S&P 500
7,656.98
+0.86% Fri
−0.80% on the week
A Fed meeting that is ~90% priced for a hike, and triple witching two days later. Between them, oil just took out its highs on a shut Saudi pipeline.
§1 — Where we start
The S&P rallied 0.86% on Friday and still finished the week down 0.80%. The Russell lost 2.41%. And the rally happened while the 2-year yield jumped 8 basis points to a fresh 52-week high — equities bought the oil relief while the bond market priced a Fed that is about to move.
That divergence did not survive the weekend. Oil has taken out Thursday's high, and the front end was right.
§2 — Session ladder
| Index | Sep 8 | Sep 9 | Sep 10 | Sep 11 | Week |
|---|---|---|---|---|---|
| S&P 500 | −0.58% | −0.48% | −0.58% | +0.86% | −0.80% |
| Dow | −1.18% | −0.77% | −0.60% | +0.98% | −1.57% |
| Nasdaq | −0.32% | −0.64% | −0.65% | +0.96% | −0.66% |
| Russell 2000 | −0.52% | −1.32% | −1.04% | +0.45% | −2.41% |
Four down days and one up day that recovered roughly half of it. The Russell's bounce was the weakest of the four — +0.45% against the S&P's +0.86% — which is the same message the session ladder gave all week: the rate-sensitive end of the market is where this is being expressed, and it did not participate in the relief.
The S&P closes the week 2.04% below its August all-time high of 7,816.70.
§3 — The fade
This is the single most important thing that happened last week, and it did not show up in any index number. Every major AI name opened strong on Friday and gave it back. Oracle gave back all of it and then some — on the best earnings report the AI-capex trade has produced this cycle.
Each bar is the round trip given back: small dot = session high, large dot = where it closed. The zero line is Thursday's close — a bar crossing it means the stock finished below where it started.
| Friday | Thu close | Session high | Fri close | Giveback |
|---|---|---|---|---|
| ORCL | $152.94 | $166.00 | $150.28 | −10.3 pp |
| AVGO | $360.83 | $366.78 | $361.99 | −1.3 pp |
| NVDA | $218.36 | $222.00 | $218.29 | −1.7 pp |
Oracle beat on revenue, EPS, cloud growth and backlog, raised full-year guidance, delivered 300,000+ GPUs, and told the market it needs no new financing. The stock traded as high as $166.00 and closed at $150.28 — below where it started the day before the print. When the best possible news produces a red candle, the news is not what is setting the price.
Treat that as the governing condition for this week. Good headlines are not being paid. Until the rate path resolves on Wednesday, the burden of proof sits with the buyer.
§4 — The weekend
On Friday, drones launched from Iraq's Maysan province struck Saudi Arabia's East–West crude pipeline, triggering fires and forcing a precautionary shutdown. No group has claimed responsibility; analysts point to Iran-aligned groups. The 1,200km line was moving roughly 4–5 million barrels per day — on the order of 4–5% of global supply.
The significance is not the volume alone, it is the routing. The Strait of Hormuz has been disrupted since March, and the East–West line is the principal way Saudi crude reaches the Red Sea without transiting Hormuz. Shutting it removes the workaround, not just a pipe. Separately, talks between Iran and Gulf states on a temporary Hormuz shipping corridor were postponed, with Saudi Arabia raising objections and Bahrain declining to take part.
Friday's equity rally was built on oil falling 3.6%. That is gone. Brent is now above Thursday's $105.37 print, which was itself the level that broke the tape last week. US diesel prices are at all-time highs, and diesel was already a quarter of August's entire PPI goods increase.
An energy shock the Fed can look through is a supply story that fades. One that keeps taking out highs while core CPI accelerates is a different problem — and it lands two days before a Fed decision that is already ~90% priced for a hike.
§5 — New variable
Nasdaq futures are down 1.2% against the S&P's 0.6%, and oil does not explain that gap. A separate story emerged over the weekend:
The read-through is being priced as a demand-timing risk for the compute complex: if the frontier labs slow their own model cadence, the capex curve that AVGO, NVDA, AMD and Micron are valued on gets pushed right. That is a thesis question, not a headline, and it will not resolve this week — but it gives the AI trade a second front while it is already fighting the rate path.
§6 — The main event
| Item | Detail |
|---|---|
| Decision | Wed Sept 16, 2:00 PM ET |
| Press conference | 2:30 PM ET · Chair Kevin Warsh |
| Also released | Summary of Economic Projections + dot plot |
| Current target | 3.50% – 3.75% |
| Hike priced | ~85–90% for 25bp → 3.75%–4.00% |
| A week ago | ~62% |
| Thursday | ~70% |
Core CPI at +0.3% against a +0.2% consensus is what did it. Economists moved with the market — EY-Parthenon's Greg Daco: "We are changing our Fed call from a hold to a 25bps hike at the FOMC meeting next week."
At ~90% priced, the decision itself is close to a non-event. The dot plot is the trade. A hike framed as one-and-done insurance against an oil shock is survivable; a dot plot showing a second and third move is a different regime, and it is the version the 2-year has been leaning toward all week.
Watch the 2-year at 2:00, not the index. It is at a 52-week high and has led every move in this sequence.
§7 — Calendar
Wednesday carries two events, not one. August retail sales at 8:30 is the first real read on whether $4.28 gasoline is eating discretionary spending — and it lands five and a half hours before the Fed decision. A weak retail number alongside a hike is the stagflationary combination nobody is positioned for.
Then Friday is triple witching. Quarterly expiry of index futures, index options and single-stock options. Volume roughly doubles, the daily range expands, and the final hour carries most of it. Positioning taken into Wednesday's decision has to be unwound or rolled by Friday, so the two events compound.
Expiry-week price action gravitates toward strikes carrying large open interest, so chart levels behave differently. The AVGO $400 call wall and the 377.5–381 zone act as pinning magnets rather than ordinary resistance. Treat any level read Monday through Thursday as provisional until Friday's expiry clears.
§8 — Levels
Last $361.99 (+0.32% Friday), after touching $366.78 intraday and failing. Q3 AI semiconductor revenue was +221% year over year to $16.7B; the chart has not cared.
Friday added a level: 366.78. The stock reached it on the best sector news available and could not hold — that makes it the near-term ceiling to beat before 369.74 even comes into play. Below, the floor is unchanged at 350.24 with $11.75 of room.
ORCL $150.28 — now below the pre-earnings close, 56% off its 52-week high of $345.72, and still falling after hours ($147.81). The gap between the fundamentals and the tape is the widest in the sector.
NVDA $218.29 — flat on the week, 52-week range $164.27–$236.54, now carrying the AI-safety story on top of the rate story.
§9 — Scenarios
The Fed moves 25bp and the SEP frames it as insurance against an energy shock, with no further hikes projected. The 2-year comes off its high, and the most rate-sensitive names — Russell, high-multiple AI — get the sharpest relief because they carry the most repricing.
Confirms: 2Y below 4.55% by Wednesday's close.
A hike, and a dot plot that keeps optionality without promising more. Nothing resolves, the 2-year stays near its high, and the tape chops into Friday's expiry with strikes doing more work than fundamentals. The most likely path, and the least tradeable.
Confirms: 2Y holds 4.60–4.70%, VIX stays under 22.
Dots showing a second and third move puts the 10-year decisively through 5.00% and tests 350.24 on AVGO. The mirror case — a hold against ~90% pricing — would spike equities initially, then raise the harder question of what the Fed sees that the market doesn't.
Confirms: 10Y > 5.00%, 2s10s under 25bp, VIX > 25.
If Brent keeps making highs, none of the dovish paths hold for long. Friday proved equities will rally on oil relief; the weekend proved that relief can vanish in 48 hours. Oil is still the independent variable — the Fed is reacting to it too.
§10 — The week, hour by hour
§11 — Scorecard
Thursday night's brief made four calls and one intraday claim. Two hit, one was right for the wrong reason, and one was wrong — including a call I published Friday morning that the day disproved by the close.
Badges: Open · Hit · Partial · Miss. Misses stay posted.
Triple witching is Friday Sept 18, the third Friday of the quarter-end month — not Sept 11, as a thread in the server had it. It is now in §7, where it belongs.
Free newsletter
Sunday's week-ahead map and Wednesday's midweek check. Free, no spam.