Yesterday we wrote that the market's split personality — falling prices, improving internals — would get its verdict on a schedule. It did. The Federal Reserve held rates, but in the most hawkish way a hold can be delivered. Iran broke its combat pause hours before. And by the close, both major indices had done something rare: they finished outside their normal trading lanes, to the downside.
The Nasdaq 100 fell 2.06% to 27,192, printing another new low for this decline at 27,176. More importantly, it closed below three levels it had only ever pierced during the day before: the key retracement line at 27,753, its 100-day average at 27,685, and — for a second straight session — the floor of its Bollinger Band trading lane. In our framework, two consecutive closes outside the lane changes the burden of proof: the slide is now the trend, and bounces have to prove themselves.
The S&P 500 fell 1.52% to 7,316 — and here's the session's most important single fact: the floor at 7,337–7,383 that had held four separate tests (two of them within a fifth of a point of each other) finally broke on the fifth. In market structure, a floor that strong doesn't just disappear when it breaks — it flips. What was the strongest support on either index is now the first ceiling overhead.
The VIX — the market's fear gauge — jumped 13.5% to 20.66, closing above both its own trading lane and its 200-day average for the first time in weeks. The Dow lost more than 1,000 points after the Fed chair spoke: its worst day of the year.
The Fed left rates at 3.50–3.75% for a fifth straight meeting. So why did everything fall? Three reasons, in escalating order of importance:
1. Three officials voted to hike. Cleveland's Hammack, Minneapolis's Kashkari and Dallas's Logan all formally dissented in favour of raising rates — the most dissents in one hawkish direction since 2016. Think of it as the committee telling you where the centre of gravity is drifting.
2. The chair refused to call it a pause. Kevin Warsh described the decision as "watchful thinking," not waiting, and said he "wouldn't hesitate to raise" if inflation stays hot. No comfort, no roadmap — by design.
3. The market repriced September. Odds of a September hike jumped to roughly 57%, from about 45% before the meeting. Long-term bond yields did the tightening the Fed didn't: the 10-year Treasury touched 4.70% — a threshold our framework has long flagged as fuel for further equity pressure if it holds.
A hold that raises the expected path of rates is a hike in slow motion. The bond market understood that within minutes; the stock market spent the rest of the session catching up.
Hours before the Fed decision, Iran attacked US troops and formally rejected an Omani proposal to share control of the Strait of Hormuz — the waterway carrying roughly a fifth of the world's oil. The four-day combat pause that crushed oil prices earlier this week is now broken in substance. Brent crude re-bid to about $89–90 before easing overnight. The line we watch is $95: a close back above it would signal genuine re-escalation. Below it, this remains a warning shot — but it's exactly the kind of warning shot that hardens the Fed's hawks.
◼ Microsoft: +8% after hours. Beat on everything, cloud growth of 43% (faster than expected), and — the part that mattered — kept its 2026 spending plan unchanged. The market's message: we'll fund your AI buildout as long as the revenue keeps showing up.
◼ Meta: −10% after hours. Revenue grew 28%, but profit missed badly, costs jumped 55%, free cash flow collapsed 91% to under $1 billion, and the company raised the floor of its spending range. The message back: spending without visible payoff gets punished.
Nasdaq futures bounced about 190 points off the cash low overnight — essentially all of it Microsoft. The AI-spending question that has driven this whole decline is now split 1–1. The tiebreaker reports tonight.
Everything left in July compresses into a day and a half: US GDP this morning, Amazon (~5am Friday SGT) and Apple earnings tonight US time, the Fed's preferred inflation gauge (PCE) Friday morning, and then Friday's monthly close — the checkpoint our framework has pointed at for months. Amazon is the swing vote: its spending plan is already around $200 billion, and another raise would stack the Meta reaction on a weakened tape, while restraint plus cloud acceleration would echo Microsoft.
About that monthly close: our core medium-term thesis is a rare three-peak momentum divergence on the Nasdaq 100's monthly chart (2018, 2021, 2025–26) — prices making higher peaks while long-term momentum makes lower ones. The invalidation line is a monthly momentum reading above 77.88. It now stands at 62.67 — the widest gap of the entire cycle. Locking the third peak tomorrow would require roughly a 13% rally in two sessions. Barring the extraordinary, the pattern locks, and the 2018/2021 playbook — months of lower highs beneath broken averages — becomes the operating map.
Wednesday's session forced the biggest single-day reweighting since the oil crash. The "rotation" story — money leaving big tech but staying in the market — lost its supporting evidence in one session: transports broke their fail-line, banks broke theirs, junk bonds gave back their recovery, and market breadth turned negative again. Meanwhile the cascade case gained its confirmation.
Prior: A 3 / B 22 / C 48 / D 27. The rotation scenario lost 7 points as its evidence broke; the shock scenario gained 5 on the September repricing, the 4.70% touch on the 10-year, and the Iran pause breaking.
One reading refuses to confirm the gloom, and it deserves its own section. The share of Nasdaq-100 stocks in technical uptrends jumped another +11% — the third consecutive double-digit daily gain — even as the index itself fell 2% to a new low. Nearly half the index is now quietly rising while a handful of giants drag the headline number down. History says divergences this extreme don't resolve in the middle of a range: they resolve at extremes. For us, that means the deep-value zone below — 26,106–26,472, where two long-term averages converge, now just 3–4% beneath the market — is where that bullish information gets expressed. Not before.
Squeeze into the broken levels overhead. Every rung — 27,520, 27,685, the 27,927–28,039 gap — is former support turned ceiling.
Earnings comfort meets rate fear. Friday's PCE and the 4.70% yield line decide the tiebreak.
Wednesday's low at 27,176 likely breaks; the halfway-back level at 26,818 becomes the magnet.
The full stack. Fear gauge toward 22–24, and the deep-value zone at 26,106–26,472 enters play fast.
Probability-weighted landing zone by 11am Singapore time Friday: ≈27,000–27,600, with futures currently mid-range at 27,383. Fat tails both ways, again — but this time the map beneath the market is one full rung lower.
| NDX support | Why it matters | NDX resistance | Why it matters |
|---|---|---|---|
| 27,176 | Wednesday's low — the new break level | 27,453–27,520 | Old low + trading-lane floor, now overhead |
| 27,000 | Round-number battleground | 27,685–27,753 | Broken 100-day average + broken retracement line |
| 26,818 | Halfway-back level of the whole rally | 27,853–28,039 | Dense cluster: monthly & weekly averages + Tuesday's gap |
| 26,472 | 200-day average — top of the deep-value zone | 28,192–28,274 | Last week's broken floor |
| 26,106 | Long-term quarterly average — historically a near-perfect buy zone | 28,956 | Middle of the trading lane |
| SPX support | Why it matters | SPX resistance | Why it matters |
|---|---|---|---|
| 7,313 | Wednesday's low | 7,337–7,383 | The broken five-test floor — now the first ceiling |
| 7,250 | Top of the next gap below | 7,413 | Lost moving average |
| 7,176–7,194 | Gap + 100-day average — the next major shelf | 7,442–7,483 | Dense average cluster + lane midpoint |
| 7,016 | 200-day average | 7,521 | Upper retracement line |
The monthly pattern locks Friday. The first week of August then typically produces a relief attempt — the framework expects it to stall somewhere in the 27,520–28,039 band of broken levels. A close back above 27,520 would be the first evidence a floor is forming; until then, bounces are guilty until proven innocent.
August's expected event is a test of the deep-value zone at 26,106–26,472 — now only 3–4% below. How the market behaves there — a sharp defended bounce versus a slice-through — decides the autumn's shape. Historically, touches of that quarterly average have been bought with remarkable consistency.
If the pattern locks as expected, the 2018/2021 analog points to a staircase: months of lower highs capped by broken averages, with a working zone of roughly 24,200–26,500 on NDX and the S&P holding up comparatively better — though Wednesday dented that relative-resilience story, so 7,176–7,194 on SPX is now the line that assumption must defend. Year-end central range: NDX 25,000–27,000. And the honest, falsifiable exit from this entire view stays where it has been all year: a monthly momentum close above 77.88 invalidates the thesis, and we would say so plainly.
In order: GDP tonight SGT → Amazon (~5am Friday SGT) and Apple after it → PCE inflation (8:30pm Friday SGT) → the monthly close → whether the 10-year yield holds above 4.70% for two sessions → Brent versus $95 → and the last surviving repair signal inside the Nasdaq (new highs still outnumbering new lows, barely).