The Sunday Sundae

Sunday, 08-02-2026
Weekly market read · week of 2026-08-03
The Milkman
OuroTaurus
Built Sun 2026-08-02 ~7:00 PM ET · anchored to Fri 2026-07-31 close · full 5-day week, no holidays · a Saturday-night Iran de-escalation headline landed after the anchor Re-confirm at Monday's open via The Early Bird Curd

01. Last Week's Carryforward

Two weeks of weekly reads are missing, and the honest scorecard is empty rather than good. The last Sunday Sundae was built 2026-07-19 for the week of 7/20–7/24. No Sundae was built 2026-07-26, so the week of 7/27–7/31 — the week the Fed held with three hike dissents and megacap cloud added roughly $1.5T — carries no weekly prediction at all. And the two setups from 7/19 remain unscored: the Nightcap has not written their outcomes, so every sundae row in the track record still reads v: null. This card therefore reports the factual price path, clearly labelled as unscored — it is not a FIRE/VOID verdict, which only the evening report may issue. confirmed [track-record.jsonl, brief archive]
Tracked setup (built 7/19, for week 7/20–7/24)StatusUnscored price path
Oversold Mean-Reversion — semiconductors, longsun-260719-OMR-SMH-L PENDING Armed Tue 7/21 — SMH traded 586.00 through the 567.41 arm and QQQ closed 708.97, above the 702.30 condition. It paid early: SMH reached 592.01 Wed 7/22, about 4.3% above the arm. Then the kill triggered Fri 7/24 when QQQ closed 684.23, under the 686.76 line. That kill was the whole value of the setup — SMH went on to trade 503.63 on 7/29, a further 6.8% below the 536.81 kill level, and the 522.65 stop would have been hit had the exit been ignored. confirmed [thinktank warehouse, adjusted daily bars]
Exhaustion Fade — extended energy, shortsun-260719-EXF-XLE-S NEVER ARMED The arm required both a genuine de-escalation headline and a failed high in energy. Neither arrived inside the 7/20–7/24 window, so nothing was risked. The headline leg arrived this weekend — see Section 13, where this thesis returns with the first of its two conditions now met. confirmed [brief archive, weekend wire]
Lens The framework's discipline held up better than its coverage: the one setup that armed was rescued by its own kill condition, and the one that never armed correctly refused a trade that had no catalyst yet. The lesson to carry is about presence, not accuracy — two unbuilt weekly reads meant the tape's most violent stretch went unforecast, and a missing brief cannot be scored, only regretted. Treat this week's calibration as thin and weight the live evidence in Sections 8 and 9 more heavily than any inherited conviction.

02. Friday → Sunday Tape

Lens Friday's close was strong at the index and weak nearly everywhere else, which is the single most important fact going into Monday. When the S&P finishes near its high while the average stock, small caps and equal-weight all finish red, the gain is being manufactured by a handful of very large names rather than earned by the market. Hunt long setups inside the megacap cloud and consumer complex where the buying actually is — and treat semiconductor strength as guilty until proven innocent after Friday's gap-up rejection. The weekend Iran headline is a fresh, unpriced input that lands squarely on energy Monday morning.

03. This Week's Regime

Two Markets, One Index — a dispersion regime with a peace-deal wildcard
Confidence: high on the diagnosis · low on the path · horizon: 2026-08-03 → 2026-08-07
35% — Dispersion persists, index grinds. Megacap earnings support holds, the damaged complex keeps bleeding on its own prints, S&P chops 740–756 while breadth stays under 50%. The base case precisely because it requires nothing new.
25% — The unwind bottoms on earnings. AMD (Tue) and SanDisk (Wed) clear a very low bar, semis reclaim Friday's 561.44 high, breadth broadens and the index breaks out above 755.58 with participation.
25% — The cracks reach the index. QQQ leads lower, SPY loses the 744.99 50-day shelf, and the 43% advance figure resolves down toward 729 rather than up. A hot payroll Friday is the most likely trigger.
15% — Peace-deal rotation. Hormuz reopens, crude breaks hard, energy is crushed, and the disinflation impulse fuels a risk-on rotation into rate-sensitives — good for the index, brutal for last quarter's energy winners.
Lens The week's strategy follows directly from the correlation number: this is a market that pays for picking and punishes buying the index in either direction. With individual names pricing large moves and the index pricing small ones, single-name and single-group setups carry the edge, while index-level directional bets are being offered at their worst odds in months. Favor group-level mean-reversion and catalyst-driven fades over broad market calls, and let the earnings calendar — not the tape's opening hour — set your entry timing.

04. Cross-Asset & Credit

Lens Nothing in credit or rates corroborates a risk-off reading, which materially lowers the odds of the bearish path turning disorderly rather than merely corrective. The live cross-asset story is instead an unwinding risk premium: crude already down 5.5% before a de-escalation headline, gold sold, the dollar soft. That points the week's best short-side hunting at energy and commodity-levered names rather than at the index, and it quietly supports rate-sensitive longs if payrolls cooperate. Confirm the high-yield spread Monday before leaning hard on any of it.

05. Macro Theme (week ahead)

Lens The dominant macro narrative is a Fed that has stopped cutting and now contains members arguing to tighten, walking into the month's most important labour print. The asymmetry matters for positioning: a soft payroll number is merely reassuring, while a hot one hands the dissenters their argument and threatens rate-sensitive and long-duration equity trades hardest. Keep duration-heavy longs smaller into Friday, and note the week's real risk sits in a single 8:30 AM release rather than spread across the sessions.

06. Geopolitical Pulse

Lens This is the week's largest unpriced input and it is deliberately two-sided: the stated terms would remove the single biggest supply constraint in the oil market, but the announcement's own track record is a string of reversals. The tradeable expression is a fade of the energy risk premium on confirmation, not an anticipatory short into a headline that has failed before — which is exactly how Section 13 frames it. Energy producers and refiners are where this lands first; airlines and transport are the second-order beneficiaries if crude genuinely breaks.

07. This Week's Calendar

  • Full five-day trading week — no holidays, no half-days, and the following Monday (8/10) is a normal session. confirmed [_shared/market-calendar.json]
  • Mon 8/3 — Palantir reports, consensus revenue near $1.81B (+81% year over year). The week's first read on whether AI enthusiasm still clears a very high bar. confirmed [TradingKey week-ahead]
  • Tue 8/4 — JOLTS job openings (June); AMD reports, watched for MI450 and Helios production detail, consensus near $1.61 EPS on about $11.3B revenue (+47%). confirmed [Kiplinger, TradingKey]
  • Wed 8/5 — SanDisk reports: the first pure-play NAND memory result since the stock fell 53% from its all-time high. The cleanest available test of whether the memory unwind is a valuation event or an earnings event. confirmed [TradingKey]
  • Fri 8/7 — July employment report. The week's dominant macro variance event. confirmed [Kiplinger]
  • Also reporting: Eli Lilly, Caterpillar, McDonald's, Disney, Costco, CVS, Pfizer, Occidental, ConocoPhillips, Uber, Airbnb, Booking, Shopify, Arista, AppLovin, Axon, Datadog, Western Digital, Astera Labs, Vistra and Oklo. confirmed [Earnings Whispers, week of 8/3]
Lens The calendar is unusually well matched to the market's open question: the exact complex that broke — AI infrastructure, memory, semiconductors, power — reports Monday through Wednesday, and then the labour market rules Friday. That argues for taking group-level risk early in the week while the resolvers are dated and specific, and reducing exposure into Thursday's close ahead of payrolls. It also means any semiconductor reversal attempt this week has a real catalyst behind it rather than hope, which is what distinguishes it from the attempt that failed two weeks ago.

08. Breadth & Internals

Lens This is the most important section in the brief and it says one thing: the rally has no crowd behind it. A market that gains 1.1% on the week while barely half its members hold their fifty-day average, and where equal-weight has flipped from outperforming to lagging, is being carried rather than led. That argues strongly against index-level longs on breakout attempts and in favour of concentrating long exposure in the specific groups doing the carrying, while treating any breadth thrust above roughly 60% as the genuine signal that the bullish path has arrived. Until then, the burden of proof sits with the bulls.

09. Sentiment Watch

Lens A volatility index in the mid-to-high teens alongside collapsing correlation describes complacency at the index level coexisting with real fear underneath, and hedges were being taken off into a weekend carrying a live geopolitical wildcard. That is a thin cushion. The practical read is that index option protection is cheap relative to the single-name risk actually present this week, and that any adverse Monday headline would have to be absorbed by a market with less hedging than it had on Thursday. Do not confuse a calm index reading with a calm market.

10. Sector Flow at Week's Start

XLY Cons. Disc.+6.11%
XLC Comm. Svcs.+1.83%
XLF Financials+1.12%
XLP Staples+1.09%
XLV Health Care−0.01%
XLE Energy−0.12%
XLK Technology−0.30%
XLI Industrials−1.54%
XLB Materials−1.62%
XLRE Real Estate−1.92%
XLU Utilities−4.19%
Lens The rotation is not defensive and it is not cyclical — it is a rotation out of everything levered to the AI buildout and into the handful of companies actually monetising it. Utilities and semiconductors falling together while consumer discretionary explodes higher tells you capital is leaving the picks-and-shovels trade rather than leaving equities. Hunt longs in communication services and the megacap consumer complex, and treat utilities, materials and real estate as the week's weakest ground; the AI-power names remain a falling knife until the memory and accelerator earnings land midweek.

11. Earnings Reaction Watch

Lens Earnings reactions, not earnings results, are the signal to watch this week — the same quarter can produce a 15% gain or an 8% loss depending purely on whether spending is matched to revenue. Because these names enter their prints already deeply damaged, the setup is asymmetric: expectations have been reset hard, so an in-line result can produce an outsized relief move. That favours waiting for the reaction and trading the confirmation rather than positioning ahead of the print, and it is the mechanism by which the semiconductor reversal in Section 13 either arms or dies.

12. Key Levels at Monday's Open

S&P 500 · SPY
747.03 confirmed [Massive]
Resistance 748.89 (Fri high) · 755.58 (20-day high — the decider) · 760.40 (52-week high)
Support 744.99 (50-day) · 737.68 (Fri low) · 729.10 (20-day low) · 700.39 (200-day)
Average daily range 8.40 · 1.76% below its 52-week high
Nasdaq 100 · QQQ
687.99 confirmed [Massive]
Resistance 695.77 (Fri high) · 715.09 (50-day — reclaim = regime change) · 726.39 (20-day high)
Support 680.05 (Fri low) · 661.14 (20-day low) · 645.04 (200-day)
Average daily range 14.70 · 8.10% below its 52-week high and beneath its 50-day
Russell 2000 · IWM
291.20 confirmed [Massive]
Resistance 292.49 (50-day) · 300.41 (20-day high) · 302.72 (52-week high)
Support 287.83 (20-day low) · 265.86 (200-day)
Average daily range 3.97 · closed just under its 50-day
Semiconductors · SMH
540.53 confirmed [Massive]
Resistance 561.44 (Fri high — the rejection level) · 596.17 (50-day) · 618.21 (20-day high)
Support 535.24 (Fri low) · 503.63 (7/29 flush low)
Average daily range 26.66 · gapped to 557.50 Friday and closed at 20% of range
Energy · XLE
59.55 confirmed [Massive]
Resistance 59.69 (Fri high) · 60.45 (20-day high)
Support 58.32 (Fri low — the trigger) · 56.67 (50-day) · 53.00 (20-day low)
Average daily range 1.15 · the sector most exposed to the weekend headline
Lens Two levels decide the week. SPY 744.99 is the fifty-day shelf the index is standing directly on top of — holding it keeps the grind alive, losing it converts a narrow rally into a broad correction with nothing beneath until 729. SPY 755.58 is the twenty-day high, and a push through it on sub-50% participation is the single highest-quality short trigger available this week. On the long side, everything hinges on SMH 561.44: Friday's rejection level is now the bar semiconductors must close above, not merely touch, before any reversion thesis deserves capital.

13. Reversal Conditions Watch

SHORT Exhaustion Fade — the energy risk premium sun-260802-EXF-XLE-S
Why now: This is the thesis carried forward from 19 July, where the arm required both a genuine de-escalation headline and a failed high in energy. The headline leg arrived Saturday night. Crude had already fallen 5.5% last week before it, and the stated deal terms would reopen Hormuz and lift the naval blockade — removing the exact supply constraint that built the premium. Energy nonetheless closed Friday up 1.0%, leaving the sector priced for a war that may be ending.
Arm (both required): confirmation the reopening is actually proceeding rather than merely announced, and XLE breaking Friday's 58.32 low on a failed-high or reversal bar.
Kill: any re-escalation or public collapse of the deal, or XLE closing above the 20-day high of 60.45 with the strait still shut.
Trump has announced strike halts several times since February and each unravelled — which is precisely why this setup demands confirmation rather than anticipation. Illustrative expression: broad energy, producers and refiners.
LONG Oversold Mean-Reversion — semiconductors and memory sun-260802-OMR-SMH-L
Why now: The group has fallen 3.68% on the week and sits 12.7% under its fifty-day average after a flush to 503.63 on 7/29. The catalysts that could mark the low are dated and specific: AMD Tuesday and SanDisk Wednesday, the latter being the first pure-play memory result since that stock fell 53% from its high. Expectations have been reset hard enough that an in-line print can produce an outsized move.
Arm: SMH closes above Friday's 561.44 high and holds it, and only after Tuesday's close — the catalyst must be delivered, not anticipated.
Kill: SMH closes below 503.63, or AMD or SanDisk guide pricing or capacity down.
Honest caveat — this exact thesis already cost a kill two weeks ago. It armed on 7/21, paid 4.3%, then triggered its exit on 7/24 and the group fell a further 6.8%. Friday compounds the warning: a 3.5% gap up was sold all day to a 20%-of-range close. That is a rejection, not a base. The bar is therefore a close above the level, not a touch of it, and the window opens midweek at the earliest.
SHORT Failed Breakout with Breadth Divergence — the index sun-260802-FBD-SPY-S
Why now: The S&P sits 1.76% below its record on a hollow base — 43.1% of names advanced on Friday's up day, only 53.2% hold their fifty-day average, and equal-weight has flipped from leading to lagging. A push to new highs on that foundation is the textbook failed-breakout condition.
Arm: SPY trades above the 20-day high of 755.58 and closes back below it, with advancers still under 50% on the breakout day.
Kill: a genuine breadth thrust — more than 60% of liquid names above their fifty-day — or two consecutive closes above 755.58.
Strictly conditional: this is not a short into strength, it is a short of a failure, and without the failure there is no trade. The kill is deliberately a breadth measure rather than a price measure, because breadth is the thesis.
Lens Note what these three have in common: not one of them is live on Monday morning. The energy fade waits on confirmation that a repeatedly-broken promise is actually being kept, the semiconductor reversion waits on Tuesday's earnings and demands a close rather than a wick, and the index short does not exist at all unless a breakout first fails. That is the correct posture for a dispersion regime with a hollow base and a weekend headline still unpriced — the edge this week is in the waiting, and the most expensive mistake available is anticipating any of the three arms before its condition prints.

14. Synthesis & Week Reaction

The week in one read

The index is winning while the market loses, and this week's calendar decides which one is telling the truth. The S&P gained 1.10% last week and closed Friday near its high, but only 43% of stocks rose that day, small caps and equal-weight finished red, and barely half the market holds its fifty-day average. Underneath, capital has rotated violently out of the AI buildout — semiconductors down 3.68%, utilities down 4.19% — and into the three megacaps proving they can convert that spending into revenue. Implied correlation at 10.20 is the market's own admission that these are now two different markets wearing one ticker.

What should happen: a grinding, two-sided week with continued dispersion, the S&P chopping between its 744.99 fifty-day and the 755.58 twenty-day high, and the real action in single groups reacting to their own earnings. The damaged complex reports Monday through Wednesday, which front-loads the week's opportunity; Friday's payroll print, into a Fed that just drew three hike dissents, back-loads its risk.

Invalidation triple — the three things that would prove this read wrong:

  • Breadth thrusts above 60% of liquid names over their fifty-day. The hollow-rally thesis dies; buy the breakout instead of fading it.
  • SPY closes below 744.99 and then 737.68. The dispersion regime has resolved downward and the narrow rally becomes a broad correction, with 729.10 the next shelf.
  • High-yield spreads widen materially. Every cross-asset confirmation is currently absent; if credit joins, this stops being a rotation and becomes a risk-off, and every long thesis here is void.
Gap risk Monday's open is a gap event by construction. Friday's close contains none of the weekend, and a live Iran de-escalation headline landed Saturday evening — energy in particular will reprice before anyone can react to it. Position sizing set on Friday's prices is stale sizing; let Monday's first hour establish the actual level before committing, and treat any Friday-afternoon carry into this particular weekend as the known leak it is.
Lens Trade the pieces, not the index. This is a market paying for group selection and punishing broad directional conviction in either direction, with two dated catalysts — midweek semiconductor earnings and Friday's jobs report — that will do more to set the tape than any level on a chart. The highest-quality opportunities this week are a confirmed energy fade and a semiconductor reversion that must earn its arm with a close, not a wick; the lowest-quality is any index bet taken before 755.58 or 744.99 resolves.