The Friday Wrap — September 4, 2026: a payroll blowout, and the chips went up

The August employment report landed Friday morning at 162,000 jobs, against a Bloomberg-compiled consensus near 55,000. The unemployment rate held at 4.1%. July's reported loss of 23,000 was revised to a gain of 21,000, which quietly erased the summer's soft patch, and June came up too. Per MT Newswires, bars and restaurants led the hiring; information-sector employment fell.

The rates market moved immediately. CME FedWatch odds of a hike at the September 16 decision went from 49% Thursday to 58% by Friday's close, per MT Newswires. Intraday prints ranged wider than that. Benzinga had the same number near 52% at midday and about 60% by the afternoon, so treat the level as a moving read and the direction as the fact. The two-year Treasury yield rose 4.3 basis points to 4.38%, its highest since January 2025. The ten-year added 2.2 basis points to 4.78%.

Here is what the index did with all of that. SPY closed at 770.19, down 0.39% on the day and up 0.11% on the week. That is the whole reaction at the index level.

Under it, the rotation was not quiet at all.

What actually got sold

The Friday gainers were memory, storage and semiconductor equipment. SanDisk rose 11.89% on our SIP bars, from 1,554.99 to 1,739.81, the biggest gain in the S&P 500 on the day; Benzinga attributed it to an accelerating NAND pricing cycle. The semiconductor ETF SOXX rose 3.52% while SPY fell. Micron, Western Digital, Applied Materials and Intel all advanced. Astera Labs led the Russell 1000 at roughly +12%, and KLA added 8.15% after Susquehanna lifted its estimates on wafer-fab equipment spending.

The Friday losers were software and discretionary retail, and every one of them fell on something specific. Guidewire dropped about 21% after a fiscal fourth quarter that beat on revenue; Benzinga's read was that a price-to-earnings multiple above 100 left no room for anything short of a clean subscription beat. UiPath fell 16.66% after growing revenue 13% to $410.3 million, beating consensus by roughly $12.5 million, and logging a fourth straight quarter of GAAP profitability. The problem was the guide. Third-quarter revenue of $440 to $445 million implies about 8% growth, roughly half the pace just delivered. Fair Isaac lost 16.55% after the FHFA director instructed Fannie Mae and Freddie Mac to accept VantageScore-underwritten mortgages, which ends a scoring monopoly by regulatory letter. Mortgage originations are about 71% of the company's B2B Scores revenue. TransUnion and Equifax fell with it.

Lululemon was the worst name in the S&P 500, down 17.48%. Second-quarter revenue fell 4% to $2.4 billion and comparable sales dropped 10%. Reported EPS of $2.92 cleared estimates, and that is the number worth pausing on: $134.5 million in tariff refunds contributed 86 cents of it. Full-year revenue guidance came down to $10.35 to $10.5 billion against $11.03 billion of consensus. The interim CEO pointed to negative social commentary and a sharper-than-expected slowdown in core categories.

Read the two lists together and the day is legible. What got sold was multiple. What held was order book.

Broadcom beat, and the tape sold it anyway

Broadcom reported Wednesday after the close. Adjusted EPS came in at $3.32 against $3.24 expected, revenue at $29.59 billion against $29.36 billion, and semiconductor revenue more than tripled to $16.7 billion versus a $15.2 billion average estimate. Per CNBC, the company pointed to expanding business with Anthropic and OpenAI. Fourth-quarter revenue guidance of $34.8 billion came in under the $35.03 billion consensus.

Our own bars are the ones that settle what happened next. AVGO closed at 367.24 on Wednesday, opened Thursday at 351.74, and closed Thursday at 357.16. That is a 2.75% Thursday decline on a quarter that beat on both lines and tripled its largest segment. Friday it added 0.20% to 357.87. Name the session when you quote a reaction, because the after-hours print and the regular session were telling different stories all week.

Dell went the other way on the same logic. Fiscal second-quarter revenue was a record $47 billion, up 58% year over year against $44.92 billion of consensus, with adjusted EPS of $7.04 up 203%. The company booked $60.9 billion in AI server orders in the quarter and $131.7 billion over twelve months. It finished the week up 15.46% per Benzinga.

FactSet puts blended second-quarter S&P 500 earnings growth at 50.4% year over year, a seventh straight quarter of double-digit growth, with ten of eleven sectors growing.

Energy, and a barrel that will not settle

Crude had the biggest week of any asset we track. WTI traded past $91 Friday afternoon and Brent near $96, with weekly gains of 9.3% and 8.8% respectively per MT Newswires. Our USO proxy, which is a front-month futures vehicle and not the barrel, rose 9.45% on the week and sits 32.07% above its 200-day average.

The supply picture behind that is unresolved. Hormuz throughput has collapsed from 21.6 million barrels a day in the fourth quarter of 2025 to 4.9 million in the second quarter of 2026, a 77% reduction. Reuters counted six commodity vessels transiting the strait Wednesday, against a ten-day average near thirteen. OPEC+ has completed the rollback of its 1.65 million barrel-a-day 2023 cut and is on paper producing at its 2022 ceiling, but traders tracking secondary-source data put actual supply 300,000 to 500,000 barrels a day short of that. Ministers meet in Vienna Sunday.

One consequence showed up in the inflation channel Friday: US diesel prices reached an all-time high, per MT Newswires, on the same day the labour data pushed hike odds up.

Two of our proxies disagree with the standard telling. Coverage had the dollar firming on the jobs print, and our UUP proxy did rise 0.25% Friday, but it finished the week down 0.35%. Energy equities also did not follow the barrel on the day. XLE fell 0.85% Friday while crude held.

The frame

SPY sits above every average we watch. It is 1.56% over its 10-week SMA, 1.04% over its 8-week EMA, 1.76% over its 50-day and 8.5% over its 200-day. The weekly MACD histogram is positive at +0.46. Friday printed a harami, a body entirely inside Thursday's, closing below its open. The traded range was 769.00 to 772.87, against an ATR14 of 4.49, so the session was narrower than an average day. The 20-day sits at 769.05 and the 50-day at 756.86, and that spread narrowed from 1.73% to 1.61% this week. Price repaired faster than the daily momentum did: the daily MACD histogram is still negative at −0.68 with RSI at 55.6.

QQQ closed 718.96, up 0.35% on the week, above its 10-week and 8-week lines. Its weekly MACD histogram is still negative at −2.85, which is the same story a step behind.

The honest divergence is IWM. It closed 296.01 and it printed the best candle of the three, engulfing Thursday's body and closing above its open on a day the S&P fell. It is also the only one of the three below its 20-day, its 50-day, its 10-week and its 8-week, all at once. Daily RSI is 48.2. Small caps had the strongest single session and the weakest structure underneath it, and both things are true at the same time.

Then zoom out, because a flat week is not a year. Quarter to date, SPY is +3.14% while QQQ is −2.37%, a 5.51 point gap on the index a reader happened to own. Year to date the order inverts completely: IWM +20.75%, QQQ +17.31%, SPY +13.55%. The weakest structure on the board is also the year's leader.

Open interest, from the OCC book as of September 2 and therefore not Friday's positioning: Friday's expiry carried its largest call line at 778 with 16,100 contracts and its largest put line at 765 with 55,453. The September 8 expiry is far thinner, with 6,447 calls at 770 and 1,781 puts at 736. This is open interest and contract counts. We do not carry greeks, so we will not call it gamma.

The board

Coil closed the week in day mode. The SPX book reads names-on and agile with SPY and six of eleven sectors open and 282 names fully qualified; the QQQ book reads names-on with three of six sectors open and 40 names qualified. The macro book reads risk-off and defensive on broken index structure, which sits oddly next to an S&P above all four of its averages, and we will show you that disagreement instead of hiding it. The crypto book has BTC above its 200-day gate by 16.8% with both sleeves in, with 41 of 81 names on the RH-tradable panel above their own 200-day. Benzinga had Bitcoin below $80,000 and down 1.8% Friday. A 200-day gate is not a daily read, and it is not supposed to be.

What we do not carry: dark-pool prints, closing-auction imbalance, spot VIX and true dealer greeks. Our VIXY line is a futures ETF and 34.67% below its 200-day, which is a fact about that fund, not about spot volatility.

What this sets up

The week re-priced policy without re-pricing growth, and the calendar decides whether that holds. US markets are closed Monday for Labor Day, so next week is four sessions. The August CPI release is scheduled for Friday, September 11 at 8:30 ET per the BLS schedule, four days before the September 15–16 FOMC meeting, and the Fed's quiet period runs from September 5 through September 17. Our own audited econ list carries CPI a day earlier, on the 10th, and two outlets published a third version; when they disagree we use the agency's own calendar and tell you that we did.

The levels are where the argument gets tested. SPY's 20-day at 769.05 sits half a point under Friday's close and its 50-day at 756.86 is 1.76% below, so the 20-day is the first line a CPI print can actually reach. Below the 20-day, the 20-day traded range bottoms at 759.48 and the 60-day at 716.58. Above, the 20-day and 60-day highs are the same number, 779.37, which is where the last three weeks stopped. For IWM the question is simpler and it has one number: 298.87, its 20-day, which it has been under all week while leading the year. Whether those levels give or hold is not ours to say. Where they are is.

---

Read the board at coil.trade/scanner · own the engine, $79 one-time, at coil.trade/buy/engine.

The Friday Wrap is impersonal market research. It is not investment advice, it is not a recommendation to buy or sell any security, and it does not take your circumstances into account.

Sources — prices, ranges, moving averages and weekly frames are our own Alpaca SIP bars through Friday's close. Open interest is OCC, as of September 2, 2026. Macro and company reporting: MT Newswires, Benzinga, CNBC, Kiplinger, FactSet via Benzinga, Reuters via Eastern Herald, and the BLS release schedule.