Friday, October 2, 2026, written at 9:30 a.m. Eastern, half an hour after the payrolls release and half an hour before the open. The September employment report and the eurozone flash estimate are both in hand. Futures and pre-market levels are indicative and can move before the bell. Mainland China and Hong Kong remain shut through October 7.
Top Call: The Labour Number Is Weaker Than It Looks in Composition and Stronger Than It Looks in Level, and the Inflation Side of the Week Is Being Ignored Entirely
Nonfarm payrolls added 29,000 jobs in September against a consensus near 90,000, July was revised down by 31,000 and August by 29,000 for a combined 60,000, and the twelve month average now sits at 45,000. Within four minutes of the release the two year fell seven basis points to 4.716%, the ten year six to 5.176%, and October hike odds went from roughly 64% to somewhere between 12% and 16%. Equity futures turned up across all four contracts with the Russell the strongest at 1.5%.
The reflex read is that the labour market is cracking and the Federal Reserve is finished. Both halves of that deserve more care than they are getting.
Start with the level. The unemployment rate was 4.2%, inside the 4.1% to 4.3% band it has occupied since March, and labour force participation rose to 61.8% from 61.6%. A labour market that adds 29,000 jobs, absorbs a rise in participation and still does not move the unemployment rate is not a market shedding workers. It is a market in which the number of jobs required to hold the rate flat has collapsed. Apollo’s Torsten Slok put breakeven job growth at close to zero against something near 200,000 two years ago, and if that is even approximately right, 29,000 is a positive number against replacement rather than a negative one. Immigration policy did most of that arithmetic and none of it is visible in the headline.
Now the composition, which is genuinely bad and is the part worth holding. Health care added 17,000 against a twelve month average of 33,000, so the one sector that has carried this expansion is decelerating by half. Financial activities lost another 7,000 and is now down 129,000 from its May 2025 peak, concentrated in insurance carriers. Construction added 11,000 and manufacturing 9,000, which are rounding errors. The U-6 underemployment rate is 7.6% and the long term unemployed are 27.1% of the total at 1.9 million people. That last figure is the tell. Firms are not firing, which is why the unemployment rate holds, and they are not hiring either, which is why anyone who does lose a job stays out for a long time. Low hire and low fire is a stable arrangement right up until something forces one side of it.
Then there is the half of the week nobody repriced. On Thursday the ISM prices paid index jumped 6.8 points to 77.9, its highest reading since the Iran war began, with 58.6% of respondents reporting higher prices against 46.2% in August and not a single commodity category reporting a decrease. This morning eurozone flash inflation printed 3.8% against 3.6% expected and 3.2% in August, the highest since 2023, with energy at 18.8% from 14.3% and core ticking up to 2.5%. And American average hourly earnings rose five cents, or 0.1%, to $37.81, which is 3.0% year over year against a headline CPI running at 3.4%.
Put those three next to each other. Input costs at the factory gate are accelerating at the fastest pace of the conflict, imported energy inflation is accelerating across the Atlantic, and the price of labour is now growing more slowly than the price of goods. That is the ordinary definition of a stagflationary pinch, and it lands on households rather than on corporate margins, because the companies have pricing power in the ISM survey and the workers do not have it in the earnings line. The equity market read the same morning as unambiguously good news and added 1.5% to the Nasdaq contract. It is good news for discount rates. It is not good news for anything else.
| Metric | Consensus | Actual | Read |
|---|---|---|---|
| Nonfarm payrolls | About 90,000 | 29,000 | Miss; 12-month average now 45,000 |
| Prior revisions | Not forecast | -60,000 | July -31,000, August -29,000 |
| Unemployment rate | 4.1% | 4.2% | Inside the 4.1-4.3% band since March |
| Participation | 61.6% | 61.8% | Rose, which is why the rate held |
| Avg hourly earnings | +0.3% / 3.2% | +0.1% / 3.0% | Real wages negative against 3.4% CPI |
| Long-term unemployed | Not forecast | 27.1% of total | Low hire, low fire, and slow to recover |
Rates: Four Prices for the Same Meeting in Five Sessions
One week ago the October meeting was priced better than 70% for a hike. On Wednesday a core PCE print that this desk showed was a statistical revision rather than disinflation took it to roughly 35%. On Thursday ISM prices paid at 77.9 pushed it back to about 64%. This morning payrolls took it to between 12% and 16%.
That is four materially different probabilities on one meeting in five sessions, derived from three data points, two of which have serious interpretation problems. The PCE surprise existed only because the consensus was built against a July figure the Bureau of Economic Analysis had already redefined. The payrolls miss is measured against a consensus that assumes a breakeven rate that no longer exists. Only the ISM print was a clean read, and it is the one the market has spent the least time on, because it pointed the wrong way for the trade everyone wanted.
We are not arguing the Fed hikes in October. Williams and Jefferson have both effectively ruled it out, and with 29,000 on the board they will not be contradicted. We are arguing that a market moving fifty points of probability in either direction on consecutive mornings is not processing information, it is chasing the most recent headline, and that this is how positioning gets built in the wrong place.
The curve itself is more informative than the probabilities. The two year fell seven basis points and the thirty year only four, so two year to ten year widens again toward 46 basis points. The ten year is now 5.176% against the 5.3338% it printed on Wednesday morning, a sixteen basis point rally in two sessions off a twenty four year high. But the thirty year at 5.569% is still a July 2002 level. The front end is pricing a growth scare. The long end has barely participated, because term premium, the deficit and the supply calendar do not care what payrolls did in September. A steepening driven by the front end collapsing while the back end holds is the market saying the Fed will stop early and the inflation will outlive the pause. That is not a dovish curve. It is a capitulating one.
Gold and bitcoin both confirm it. Gold is up about 1.1% near $4,250 and bitcoin 3.3% to roughly $86,675, with the dollar softer at 101.79. Nominal yields fell with no obvious fall in breakevens, so real yields came down and both carry sensitive assets rallied together. The framework we marked mixed yesterday works cleanly today.
Energy: Crude Is Loosening While Product Is Not, and Only One of Those Is in the Headline
WTI fell 4.22% to $88.95 on the November contract, through $90 for the first time in weeks, with Brent near $99.68 and the spread back around $11. On the week WTI is down roughly 7%.
Three things are doing this and none of them is a ceasefire. A large American inventory build, downgrades to 2026 demand forecasts from both OPEC and the IEA, and reports of partial recovery in Persian Gulf flows. Set against that, Washington is still floating a diesel export ban and freight rates are elevated, which depresses the landed value of American barrels specifically. That is why WTI is doing the heavy lifting on the downside rather than Brent.
The important point is the one that keeps not showing up in crude. The diesel crack has been above $100 a barrel since August and set a record above $106 in early September, and distillate inventories are the lowest on record for the season at 14% below the five year average. Crude is loosening. Refined product is not. A market reading $88 WTI as energy disinflation is reading the wrong barrel, and the ISM prices paid print on Thursday, driven in part by petroleum based products, is the evidence that the pass through runs through the product crack and not the crude screen.
On the diplomatic track there is nothing new. Day two hundred and sixteen of the war finds Araghchi in the United States with Qatari mediation continuing on an updated version of the seven day roadmap Trump already rejected, and Tehran repeating that Hormuz reopens only on its conditions. OPEC+ has left October quotas unchanged and has very little physical lever to pull while the strait is constrained.
Europe: The Highest Inflation Since 2023 and the Equity Market Went Up Anyway
Eurozone flash inflation came in at 3.8% year over year against 3.6% expected and 3.2% in August, with the monthly rate accelerating to 0.6% from 0.4%. Energy at 18.8% supplies essentially all of the acceleration. Services are 3.2%, core 2.5%, food 1.4% and non energy industrial goods 1.1%.
The divergence this desk has run for three sessions is now at its widest. American inflation was revised down on Wednesday by a change in how three service categories are measured. European inflation has risen six tenths in a month because the continent imports every barrel it burns and has no domestic offset. One of those is a definitional event and one is a physical one, and only the physical one shows up in what anyone actually pays.
European equities rose anyway, with the Stoxx 600 up 0.86% to 632.06, the DAX up 0.95% near 25,180 and the CAC up 0.73% to 7,892. They are trading the American payroll number and the global discount rate rather than their own price level, which tells you the ECB is still not regarded as a live threat. If a 3.8% print with core rising does not change that, the question is what would, and the answer is probably the October 16 confirmation with the full detail attached.
Asia: Tokyo Gives It Back, Seoul Reclaims 7,000
The Nikkei fell about 0.9% to roughly 68,310, surrendering part of Thursday’s 3.3% memory driven surge. The simplest explanation is the better one. Thursday’s move was a single day repricing of the semiconductor supply chain on the Micron guide, Micron itself then closed up only 3.03% rather than the 8% it was indicated, and Tokyo marked itself back toward where the cash market actually settled.
Seoul went the other way, with the Kospi up 32.4 points, or 0.5%, to 7,003.7, back above 7,000 for the first time since before Chuseok, and foreign investors sold for a sixth consecutive session while doing it. That is the pattern we flagged yesterday holding for another day. The domestic bid is absorbing persistent foreign selling in the two names responsible for a record $60.3bn September semiconductor export month. It remains a currency and rate differential story rather than a memory story, and a softer dollar this morning is exactly the condition under which it resolves.
Corporate: Nike Guides Below Every Estimate on the Street and Announces Layoffs for 2027
Nike reported after Thursday’s close with adjusted earnings of $0.48 against $0.44 expected and revenue of $11.21bn against $11.35bn, and the shares fell 6.24% to $32.96 after hours from a close of $35.15 that was already near a twelve year low. Gross margin improved to 42.8%. North America returned to 2% growth. Greater China fell 26% in a region worth about 15% of annual revenue, a ninth consecutive quarterly decline.
The quarter is not the story. The full year guide is revenue down a high single digit percentage and earnings of $1.15 to $1.35 against a $1.61 consensus, which is below every published estimate rather than below the average of them. Alongside it the company set out a restructuring called Pace running through 2031, targeting $2.5bn of savings over four years including $1bn of employee related costs, reorganising into Americas, APGC and EMEA with Asia Pacific and Greater China leadership moving to Singapore, and with workforce decisions beginning in calendar 2027.
Read that against this morning’s payroll report. The employment number describes what happened in September. An 8-K from the largest footwear company in the world, announcing a four year cost programme with a billion dollars of labour in it and a start date fifteen months out, describes what is being planned. The household side of this economy is not weakening in the hiring data yet because the decisions that will weaken it have not been executed. They have been disclosed.
| Metric | Consensus | Actual | Read |
|---|---|---|---|
| Revenue | $11.35bn | $11.21bn | Miss; China -26%, a ninth down quarter |
| Adjusted EPS | $0.44 | $0.48 | Beat on margin and SG&A, not on demand |
| Gross margin | Below 42% | 42.8% | Supply chain actions working |
| FY27 EPS guide | $1.61 | $1.15 to $1.35 | Below every estimate on the street |
Micron closed at $1,097.39, up 3.03%, after being indicated roughly 8% higher pre-market. Rosenblatt went to $1,900 from $1,500, DA Davidson to $2,100 and Mizuho to $1,400, and the average twelve month target sits near $1,520 across 49 analysts rated Strong Buy. A 3% close on that set of numbers and those revisions says the information was already in the price, which is the clearest evidence yet that memory has moved from a contrarian position to a consensus one. TrendForce on September 30 put conventional DRAM contract prices up another 10% to 15% in the fourth quarter and NAND up 15% to 20%. The physical shortage is real and getting tighter, and the easy money in expressing it is behind us.
Alphabet launched Gemini 4 Argon, its first frontier model since February, and the shares rose about 2% on Thursday. The benchmark claims are strong, with a reported 77.9% on DeepSWE v1.1, 68.9% on the Vals Index and a tie for the top cybersecurity score, and the model can emit up to a million output tokens in one trajectory against 64,000 previously. The commercially interesting line is the price. Introductory rates are $2 per million input tokens and $10 per million output, with cached input discounted 95%, and Google has disclosed in advance that those rates eventually double to $4 and $20. A frontier lab publishing the step-up at launch is telling you the launch price is a customer acquisition cost rather than a cost of service, and a million token trajectory is a very large amount of compute to sell at a promotional rate. That is the capacity thesis appearing in a price list.
onsemi restructured its acquisition of Synaptics into an all-cash deal at $123 a share for roughly $5.7bn, replacing the all-stock agreement signed in June that valued the target near $7bn, after an unsolicited competing proposal. A buyer switching from paper to cash while cutting the headline value by $1.3bn is making two statements at once: that it regards its own equity as the more valuable currency to retain, and that the target’s board preferred certainty to participation. In the hottest semiconductor tape in two decades, an analog and IoT asset is being bought at a discount to a price agreed three months ago. The complex is not one trade.
Rocket Lab signed a twenty mission Electron agreement with Tokyo’s Synspective running from 2028 to 2031, its largest commercial launch contract to date, taking the Electron backlog past 100 launches and total contracted backlog above $2.3bn, plus a dedicated Neutron mission for Kepler Communications ahead of the rocket’s expected fourth quarter debut. Accenture held most of its post-earnings move at about $214. In healthcare, Ultragenyx won the first FDA approval in Sanfilippo syndrome type A and Eli Lilly reported that oral orforglipron met its non-inferiority endpoint on major adverse cardiovascular events against titrated insulin glargine in ACHIEVE-4.
Key Events Today and Ahead
| When | What | Why It Matters |
|---|---|---|
| Fri, done | Payrolls 29,000, U3 4.2% | 60,000 of revisions; October effectively off |
| Fri, done | Eurozone CPI 3.8% | Energy 18.8%; highest since 2023 |
| Thu, done | ISM prices paid 77.9 | +6.8 points, the week’s cleanest signal |
| Mon | ISM services | The prices component is the one to read |
| Oct 8 | China and Hong Kong return | Absent for this entire data week |
| Oct 16 | Eurozone final CPI | Confirms or revises this morning’s 3.8% |
| Oct 27-28 | FOMC | Priced 12% to 16%; four prices in five sessions |
The ISM services prices index on Monday is the line that matters next. Manufacturing told us on Thursday that goods inflation is accelerating at the factory gate. If services prices follow it higher into a labour market adding 29,000 jobs a month, then the committee is being asked to hold policy into rising inflation and falling employment at the same time, which is the one configuration it has no good answer for.
Accountability
We flagged Micron as indicated up roughly 8% pre-market and it closed up 3.03%. The call on the quarter was right and the call on the reaction was not, and the gap is the more useful thing. We should have asked why a disclosed, contracted, sold-out business with a $10bn increase in customer commitments was worth only three points on the day, and the answer we now favour is that the memory trade is consensus. That is a meaningful amendment to a thesis we took a victory lap on yesterday.
The anchoring warning we wrote yesterday applies to us again this morning, in reverse. We spent the week arguing that the market was overreacting to a PCE revision by cutting hike odds, and this morning a payroll miss cut them much further. It would be easy to describe that as vindication. It is not. The direction we were sceptical of has now been confirmed twice by prints we did not predict, and the honest position is that our objection was always to the quality of the reasoning rather than to the destination. We are keeping the objection and dropping any claim to have called the path.
The gold and bitcoin carry framework passes cleanly. Yesterday it was mixed and we declined to claim it. Today nominal yields fell across the curve, the dollar softened, and gold rose 1.1% with bitcoin up 3.3%. Co-movement against the real rate held. We restore it to full confidence and will say so again when it fails.
The distillate thesis is correct on mechanism for a fourth week and is finally showing up where we said it would. We have argued the shortage lives in refined product rather than crude. WTI fell 4.22% this morning while the diesel crack stays above $100 and distillate inventories sit at a record seasonal low, and Thursday’s ISM prices paid print cited petroleum based products among its drivers. After three weeks of naming the wrong trigger, the structure is now visible in a macro survey rather than only in a crack spread.
The crude discipline rule pays for a seventh week. WTI fell more than four percent on inventories and demand forecasts in a week that produced no diplomatic movement at all. Carrying no directional view on the barrel remains the correct posture.
The breadth question gets its counterexample. We closed September saying the index was one trade and the Russell was the casualty. This morning the Russell contract is the strongest of the four at 1.5%. One session is not a reversal, and a rate driven bounce in the most rate sensitive index is exactly what should happen when the front end falls seven basis points. But we said breadth was bad and structural, and we owe it to the record to note the first day it behaved otherwise.