S&P 500
7,743.41
+0.51% Fri · +1.21% week
−0.94% vs ATH
Business activity at a five-year high. Consumer sentiment at a four-month low. The gap between them is the trade, and it shows up in the index spread.
PCE Wednesday · Payrolls Friday · Quarter-end§1 — The tape
Technology led the week at +3.52%. Utilities −3.87%, real estate −2.28% — both squeezed directly by the yield move — and energy −3.53% alongside crude. Microsoft added 4.53% on the week and Meta 12.90%.
§2 — The weekend
Friday's rally had a specific cause: reports that Iran had proposed reopening the Strait of Hormuz within seven days and restarting nuclear talks, conditional on the US lifting its naval blockade and oil sanctions. Crude fell hard — Brent under $98, WTI to $92.44, down 3.79% on the day.
Over the weekend the administration rejected the proposal, while signalling talks could resume. Brent is back at $106.02, having traded above $108 at one point this morning. Saudi-led forces also intercepted Houthi drones aimed at Riyadh and areas where Aramco operates.
Brent went $104.70 → under $98 → $106.02 between Thursday and this morning. Anyone who took Friday's break of $100 as a trend change got run over by a Sunday headline. This is what an unscheduled catalyst does — and it is exactly why the level, not the news, is the thing to trade.
§3 — The thing that got confirmed
Friday was the cleanest possible test of the call Edition 05 made. Crude fell nearly 4%, the Hormuz premium drained, and the 10-year went to 5.184% — its highest since July 6, 2007. The 30-year reached 5.504%.
If this were still an oil-driven inflation story, that could not happen. It happened because the driver has changed: composite PMI at a five-year high, core capital goods orders up 1.6%, weak demand at debt auctions, and heavy government borrowing. Rates are now a growth-and-supply story. Oil is noise around it.
Stop trading equities off the crude screen. For three weeks oil was the leading indicator for everything; as of last week it isn't. The chain now runs from activity data and auction results to yields to multiples. PCE Wednesday and payrolls Friday are the inputs that matter, and neither is an energy number.
§4 — The divergence
Large-cap tech and small caps have been separating since the first week of September. It is not noise any more — it is the single most persistent thing in this market.
| Week | Nasdaq | Russell | Spread |
|---|---|---|---|
| Sep 8–11 | −0.66% | −2.41% | +1.75 pp |
| Sep 14–18 | +0.72% | −0.93% | +1.65 pp |
| Sep 21–25 | +2.06% | −0.80% | +2.86 pp |
| Cumulative | +2.12% | −4.64% | +6.76 pp |
This is the two-economy trade. Small caps are domestic, cyclical, floating-rate borrowers — they are the purest expression of a 4% funds rate and a 5.18% ten-year. Mega-cap tech funds itself from cash flow and sells into a demand curve that does not check the mortgage rate. Same economy, opposite exposure, and the market is sorting them accordingly.
§5 — The crack
Composite PMI 58.4, a five-year high. Manufacturing 57.0, best in 52 months. Core capital goods orders +1.6%.
UMich sentiment 48.1, a four-month low, down from 51.7 — with respondents naming grocery and gas prices. Mortgage rates above 7%.
Those two readings describe different countries. The resolution matters more than either number: if business investment is running hot while households are this negative, either sentiment recovers or spending eventually follows sentiment down. Consumer confidence Tuesday and payrolls Friday are the first real evidence either way.
Reminder from the Sept 16 edition: August retail sales were +1.2% on the headline, but gasoline stations were +21% YoY. Nominal strength, thinner underneath.
§6 — Levels
Broke it Sept 15, reclaimed it Sept 18, tested it again Sept 24 and closed Friday at $352.81 (+0.70%). Friday's low was $349.43 — a marginal break intraday that did not hold. A level defended three times in eight sessions is real.
| Ticker | Fri close | Fri | From 52wk high |
|---|---|---|---|
| NVDA | $225.07 | +0.22% | −4.85% |
| AVGO | $352.81 | +0.70% | −28.73% |
| ORCL | $137.10 | −1.75% | −58.39% |
Oracle made another low. It is now 58% below its 52-week high and has fallen in every session since the force majeure notice on Project Jupiter. Nvidia is within 5% of its high. The complex has not re-converged, and nothing this week suggests it will.
Pre-market Monday: AVGO $352.98 · NVDA $225.07 · ORCL $138.28.
§7 — The week
After two weeks of nothing, four catalysts in four days. PCE Wednesday is the first inflation print the Fed sees since hiking — and the one that decides whether the 16-of-19 majority gets its second move. It lands on quarter-end, which adds rebalancing flow to an already busy session.
Payrolls Friday is the bigger number. Given the split in §5, the employment report is the tiebreaker between the strong-business and weak-consumer readings — and with a Fed that has explicitly said financial conditions are still too easy, a hot print is not good news for multiples.
Micron Wednesday is the AI-capex read of the week. After Oracle's force majeure, memory pricing and datacenter commentary carry more weight than usual.
§8 — Scenarios
Edition 05 committed to reserving one branch for a driver that is not currently in the narrative. That's Scenario D, and it stays from now on.
Inflation cools and hiring moderates. Yields finally come off 5.18%, the curve steepens from a very flat base, and the four-week rotation reverses hard — the Russell is where the value is if rates relent, precisely because it has been punished most.
Confirms: 10Y < 5.05%, Russell outperforms Nasdaq on the week.
PCE firm and payrolls solid confirm what the PMI already said. Yields stay above 5.15%, a second hike gets priced further, and the tape keeps doing what it has done for three weeks: large-cap tech bid, small caps and rate-sensitives sold.
Confirms: 10Y > 5.15%, Nasdaq–Russell spread widens a fourth week.
The worst combination and the one the §5 split points at: PCE firm while payrolls disappoint. That removes the growth justification for high yields without removing the inflation problem. Everything sells, and the Fed has no good answer.
Confirms: 10Y > 5.15% with payrolls under consensus; VIX > 22.
Candidates this week: credit — a 5.5% thirty-year and AI capex funded with debt is a spread-widening setup nobody is watching. Quarter-end flows — Wednesday's rebalance out of a quarter this lopsided could be violent in both directions. Micron — a bad AI-demand read would hit harder than any macro print. Or Hormuz resolving for real.
Watch: IG/HY spreads, the 30-year, and Wednesday's closing auction.
§9 — Scorecard
Badges: Open · Hit · Partial · Miss. Misses stay posted. Running total across six editions: 11 hits, 5 partials, 5 misses.
Free newsletter
Sunday's week-ahead map and Wednesday's midweek check. Free, no spam.