its own constituents in both directions at once.

Three releases landed today and all three beat consensus. That is the fuel behind the yield move, and it is the signature feature of this read: the morning brief could only reconcile the 08:30 release, and two more have printed since.

A consensus discrepancy worth recording rather than smoothing over. The morning brief carried the Employment Cost Index consensus as +0.9%, which would make the print exactly in line; the source used for this read gives +0.8%, which makes it a modest hawkish beat. The actual figure of +0.9% is not in dispute and is confirmed from the statistical agency. The consensus is the contested value, it is web-sourced in both cases, and it is labelled as an estimate accordingly. The distinction matters only at the margin here, because the other two prints beat unambiguously.

Taken together the three describe firm activity with firm wages — slower growth than hoped but no disinflation to justify easier policy. Against a backdrop where this week’s policy hold already carried three dissents preferring an immediate hike, that combination is what pushed the long end to its highest since 2007.

Still ahead. Today is the final trading day of July, so month-end rebalancing flows are a live factor into the close and are a genuine reason to discount the directional information content of the last thirty minutes. No further scheduled releases of consequence remain today; any Federal Reserve speaking events this afternoon were not retrievable for this run and are flagged rather than assumed absent.

Breadth is negative and has not improved with the cap-weighted recovery, which is the single most important qualification on the afternoon.

The cleanest measure is the new-highs-to-new-lows balance, and it is decisively negative across the whole market: 65 new fifty-two-week highs against 108 new lows. Split by venue it is more revealing still — the New York exchange is roughly balanced at 31 highs against 23 lows, while the Nasdaq is 33 highs against 78 lows, better than two-to-one negative. The weakness is concentrated precisely where the index looks strongest.

The equal-weight differential says the same thing from a different angle. The equal-weight S&P proxy is -0.23% against the cap-weighted +0.23%, a gap of 0.46 percentage points with the average company on the wrong side of zero. Premarket that gap was effectively nil. Small caps are -0.87%. A composite momentum measure across roughly five thousand names reads -1.91%, with 41.78% of constituents above their five-day average and 56.24% above their two-hundred-day.

The live tick, Arms index and advance-decline line — normally this report’s advantage over the premarket edition — were not retrievable from the quote source on this run. Breadth here is therefore measured from five confirmed substitutes rather than inferred: the highs-lows balance by venue, the equal-weight differential, the momentum composite, the moving-average participation shares and sector participation. Nothing has been estimated to fill the gap.

The volatility gauge reads 17.28, and it needs a caveat before it can be used. The quote source reports a net change of +1.11 alongside a percent change of +6.43%, and those two figures are not consistent with each other or with any single prior close — the point change implies a prior near 16.17, while the percentage is the point change divided by the current level. Against the 17.09 prior close confirmed from the local warehouse and carried by the morning brief, 17.28 represents a rise of roughly 1.1%: a mild bid, not a spike. The level is treated as confirmed; the change fields are treated as unreliable, and Section 3 documents how much the day-type label moves depending on which is believed.

At 17.28 the gauge sits in the mid band — above the fifteen calm threshold, below the twenty elevated line. A market that saw its technology index round-trip 2.3% intraday while the volatility gauge rose about a percent is a market pricing rotation rather than fear, which is consistent with credit staying firm.

The volatility term structure — spot against three-month — was unavailable for a sixth consecutive session, which leaves the volatility backwardation archetype formally unevaluable again. Its context would be wrong at a gauge reading of 17 regardless.

Retail survey positioning is carried from the 07-23 week and labelled as such: 29.6% bullish against 42.3% bearish. Net bearish retail sentiment is a mild contrarian positive and, importantly, rules out the euphoria that typically accompanies a genuine top. The intraday put-call ratio and the composite fear-greed reading were not retrievable this run.

The leaderboard is not what it appears to be, and the distortion runs in both directions.

Consumer cyclicals lead at +3.00%, but Amazon is roughly a fifth of that fund and is up 14.9%. Strip that single position and the sector is approximately flat. This is a single-name event wearing a sector wrapper — the identical distortion the morning brief flagged premarket, now larger.

At the other end, technology is -0.81% and tenth of eleven, yet Microsoft is +2.5%, Alphabet +5.8% and the chip designer Nvidia +0.7% within it. That reading is almost entirely Apple at -9.6%. The sector aggregate misrepresents its constituents at both ends of the table simultaneously.

Materials at -2.30% is the one genuinely broad move on the board, and it is the clean read-through from the yield spike rather than a company-specific story.

Semiconductors are the most instructive line item of all. The basket is +0.41% on the day — apparently green — but it opened at 557.50, reached 561.44, and has since fallen to a low of 535.24, now 541.13 and 4.35 below its average price. A basket that has given back a 4.2% gain to sit fractionally positive is not participating; it is distributing.

Multi-period context confirmed (vendor, 07-31 midday). The weekly column shows consumer cyclicals +6.67% and communication services +4.17% leading, with utilities -3.19% and industrials -1.98% worst. The contradiction the morning identified persists: consumer cyclicals post the best week on the board while remaining the worst year-to-date at -3.55%. Technology is -0.27% on the week despite two extraordinary earnings nights — the damage from the earlier de-rate has not been undone. Energy is the strongest year-to-date at +32.17% but has rolled over on the week at -0.31%.

The overnight reports have produced the widest single-day dispersion of the season so far, and how each one is trading validates the standard the morning brief proposed: demonstrated return plus disclosed forward commitment.

The spread between the best and worst mega-cap reaction today is roughly 24.5 percentage points. That, rather than any index level, is the clearest single statement of what kind of market this is.

Every level the morning brief marked as breached to the upside has since been retested, and they have not resolved the same way.

Considered and declined, with reasons.

The through-line of this session is that a bond-market repricing interrupted an earnings-driven rally, and the two forces have since split the market in half rather than resolving.

The chain is traceable end to end. Three economic releases beat consensus between 08:30 and 10:00 — employment costs, Chicago activity and consumer sentiment. The ten-year yield rose more than six basis points to its highest since January 2025, the thirty-year to its highest since 2007. Within an hour the entire long-duration and non-yielding complex was marked down together: gold, silver, the crypto proxy and materials, with the dollar barely moving — the signature of a real-rate move rather than a currency one. The technology index gave back 2.26% from its high and has failed at its average price four times since. Meanwhile Amazon rose 14.9% on cloud revenue that accelerated 37%, and the cap-weighted index quietly ground back above its own average price on a two-hour series of higher lows.

Both things are true at once, and that is the regime. Four independent measures agree the advance is narrow — the equal-weight proxy negative at -0.23%, new lows outpacing new highs 108 to 65, small caps at -0.87%, and eight of eleven sectors red. What argues against reading that as an imminent break is equally concrete: credit is firm at -0.13%, the majority of companies remain above their two-hundred-day averages at 56.24%, retail positioning is net bearish rather than euphoric, and the broad index has reclaimed and held its levels. Narrow is not the same as fragile.

Predicted path, now to 4:00 PM. The base case is a continued split tape: the broad index holding its 741.80 average price in a slow drift, and the technology index remaining capped beneath 686.14 without breaking down decisively. Two specific factors argue against a clean directional close. First, this is the final trading day of July, so month-end rebalancing will dominate the last half hour and should be read as flow rather than conviction. Second, the yield move that drove the session is itself the variable — if the long end settles, the pressure on the long-duration complex eases mechanically and the semiconductor basket is the first thing to lift.

Same-day invalidation, both directions. The bearish trigger is the broad index losing 741.80 on expanding volume, which would end the only constructive structure of the day and likely pull the technology index back toward its 680.05 low. The bullish trigger — the one that would invalidate the caution running through this entire report — is the technology index closing back above 686.14 with the equal-weight proxy turning positive. Price recovering while the equal-weight tape stays red is a continuation of the same narrow structure, not a resolution of it.

Into the weekend. Positions carried past 4:00 PM today carry a two-day gap with Middle East front lines described as expanding and crude bid at $85.10. That exposure cannot be managed once the bell rings. Everything surfaced in this report is same-day by construction.