Payrolls missed by 61,000, October was priced out, and the long end sold off anyway

Friday 2 October — the payroll print, and the bond market's answer to it

The BLS released the September Employment Situation at 08:30 ET. Non-farm payrolls +29,000 against a +90,000 consensus; private +46,000, implying government −17,000. July was revised from +21,000 to −10,000 and August from +162,000 to +133,000, a combined −60,000. The unemployment rate rose to 4.2%, U-6 fell to 7.6%, average hourly earnings rose 5 cents to $37.81, +0.1% m/m and +3.0% y/y, and participation was 61.8%. The industry detail is the soft part: temporary help −10,900, professional and business services −9,000, information −10,000, financial activities −7,000, with the gain almost entirely health care (+16,700) and construction (+11,000).

The household survey complicates the recession reading and is the reason this desk is not treating 4.2% as a growth signal: household employment rose 406,000 while the labour force rose 485,000. The rate went up because supply grew faster than hiring, not because employment fell. Set beside the Cleveland Fed's nowcast — core PCE 3.02% for September rising to 3.10% for October, headline PCE 3.56% to 3.66% — this is a soft-labour, sticky-inflation print rather than a disinflationary one.

The reaction was in two parts and the second part is what matters. Policy-path pricing repriced dovish on contact: the October hike went from 36–39% to 17–23% across three independent routes, and Reuters reported rate futures moving from "more than one-in-four" to "less than one-in-five". Yields initially fell. They then reversed completely, and the official par curve closed higher at every tenor: 2y 4.83% (+5bp), 3y 4.96% (+5bp), 5y 5.06% (+5bp), 7y 5.17% (+5bp), 10y 5.28% (+4bp), 30y 5.63% (+2bp). A secondary source reported 2 October "closes" of 2y 4.74 / 5y 4.94 / 10y 5.18 / 30y 5.57, down 5–7bp, which flatly contradicts the primary; those figures are rejected as closes and are plausibly the post-print intraday lows, which would imply a round trip of roughly ten basis points in the ten-year on the day. No Fed official spoke on 2, 3 or 4 October — verified by absence from the Fed's own 2026 speeches index, whose latest entry is 1 October — so the Committee's reaction function to +29,000 is genuinely unknown going into Wednesday's minutes.

Equities took the dovish half and ignored the rest. S&P 500 7,722.72 (+56.27, +0.73%), Nasdaq Composite 27,190.86 (+1.19%), Dow 51,176.96 (+0.49%), Russell 2000 2,832.90 (+0.94%), VIX down 6.6% to 15.31 after spiking to 17.59 intraday on Thursday as the pre-payrolls hedging bid. Ten of eleven sectors rose, consumer discretionary and technology leading and health care the only decliner. Movers: Tesla +4.65% on Q3 deliveries of 486,532 against roughly 461,000 expected; Rivian a record 19,248; Broadcom +3.35% on an AI chip funding report with Nvidia reinstated a Morgan Stanley top pick; against that Seagate and Western Digital −10.2% each on Toshiba HDD capacity expansion, and Nike −3.64% on a 4% fiscal Q1 revenue fall. The Composite's +0.5% week against the Dow's −1.3% is a 180bp growth-over-value spread in five sessions, and the AI complex carried the index single-handedly.

Europe: a relief rally that did not repair anything

Every one of the eight European indices closed higher on Friday and every one still finished the week down. Stoxx 600 631.35 (+0.75%, week −1.14%), DAX 25,231.20 (+1.17%, week −0.7%), CAC 40 7,897.19 (+0.79%, week −2.24%), FTSE 100 10,461.95 (+0.32%, week −2.18%), IBEX 19,085.30 (+0.42%), FTSE MIB 50,483.21 (+0.49%), SMI 13,660.92 (+0.28%), Euro Stoxx 50 6,238.50 (+1.02%). Each level was reconstructed as level minus change against the verified 1 October close and every one tied. The move was a pure semis squeeze — Infineon +8.78%, BE Semiconductor +7.00%, STMicroelectronics +6.54% — with banks lagging and on course for their worst week since April. The driver was American, not European: the payroll print broke the Fed-hike bid in the long end for part of a session. Against it, the euro-area September flash HICP came in at 3.8% y/y, up from 3.2% and above a 3.6% consensus, the highest since September 2023, with energy +18.8% and core 2.5% — which is why banks could not rally with the index and why December ECB hike pricing sits near 82%.

The week's event was the gilt long end, and it needs restating correctly: the 30-year printed 6.029–6.04% intraday on Thursday 1 October, the first time above 6% since 1998, and closed 5.94%; Friday closed 5.91%. The 10-year's high was 5.44% on Monday 28 September, a 19-year peak, not 5.48% on 1 October. No DMO, Treasury, Chancellor or Bank of England comment on the breach could be found across two sessions — silence through a 28-year yield high is itself the datum. France cleared €11.999bn of OATs on 1 October in which the on-the-run 2036 benchmark drew a bid-to-cover of 2.00 against 2.48–3.02 on the three off-the-runs, the cleanest available evidence that the problem is French-specific rather than a general duration problem.

Asia on Friday: one theme, two signs

The session split cleanly along the rates axis. Hong Kong broke on reopening: Hang Seng 23,972.29, −2.60%, with the open as the high — financials the drag, HSBC −5.7%, Standard Chartered more than 5% lower, AIA more than 4% — and HS Tech −2.26% to 4,157.94. New Zealand −0.94% and Japan's TOPIX −0.99% went the same way. The markets that rose did so on silicon: Australia +0.79% with IT +4.45%, Korea +0.45% to 7,003.74 on bargain-hunting in major tech, Taiwan +0.25% to a record close of 48,475.74. Long duration got hurt; long silicon got paid.

Japan deserves its own line, because the week's rally was an index artefact. Over the week to Friday the Nikkei rose 2.93% while the TOPIX fell 0.91% — a 384bp divergence with opposite signs. On 1 October alone the Nikkei rose 3.30% against the TOPIX's 0.57%, a 273bp single-day spread. The average Japanese company had a down week; the price-weighted index had a melt-up on a handful of high-nominal chip and SoftBank names. Friday was broad by contrast (Nikkei −0.94%, TOPIX −0.99%), which is what makes the melt-up day the suspect one.

The weekend, 3–4 October

OPEC+ held. The core seven kept November required production at September levels — a joint 31.0m bpd, Saudi Arabia 10.478m, Russia 9.949m, Iraq 4.431m — with the 1.65m bpd voluntary tranche fully unwound in September and the 2.0m bpd additional cuts maintained through year-end. Next meeting 1 November. Compensation cuts were not addressed in any reachable account, which is not the same as unchanged, and the communiqué carried no Hormuz language at all. The official text could not be read: opec.org returned HTTP 402 through this session's egress, so the decision is wire-sourced.

The G7 release is the bigger number. Announced Friday: up to 100m barrels of crude and diesel from G7 and partner stocks over four months, coordinated through the IEA, with a "substantial" diesel release inside 20 days and the crude/diesel split undisclosed. Macron framed it as "a clear signal to the markets". The market answered immediately — European diesel futures slumped, diesel's premium over crude fell from $76.77 to $69/bbl and European refining margins fell about 10%. Two qualifiers belong with it: many barrels pledged under an earlier 400m-barrel release in March have still not reached markets, and the US contribution is arithmetically capped — the SPR holds 283.8m barrels at 30 September against a 252.4m non-emergency floor, so 31.4m of lawful headroom against the 40m being offered, which is precisely why the 6 October tranche is an exchange and not a sale.

Iran declared the Strait of Hormuz closed on 4 October, pending seven conditions its negotiator attributes to a "June Islamabad deal"; the conditions are not enumerated in any accessible text and this desk cannot establish whether that instrument is the collapsed 17 June memorandum or a separate one. Treat the declaration as a political act, not a measured fact, because the evidence cuts against it: the same 48 hours produced reporting that Persian Gulf crude exports have resumed near pre-war levels and that LNG shipments through Hormuz hit their highest level since the war began. On transits the providers disagree by roughly a factor of ten and the disagreement is methodological: Windward counted 119 transits in the week to 30 September, up 116%, with 16 on the 30th of which 6 were running dark, while AIS-geofence trackers give 1.6 per day on a seven-day average, against a pre-crisis baseline near 85 per day. No closure probability is published, and no barrel or ship figure sourced to a government official appears anywhere in this note — four such claims have now failed against vessel tracking. One dashboard headlines the strait "closed, day 218" but qualifies it as closed by war-risk insurance, with premiums at roughly 40 times normal and about $10.0m per VLCC voyage. That is an insurance statement, not a blockade.

The vessel tally and the date both need correcting. The three tankers were struck 29 September, not around 1 October: Mersin Prosperity, Sinbad and Al Ruwais, after Al Funtas on the 28th. Two more were hit on 1 October — the KOTC VLCC Kazimah III and an unnamed Panama-flagged crude tanker, both roughly five miles off the coast of Oman — making six in a week and 110 incidents in the conflict. No casualties. Attribution for the 29 September three remains unestablished; the nearest thing is Iran's Persian Gulf Strait Authority claiming the Kazimah III was on a "non-compliant list", which is a claim of targeting rather than of the strike. Separately Yanbu port on the Red Sea was struck on 1 October with loading briefly suspended then resumed, and on 4 October the Houthis claimed a ballistic-missile and drone strike on an Aramco facility south of Riyadh, which Saudi Arabia disputes as "misleading". Keep the theatres apart: the East–West pipeline strikes were launched from Maysan province, Iraq; the Houthi chokepoint seizure is at Mayun (Perim) Island, Yemen, about 2,000km away; and the tanker strikes are off Oman. All three are live and none is the other.

Russia–Ukraine. Zelenskyy said on 4 October that Ukrainian intelligence has obtained documents showing Putin issued a new doctrine authorising broader attacks on civilian targets, and that Ukraine will respond by targeting Russian energy infrastructure, particularly refineries. Russia's foreign ministry told foreign nationals to leave Ukraine. On 3 October Russian forces hit a Dnipro bridge in Kyiv and a cargo ship in Odesa; Moldova reported five Russian air weapons violating its airspace. No overnight intercept tally is published because none is reliably available. A declared Ukrainian campaign against Russian refineries, running into a G7 that is releasing barrels to fight a diesel crisis, is a product-crack story before it is a crude story.

Elsewhere. Japan's PM Sanae Takaichi marked one year as LDP leader with cabinet approval at 45%, down from above 65% at inauguration, no leadership challenge and no reshuffle; separately the Economy Minister said on 2 October that Japan "is no longer in a deflationary period" and does not need excessively loose policy. The market-relevant item is dated Friday: Takaichi was told in talks with President Trump that a weak yen is hurting US trade, verified across three outlets at headline level though the body text could not be retrieved. That removes her cheapest policy option and is the cleanest explanation for BoJ October hike pricing going from roughly 15% to 27% against a Tankan that argued for a hold. Brazil's presidential first round was Sunday 4 October and the live prediction book priced it as a two-horse race, Flávio Bolsonaro around 55.7% against Lula around 42.5%; the contract resolves at 14:59 AEDT today. Italy's Meloni formally asked Brussels on 2 October to build higher-than-expected inflation permanently into the deficit calculation, alongside national escape-clause requests worth roughly 1.5% of GDP through 2028; the Commission pushed back. In Australia, NSW Premier Chris Minns said governments should moderate spending in response to RBA advice, which the Treasurer reframed and the Opposition seized on — a fiscal-versus-monetary argument now running inside the governing party four weeks before the Statement on Monetary Policy.

This morning's setup. Bitcoin is $85,266 at 09:39 ET on 4 October, up 0.62% from Thursday's close, with total market cap $2.981tn and dominance up half a point to 57.5% — capital rotating into BTC out of the long tail, consistent with four consecutive sessions of ETH ETF outflows against BTC flipping positive. Funding is negligible on both and market-wide 24-hour liquidations are $38.7m, of which BTC and ETH are only $4.5m; positioning is clean, so a long-end shock would transmit through spot and ETF flows rather than through forced liquidation. The regional session ahead is structurally thin: Korea and mainland China shut, Stock Connect suspended until Thursday, Australian banks on a NSW holiday with no RBA fix, and Hong Kong trading a −2.60% break with no southbound bid for three more sessions.

Fed — the minutes are the only window, because nobody spoke. With the meeting on 27–28 October, the standard blackout runs Sat 17 Oct to Thu 29 Oct, so speakers are free all of this week and next. They did not use the weekend: the Fed's own 2026 speeches index has nothing later than 1 October (Bowman on regulation, Jefferson on the economy, Waller on AI), which means the Committee's reaction function to a +29,000 print is entirely unrevealed. That makes the minutes of the 15–16 September meeting, Wednesday 7 October at 14:00 ET (05:00 AEDT Thursday), the single most important Fed event of the week and the only look inside a 12–0 hike that 16 of 18 officials followed with a dot implying at least one more this year. Against that dot distribution, 17–23% priced for October is a large gap, and December at 84% is where the Committee's own projection lives.

RBA — the dispersion is the trade and there is nobody to narrow it. The 29 September hike to 4.60% was unanimous, citing higher global energy prices, stronger-than-expected recent inflation, domestic capacity constraints and AI-driven technology price increases with firms reporting pass-through intentions — against cooling consumer spending, falling house prices in most capitals and an easing labour market. The guidance was explicit: further tightening "may be needed to prevent inflation from becoming embedded". The market now prices 18% for 3 November. Westpac and ANZ both have a November hike as base case at 4.85%, Westpac with an explicit condition — "absent a lasting resolution of the Middle East conflict beforehand", which with Brent above $100 is not satisfied. CBA holds near term while flagging November risk; NAB has the peak at the current 4.60%. Outside the big four, EQ Economics argues the cash rate needs to go above 5%. And there is no scheduled RBA communication before the minutes on 13 October — the diary's first speaking engagement is 18 October (Kent, a News 24 interview), correcting the 19 October date this desk published on Friday, then Kent again on the 19th, Connolly 20th, Hauser 21st, Jones 26th. The Board goes into the 3 November Statement with no Governor-level word at all.

ECB — the accounts on Thursday, and the TPI question nobody has answered. The decision date in Sydney time needs correcting: the Governing Council meets 29 October local, announcing at 14:15 CET, which is 00:15 AEDT on Friday 30 October — the carried cell conflated the local date with the AEDT clock. (The Fed and BoC cells, which look like they contradict a 28 October local date, are correct and should not be "fixed".) October hike pricing has collapsed to roughly 7–9% and December sits near 82%; Schnabel, making the hawkish case, said it "points away from an October hike", which from that quarter is the decisive datapoint. Rehn's line on 2 October is the one to carry: energy price surges and rising long-term rates create "unusually high uncertainty" for the projections. And the live question under V025 is unanswered: on 1 October Nagel said the TPI targets the effective conduct of monetary policy "rather than specific sovereign spread levels" — on the day French spreads set a 12-month high. That is a pre-emptive refusal to name a trigger. He speaks again today at 18:45 AEDT, and the accounts of the September meeting land Thursday at 22:30 AEDT, the only look inside the Council before the decision.

BoJ — the reaction function acquired a new input on Friday. October pricing nearly doubled to 27% against a Tankan that argued for a hold (large manufacturing +24 against +25 expected, large non-manufacturing backwards from +37 to +35, one-year inflation expectations 2.7% to 2.6%). The explanation is not the survey. Tokyo core CPI printed 2.7% in September, above the 2% target for the first time in nine months; the September meeting summary showed policymakers shifting toward preventing an inflation overshoot, with one member warning that delay "could force double shocks"; and on 2 October the Prime Minister was told in talks with President Trump that a weak yen is hurting US trade. Takaichi's established position is that a weak yen benefits exporters, and her government is preparing a record budget with JGBs already under pressure. US pressure on the currency removes her cheapest option and raises the political cost of a BoJ that lags — which is the cleanest available explanation for 15% going to 27%. Hold the trio together: a 30-year one basis point from its all-time high, a record budget, and a currency now under foreign political pressure, into a 30 October decision with an Outlook Report. ⚠ The yen-talks detail is three-source at headline level only; the body text could not be retrieved.

THE ITEM THAT MATTERS MOST: a view closed itself on a pre-committed condition, and it closed WRONG. V029 was opened on 1 October with the condition written in advance that "a September payrolls print below +50K on Friday 2 October closes it immediately". The print was +29,000. The view is closed, at 4.83% on the official par 2-year against a 4.89% entry — six basis points against a bias of higher yields, so it is marked WRONG, notwithstanding that the instrument closed higher in yield on the day itself. The pre-commitment removed any scope to argue that the subsequent bear flattening vindicated the view. Six consecutive closes have now fired on conditions written in advance.

THE SECOND ITEM IS THE 5s30s OVERRIDE, WHICH FIRED AFTER FIVE DECLINES AND TWO ONE-BASIS-POINT MISSES. The re-entry was written as 56bp on the official par curve and nothing else — not a vendor mark, not a rounded one. 1 October printed 60bp and 2 October printed 57bp. Two consecutive official closes above the trigger, verified from the Treasury's own table this morning. V030 is opened at 57bp. The rule that declined this re-entry five times, twice at a one-basis-point cost, is the same rule that just took it without a discussion.

THE THIRD IS THREE CONVICTION CUTS, AND ALL THREE ARE ARGUED FROM THE VIEW'S OWN PRIMARY LEG RATHER THAN FROM THE PRICE. V017 High to Med, because the thing confirmed on 30 September — IG pinned flat while the distressed tail blew out — did not repeat one session later. V028 Med to Low, because the mechanical engine in its rationale is at −0.80 standard deviations rather than long across all speeds. V006 Med to Low, because RBA November pricing halved to 18%. Each of these reverses or mirrors a change made in the last two editions, which is what a convention that cuts both ways looks like in practice.

Book and scorecard. Seven open, twenty-three closed. One closed and one opened, both on pre-committed conditions. The trailing record is 3 right / 13 wrong / 7 scratch across 23 closed — 3 of 16 decided, which is bad and is stated plainly rather than framed. Three convictions were cut this morning and none was cut on the price: V017 because its confirming evidence did not persist a session, V028 because the flow mechanics in its rationale resolved before the catalyst arrived, V006 because its primary leg halved. Set against the two upgrades made on Friday, argued the same way in the other direction, that is a convention working rather than a convention being applied selectively.

Portfolio-level read, and the concentration has changed shape. The book remains one idea expressed several ways — that the term premium is the binding constraint and the policy path is not — and Friday was as clean a test of that idea as the calendar offers: policy odds halved, the curve rose at every tenor. The idea is now expressed most directly in V030, which is the new centre of the book, with V025 the same trade in European sovereign credit and V006 in an equity market that has no offset to it. V017 has drifted from a dispersion trade toward a beta trade and has been marked down accordingly. What was removed is the front-end leg: V029 is closed, so the book no longer carries a direct policy-path position — which is the right place to be on a tape where a 61,000 miss moved the curve the wrong way. V023 and V024 remain the two genuinely independent legs. The single largest risk to the book is now a bull flattener: a successful 30-year reopening on Thursday that takes 5s30s back inside 50bp hurts V030 and V025 together, and the 2 October credit print decides whether V017 is a third casualty of the same move.

These are analytical framings for a professional reader, expressed in the vernacular of a macro desk; they are not personalised investment advice and carry no position sizing. The "what changes the view" column is the accountability mechanism — every view is logged and scored in the project's views ledger, and the hit rate above is published whether it flatters or not.