The bottom line
- The RBA hiked to 4.60% unanimously and then Bullock talked the market out of the next one. The statement hardened — the conditional “if upside risks materialise” became the flatter “including increasing the cash rate target further if needed”, and the Board dropped its dated inflation-return guidance (August said “until late 2027”; September says only “a reasonable period”). The press conference then went the other way: Bullock disclosed that a hold had been debated, and said “what is the hope here, is that this will be restrictive enough… then maybe there doesn't need to be any more interest rate rises.” The curve believed the press conference. The 3-year rallied 2bp to 5.02% while the 10-year cheapened 2bp to 5.43%, steepening 3s10s 4bp to 41bp — which is the exact mechanism this desk named as Risk #1 yesterday, and it fired V004's invalidation. Only 43% is priced for 3 November and only ANZ among the majors formally forecasts it. The decisive date is not the meeting: it is the quarterly CPI on 28 October, six days before it.
- The S&P closed 7,671.01, nine points through the 7,680 line, and V028's confirm trigger fired on a level this desk deliberately kept as written the day it discovered the level's rationale was false. Yesterday's edition established that both legs of the “two-mechanism” confluence were wrong — the 200-day is ~7,209, not 7,679.54, and the gamma flip is ~7,700 — and then declined to move the trigger, on the principle that a restatement must be published in advance of a session rather than after a move. One session later it fired anyway. Conviction rises Low → Med. The supporting structure got worse, not better: net gamma negative on all four sources with 29 September data, a call wall that has collapsed from 7,800/8,000 to ~7,700 against a put wall at ~7,670, and 47.90% of the index above its own 200-day while the index sits 6.4% above its.
- Crude gave back the Hormuz premium, and the carried physical picture was wrong in the house's favour. Brent fell to ≈$102.70 (−4.65%) and WTI to ≈$89.32 (−7.18%) from verified Monday settles. Three named catalysts, not one headline: the Saudi East–West pipeline is running at ~3.5 mbd, about half its 7.0 mbd capacity — not the “100% with zero redundancy” this desk published, and Red Sea loadings have resumed; the US ordered an SPR release of up to 40 million barrels with the reserve at its lowest since 1982; and Qatar is carrying new Iranian proposals to Washington with a US response expected today. Against that, Trump floated and then publicly disowned sanctions relief within three hours on Monday night. The curve still says acute-but-finite: Nov–Dec backwardation over $7 in one month.
- The credit blackout broke after three dark sessions and the house view does not confirm — because the second condition fails outright. All three series posted: IG 83bp, HY 302bp, CCC 1,146bp at 28 September, from 79/280/1,112 on the 24th. CCC is four basis points from the 1,150 confirmation level. But the trigger reads “CCC through 1,150bp with IG flat”, and IG is 4bp wider, not flat. Everything gapped together — the CCC/IG ratio actually fell from 14.1× to 13.8× — so this is broad rates-driven beta, not the idiosyncratic low-quality decompression the view is built on. The short leg is working and the long leg is bleeding. V017 stays open, unconfirmed, at Med.
- The bond market and the equity market are pricing different worlds, and the gap is now measurable. The 30-year set a 24-year intraday high at 5.612%, the highest since June 2002, while the front end rallied — a clean bear steepener driven by a Conference Board confidence print that collapsed to 81.9 against 89 expected, a 12-year low, with JOLTS openings down to 7.079M. The detail that matters for the Fed: 12-month inflation expectations rose to a 6.1% mean. Meanwhile MOVE at 101.82 tops 59 of the previous 60 readings while the VIX sits at ~16.03 in its 120th consecutive day of contango. Four Fed speakers after the bell split on where it ends — Williams: “one more hike this year is likely enough”; Goolsbee: five and a half years above target is “playing with fire”. Core PCE lands tonight at 22:30 AEST.
- Asia's semiconductor break narrowed instead of broadening, and Taiwan's feared catch-up never came. The TAIEX reopened after two closed sessions having priced none of the global weakness and gave back only 0.82% — with TSMC closing unchanged and the damage concentrated in MediaTek at −7.10%. Its intraday high was above the 24 September close, so it did not gap down at all, which vindicates yesterday's decision to decline a catch-up short as “a bet on an opening print”. The KOSPI opened −0.66% and closed −0.27%, recovering more than half its gap even with semis down ~5%; mainland AI-chip names rose. Micron reports Thursday ~06:30 AEST and is the real test.
Overnight recap
Tuesday 14:30 AEST — the RBA hikes, and the press conference is the trade
The Monetary Policy Board raised the cash rate target 25bp to 4.60%, unanimously — the statement says so in terms — the fourth hike of 2026, 100bp of cumulative tightening this year, and the highest cash rate since 2011. It met consensus exactly: all four majors and, on one survey, 29 of 29 economists. Because the decision was fully discounted, the information was all in the language, and the language moved in two directions at once.
The statement hardened. Three drafting changes matter. The conditional construction used in August — the Board would hike further “if upside risks materialise” — became the flatter “including increasing the cash rate target further if needed”, which is a lower bar. The Board dropped its dated inflation-return guidance: August said inflation would not return to target until late 2027, September says only “a reasonable period”, and removing a date removes a constraint. And on energy the statement used unusually strong adjectival language, calling global energy prices “much higher” than assumed in the August forecasts, alongside “some of the upside risks to inflation are materialising” and the standing formulation that “growth in aggregate demand needs to remain subdued for a period”.
The press conference softened. At 15:30 Bullock disclosed that the Board had actively debated a hold — which is compatible with a unanimous vote and is the only genuinely new information in the event — and then declined to validate a November follow-up. The quote the market traded: “What we are predicting, what is the hope here, is that this will be restrictive enough… then maybe there doesn't need to be any more interest rate rises.” She said the Board wanted to see how four rate rises “feed through”, argued the stance is already restrictive (“they are up where they were probably when we had interest rates at 4.35%”), and pointed at weak productivity rather than wages as the inflation driver.
The curve believed the press conference, and the equity market believed it twice. ACGB 3s rallied 2bp to 5.02% while 10s cheapened 2bp to 5.43% — 3s10s steepened 4bp to 41bp. The ASX is the cleaner tell: the same publisher's live blog had the index down 6 points at 14:50, twenty minutes after the decision, and it closed up 29.6 points. Essentially the whole day's gain was made across and after the press conference. AUD/USD spiked to 0.7029 on the statement, broke 0.7000 on the presser to a Sydney close of 0.6993, and then recovered through the US session to ≈0.7018–0.7020. Pricing for 3 November sat at ~43% immediately afterwards — minimal change from pre-announcement.
The street split, and the split is wider than the pricing. ANZ is the only big-four bank formally calling a consecutive November hike, to 4.85%. Standard Chartered read Bullock as a dovish shift and moved to an extended hold; ING expects one final 25bp in Q4 contingent on oil, explicitly flagging that sharp crude declines would prompt a pause — which is exactly what crude then did overnight. Commentators are calling November the base case; the instruments are not. The gap between the narrative and the 43% is the thing to write about, and the quarterly CPI on 28 October is what closes it.
Asia — the semiconductor break narrows, and the mainland reverses
Taiwan reopened after two closed sessions (25 and 28 September, Teachers' Day) having priced none of two sessions of global semiconductor weakness, and the feared catch-up did not materialise: the TAIEX closed 47,631.96, −0.82%, its chain tying to the cent against the 24 September close, on turnover of NT$785bn. Two details matter more than the level. Its intraday high of 48,045.13 was above the 24 September close, so it opened firm and then faded — this was not a gap-down. And TSMC, over 40% of market cap, closed unchanged at NT$2,475, as did Hon Hai; the damage was MediaTek at −7.10%. Against KOSPI −2.70%, Shenzhen −3.44% and a 1,157-point Nikkei reversal over the same window, Taiwan's catch-up was a fraction of what those tapes implied.
Korea stabilised at the index level while the sector kept bleeding. The KOSPI opened −0.66% and closed 6,870.81, −0.27% — both the open and the close reconstruct independently to the verified prior, which is the strongest chain evidence available — with the KOSDAQ at 843.87 (−0.32%) and the won at 1,365. Underneath, semiconductors “broadly plunged 5%”, with Samsung Electronics also trading ex-dividend, and foreign plus institutional investors net sold roughly $3bn. An index that recovers more than half its opening gap while its largest sector falls 5% is a rotation being absorbed, not a market breaking.
The mainland rose and Hong Kong fell — an exact reversal of Monday. Shenzhen +0.34% off its nine-month low, CSI 300 +0.10%, Shanghai +0.18%, against a Hang Seng at 24,523.57, −0.48%. The mainland's driver was stimulus expectation — accelerated Q4 government bond issuance plans and possible activation of unused local-government borrowing quotas — plus positioning into today's PMIs; Cambricon, a domestic AI-chip name, rose 2.55% on a day semis fell elsewhere. Hong Kong's was higher oil, a US 10-year above 5.2%, continued Chinese-tech weakness and stated caution ahead of the National Day closure: Tencent −1.8%, Xiaomi −2.6%, Meituan −2.0%.
Japan's headline divergence has a benign explanation and should not be read as risk-off. The Nikkei closed 65,481.27, −0.60% against a TOPIX at 4,041.13, −1.72% — a 112bp gap. More than half of TOPIX constituents traded ex-dividend on 29 September, Japan's fiscal-half record date, which is the entire story; the named decliners (Sumitomo Mitsui −4.0%, Mitsubishi UFJ −3.5%, Toyota −3.5%) are among the largest ex-dividend names and those are gross, not clean, moves. JGBs went flat-to-nothing after two sessions of front-led bear flattening: 2y 1.97%, 10y 3.09%, 30y 4.18%. That is the first session in three that has not added to October BoJ pricing. The exchange archive also posted Monday's row, converting the carried 65,878 to an exchange-verified 65,877.62 and confirming the reversal was 1,157.12 points off the high — and that the 28th closed exactly at its low, with no bounce into the bell at all.
Europe — a semiconductor melt-up sold into the close by two shocks
The Stoxx 600 closed 638.08, −0.09%, having traded as much as 0.7% higher: a semiconductor rally on an Anthropic IPO headline was given back on the US bond move and the consumer-confidence collapse. Infineon +4.8% (verified three ways), ASML +4.18%, BESI +4.42%, ASM International +4.24%, ams OSRAM +8.7%, Aixtron +4.4%; Julius Bär +7.2% and Legrand +6.0% led outside tech. The other side: Fresenius −4.9% on an FDA manufacturing warning, Repsol −4.76% as crude de-rated, Vistry −7.82% — reversing a UK housebuilder rally that had led Monday — and Bayer −3.8%. DAX 25,399, +0.10%; CAC 40 8,036, −0.53%, an 18-week low; FTSE 100 10,623, −0.57%; IBEX 19,517.50, −0.42%; SMI 13,912.19, −0.22%; FTSE MIB 51,804, +0.09%.
The real European story was in the auctions, not the secondary marks. The Italian 10-year BTP auction cleared 4.58% against 4.10% previously — a 48bp jump in clearing yield — and the UK 10-year gilt auction cleared 5.38% against 5.16%, up 22bp, both on the same morning, with cover flat to lower. BTP–Bund went back through 100bp. Peripheral and semi-core supply is being repriced hard into quarter-end, and that is the cleanest available evidence for the house's OAT–Bund widener on a day the spread itself could not be marked. Spain's flash CPI was the data surprise: 4.9% against 4.6% expected, up from 4.3% — a 30bp beat and a 60bp acceleration, three days before the euro-area flash.
The US session — a bear steepener on a consumer shock
Two 10:00 ET prints did the work. Conference Board consumer confidence collapsed to 81.9 from a revised 88.6, against 89 expected — the lowest since 2014, a 12-year low and a 7.1-point miss. Present Situation 109.3 (−7.9), Expectations 63.6 (−5.9), a third consecutive decline and deep below the 80 recession-signal threshold. The line that matters for the Fed runs the other way: 12-month inflation expectations rose, mean to 6.1% and median to 5.1%, with the Board's chief economist noting references to “prices, the high cost of goods and services, and oil and gas prices in particular” at new heights. The survey window, 1–23 September, captured the 16 September hike. JOLTS openings fell to 7.079M against ~7.24M expected, openings −256k, hires +46k, layoffs −61k, quits 3.066M. Case-Shiller and FHFA both ran hot on house prices; Dallas Fed services turned negative.
The curve steepened violently on it. The front end rallied — vendor reads put the 2-year around 4.90%, roughly 2bp richer than the official 28 September par close of 4.92% — while the 30-year set a new 24-year intraday high at 5.612%, the highest since June 2002, and the 10-year pushed to ≈5.26%. 2s10s re-steepened ~4bp to ≈36bp, exactly reversing Monday's flattening, and 5s30s widened ~2bp to ≈52bp. The context is three weeks of ugly supply: the 5-year auction on 23 September cleared 5.033% with the second-largest tail on record and the weakest bid-to-cover since December 2018, and the 7-year cleared 5.085%, the highest since April 1993, with indirects at 57.2%. Tuesday's own bills told the same story — the 52-week bill cleared 4.400% against 3.980% previously, 42bp in one cycle. All of this is happening with Treasury's long-end buybacks running at double size since 9 September.
Equities were small and split. S&P 500 7,671.01 (−0.17%), Nasdaq Composite ≈26,796 (−0.09%), Dow ≈51,364 (−0.23%), Russell 2000 2,809.54 (−0.30%), VIX ≈16.03. Mega-cap tech held — QQQ, tracking the Nasdaq-100, rose 0.20% — while everything broader sagged, which is the whole shape of the session. Utilities led sectors at +0.8%, which on a day the long bond made a 24-year high reads as a defensive bid rather than a rate-sensitive one. Single names: Fair Isaac −21.6% on FHFA mortgage credit-pricing changes, dragging Equifax −2.65%; Summit Therapeutics +18% on a $2bn AstraZeneca investment; Carnival +11.65%; AMD +1.2–1.4% on its acquisition of World Labs. After the bell, four Fed speakers: Williams — “one more hike likely this year is likely enough”, the dovish cap; Goolsbee — five and a half years above target is “playing with fire”, and AI productivity expectations “could create a high danger of overheating now”; Musalem and Barr both leaning hawkish on AI capex as a demand pressure. That AI-capex-as-inflation-source argument now runs through two Fed speakers and the RBA statement on the same day.
Market dashboard
Tuesday 29 September — cross-asset change
| Equities | Close | 1d | Note |
|---|---|---|---|
| S&P 500 | 7,671.01 | −0.17% | ⭐ Three routes converge and the trigger fired. Quote page “Closed·15:59:58”; SPY's settled −0.18% implies 7,669.5; two gamma vendors' near-bell spot 7,670.04 and 7,671.59. 8.99pt BELOW V028's 7,680 confirm line. A circulating 7,683.73 is a 09:53 ET morning mark and is rejected |
| Nasdaq Composite | ≈26,796 | ≈−0.09% | Derived from ONEQ (the Composite ETF), settled −0.09%. ⚠ QQQ tracks the Nasdaq-100 and rose +0.20%; a circulating 26,870.44 matches the NDX, not the Composite. −0.09% sits coherently between NDX +0.20% and Russell −0.30% |
| Dow Jones | ≈51,364 | ≈−0.23% | Derived from DIA's settled −0.2276% on a verified 51,481.51 prior. ⚠ The quote page still serves Monday's close under a “Closed” header |
| Russell 2000 | 2,809.54 | −0.30% | ⭐ A TWO-EDITION GAP CLOSED. The chain ties exactly — 2,809.54 + 8.37 = 2,817.91, the AP-verified prior. Small caps the session's worst major US index |
| VIX | ≈16.03 | −0.25% | A 16:13 ET read against a verified 16.07 prior; VIX calculates to 16:15. ⚠ FRED's official VIXCLS has not posted 29 Sep, so this is corroborated, not settled. No 29 Sep VIX3M, IVTS or contango count |
| Stoxx 600 | 638.08 | −0.09% | ⭐⭐ OBTAINED AFTER FIVE EDITIONS — and the reason it kept failing was our own prior. Traded +0.7% intraday. Route that worked: a wire story 24 minutes after the close, not a quote page |
| DAX · CAC 40 | 25,399 · 8,036 | +0.10% · −0.53% | Both tie against corrected priors. CAC at an 18-week low. Semis led Europe: Infineon +4.8%, ASML +4.18%, BESI +4.42% |
| FTSE 100 · IBEX · SMI | 10,623 · 19,517.50 · 13,912.19 | −0.57% · −0.42% · −0.22% | IBEX ties exactly; SMI ties to 0.04pt. ⚠ The Spanish page's narrative is a CFD proxy and its quote box ties — the mirror of the Italian page. Read both fields, take whichever reconciles |
| FTSE MIB · Euro Stoxx 50 | 51,804 · not published | +0.09% | ⚠ MIB is three-camped (+0.09% / “above parity” / −0.21% / +0.63%); the published figure is the only one that states a level and reconciles. ⚠ ESTX50 withheld — the only read was a 15:45 CET intraday on a day every index faded into the bell |
| Nikkei 225 | 65,481.27 | −0.60% | Chain ties exactly to an exchange-verified 65,877.62 prior, corroborated by a second unrelated source. ⚠ The archive has not posted 29 Sep |
| TOPIX | 4,041.13 | −1.72% | ⭐ THE 112bp GAP TO THE NIKKEI IS AN EX-DIVIDEND ARTEFACT, NOT A SELLOFF. More than half of TOPIX constituents traded ex-div on Japan's fiscal-half record date. Anyone quoting this as risk-off is wrong |
| Hang Seng · HS Tech | 24,523.57 · not obtained | −0.48% | HSI chain ties and the page states its own prior. ⚠⚠ HS TECH LOST AGAIN ON TWO ROUTES — the historical table stops at 28 Sep and the quote page serves a 28 Sep intraday against a 25 Sep prior. Tencent −1.8%, Xiaomi −2.6%, Meituan −2.0% say the divergence widened; that is inference, not a figure |
| CSI 300 · Shanghai · Shenzhen | 4,345.21 · 3,830.45 · 12,902 | +0.10% · +0.18% · +0.34% | ⭐ All four China/HK chains tie and both pages state their own priors — the first clean session in several. The mainland ROSE against a falling Hong Kong, an exact reversal of Monday. Driver: accelerated Q4 bond issuance and unused local-government quotas. Last session before Golden Week |
| KOSPI · KOSDAQ | 6,870.81 · 843.87 | −0.27% · −0.32% | ⭐⭐ OPEN AND CLOSE BOTH RECONSTRUCT to the same verified prior — the strongest chain evidence in the edition. Opened −0.66%, closed −0.27%: it recovered more than half its gap while semis fell ~5%. Foreign + institutional net sold ~$3bn |
| TAIEX | 47,631.96 | −0.82% | ⭐ Chain ties to the cent across a two-session closure. Day high 48,045.13 was ABOVE the 24 Sep close — it did not gap down. TSMC unchanged at NT$2,475; MediaTek −7.10%. Turnover NT$785.04bn |
| Sensex · Nifty 50 | 72,529.07 · 22,716.20 | −0.33% · −0.28% | ⭐ Both legs reconstruct exactly. 13 of 16 sectoral indices lower; Nifty IT −1.48%; only Pharma/Healthcare supported. FII selling >US$2.17bn in September; the Nifty is −5.7% on the month — week nine after a seven-week losing streak |
| S&P/ASX 200 | 8,709.3 | +0.34% | ⭐ Chain ties: 8,679.70 + 29.61. ⚠ A vendor printed the level as 8,689.70 while its own change and percentage both imply 8,709.3 — caught by arithmetic. IT +4.61% carried it; Financials −0.13%, reversing Monday's +1.16%. Codan +23.9%, Megaport +9.4%. Breadth 115/74/11 |
| NZX 50 | 13,683.63 | −1.06% | Chain ties to the cent and the source states its own prior. Lowest since 16 Sep; erased all of Monday's gain. Attributed to global yields and oil, not the RBA |
| Jakarta (28 Sep) | 6,147.86 | −1.51% | ⚠ CORRECTION to yesterday's ≈6,151 / −1.45%, which was wrong on both legs. Now verified. No 29 Sep close obtained |
| Rates & credit | Level | 1d | Note |
|---|---|---|---|
| ⭐ Treasury's official par curve for 28 September has POSTED, discharging yesterday's first verification action. 29 September has not (it posts ~18:00 ET). US yields below for the 29th are vendor levels; the 28th is primary. | |||
| Official par curve, 28 Sep | 2y 4.92 · 5y 5.06 · 10y 5.24 · 30y 5.56% | 2y +11 · 10y +7 · 30y +7bp | ⭐⭐ THREE CARRIED UNCERTAINTIES SETTLED AT ONCE. The 30-year published as a 5.53–5.58% range is 5.56%; the 2-year basis question resolves at 4.92% (top of the carried band); 3y 5.01, 7y 5.15, 20y 5.60. 2s10s 32bp, 5s30s 50bp — both flatter than 25 Sep's 36/51 |
| UST 2y (29 Sep) | ≈4.90% | ≈−2bp | The front end RALLIED on the confidence collapse — the first meaningful richening in September, a month in which it added ~56bp |
| UST 5y · 10y (29 Sep) | ≈5.07% · ≈5.26% | ≈+1 · ≈+2bp | The 5y has still never closed above 5% on the official series; it is now 6bp above it on vendor marks |
| UST 30y (29 Sep) | ≈5.59% | ≈+3bp | ⭐ A NEW 24-YEAR INTRADAY HIGH AT 5.612%, the highest since June 2002, corroborated three ways. And it happened with Treasury's long-end buybacks running at double size since 9 September — that is the structural fact of the session |
| 2s10s · 5s30s (29 Sep) | ≈36 · ≈52bp | ≈+4 · ≈+2bp | A clean bear steepener, exactly reversing Monday's bear flattener. ⚠ The 5s30s steepener candidate was declined for a third consecutive edition and this time the decline cost money — see §08 |
| Bills (29 Sep auctions) | 6-wk 3.970% · 52-wk 4.400% | prior 3.870 · 3.980 | +42bp on the 1-year in a single auction cycle. The cleanest read of the day: the bill market is pricing the hiking path outright |
| Recent coupon auctions | 5y 5.033% · 7y 5.085% | tails 3.1bp · ~0.7bp | ⚠ The 5-year's tail was the second-largest on record with a bid-to-cover of 2.212, the weakest since December 2018; the 7-year cleared the highest since April 1993 on indirects of 57.2%. Two consecutive tailing auctions across tenors is a pattern. All three settle today |
| JGB 2y · 10y · 30y | 1.97 · 3.09 · 4.18% | 0 · 0 · +1bp | The bear flattening STALLED — the first session in three that has not added to October BoJ pricing. The 2y still embeds ~72bp over a 1.25% policy rate, close to three hikes |
| Bund 2y · 10y · 30y | 3.27 · 3.61 · 3.95% | 0 · −1.9 · −1.0bp | Bunds bid against everything else — a safe-haven bid inside a European supply shock. ⚠ Changes carry ±0.5bp of rounding against four-decimal priors |
| OAT 10y · BTP 10y | 4.81 · 4.59% | +4.0 · +2.0bp | ⚠ The vendor's BTP change column had the sign wrong (it printed −2bp against a recomputed +2bp). OAT at its highest since July 2008 |
| OAT–Bund | no mark | — | ⚠⚠ UNMARKABLE, AND THE ROUTE IS NOW DEAD, NOT MERELY LAGGING — the dedicated same-page series 404s. Last established level 105.4bp (25 Sep). OAT +4.0bp against Bund −1.9bp implies ~6bp of widening, i.e. the low 110s — published as direction only. Mixing legs across pages is the documented artefact that once gave 110.5bp against a true 97bp |
| Auction tails — the real European story | BTP 10y 4.58% · Gilt 10y 5.38% | +48bp · +22bp | ⭐ Both cleared on the same morning against previous auctions at 4.10% and 5.16%, with cover flat to lower. BTP–Bund back through 100bp. This is the evidence for the widener on a day the spread itself cannot be marked |
| Gilt 2y · 10y · 30y | 4.74 · 5.36 · 5.88% | −3.0 · −4.3 · −1.0bp | Gilts gave back some of Ramsden's Monday move (+8bp at the 2y). ⚠ The vendor's own 2y change read −15.1bp against a recomputed −3.0bp — a 12bp error. Taylor pushed back publicly: policy “should not react mechanically to energy price movements” |
| ACGB 2y · 3y · 10y | 5.06 · 5.02 · 5.43% | +1.0 · −2.0 · +2.0bp | The RBA's own curve disagreed with its statement. Premium over the new 4.60% cash rate: 2y +46bp, 3y +42bp, 10y +83bp. The 3y still embeds roughly 1.7 further hikes — the premium compressed but did not collapse |
| ACGB 3s10s | 41bp | +4bp | ⭐ STEEPENED 4bp — V004's invalidation condition, written as “a dovish RBA with a sticky 10y”, fired on its terms. The view is closed 2bp in the money from a 43bp entry, having been 6bp in the money on Monday. See §08 |
| US IG · HY · CCC OAS | 83 · 302 · 1,146bp (28 Sep) | +4 · +22 · +34bpover 24→28 Sep | ⭐⭐ THE THREE-SESSION BLACKOUT BROKE. 25 Sep: 81 / 293 / 1,128. 28 Sep: 83 / 302 / 1,146. All three verified; the 24 Sep values match the carried figures exactly, validating continuity. CCC is 4bp from the 1,150 confirm level — but IG is 4bp WIDER, not flat, so the second condition fails outright. The CCC/IG ratio fell 14.1× → 13.8×: this is rates beta, not decompression. ⚠ No 29 Sep observation yet |
| SoftBank | $11.1bn priced 23 Sep | 7.5y at 9.75–9.875% | ⭐ SIX EDITIONS OF FAILURE ANSWERED FROM THE PRIMARY SIDE INSTEAD. The largest high-yield corporate bond sale on record — 3.5y $1.0bn at 8.75–8.875%, 5.5y $4.5bn at 9.375–9.5%, 7.5y $4.5bn at 9.75–9.875%, book >$20bn, BB+ at both S&P and Fitch, funding the $65bn OpenAI commitment. A BB+ credit clearing 7.5-year paper at 9.875% into a widening tape is itself the mark. ⚠ Secondary levels failed a seventh time and the chase is formally abandoned |
| FX | 29 Sep close | 1d | Note |
|---|---|---|---|
| ⭐ The 25 September FX protocol PASSED this session, having failed entirely on Monday. Levels are same-vintage vendor closes whose implied priors were checked against the carried 28 September set; the DXY reconciles to the verified prior to the fourth decimal. | |||
| DXY | 101.3692 | +0.17% | ⭐ 101.1973 + 0.1722 = 101.3695 — an exact tie to the verified prior. ⚠ But the mandated residual FAILS on magnitude: weight-summing the six constituents gives +0.104pp against a stated +0.170pp, so 39% of the index move is unexplained by its own legs. Direction is coherent (all six contributed positively); the yen leg is not the explanation. Most likely a snapshot-timing mismatch between the index feed and the currency pages. Published, not reconciled away |
| EUR/USD · GBP/USD | 1.13619 · 1.32372 | −0.08% · −0.13% | Implied priors within 2 and 9 pips of the carried set |
| USD/JPY | 157.39 | +0.01% | ⚠ Implied prior 14 pips from the carried 157.23 — the weakest chain in the set, and the one leg not sourced from the same vendor family. ~2.6 yen below the 160 policy line. No MoF commentary |
| AUD/USD | ≈0.7018–0.7020 | ≈flat | ⭐ The intraday path is the story: ~0.7008 pre-decision, 0.7029 high on the statement, BROKE 0.7000 on the press conference, Sydney close 0.6993, then recovered through the US session. Two vendors give 0.7018 and 0.70202. ⚠ A third page was rejected — it presented the post-decision spike as a “previous close” under pre-decision commentary |
| NZD/USD | 0.56519 | −0.27% | Implied prior within 6 pips of carried. The Kiwi did the work in the cross, not the Aussie |
| USD/CAD · USD/CHF | 1.4183 · 0.8342 | +0.06% · +0.26% | Canadian GDP flat m/m against a +0.4% prior |
| USD/CNY | 6.7083 | −0.07% | ⚠⚠ THE PBoC FIX IS SINGLE-SOURCE AND INTERNALLY IMPLAUSIBLE — do not lead with it. A reported 6.7411 sits 234 pips from the stated Reuters estimate and 328 pips weaker-yuan than the onshore spot close. A 234-pip miss on a fix normally estimated to within a handful is more consistent with a transcription error than with policy. Direction only: the fix was set marginally weaker day-on-day and spot closed stronger than the fix |
| AUD/NZD | 1.24211 | +0.31% | ⭐⭐ THE CLEANEST MARK ON THIS VIEW IN FOUR EDITIONS. Cross-computed from two legs of the same table; the identical construction on 28 September gives 1.23824 against the carried 1.2384 — a 1.6-pip match that validates both legs and the method. ≈+0.86% from a 1.2315 entry |
| USD/MXN · USD/KRW | 17.9716 · 1,360.28 | 28 Sep · 28 Sep | ⚠ Both pages stamped 28 September — no 29 Sep print. The peso's +2.89% on Monday is the largest FX move in the set and could not be confirmed as extended |
| USD/INR · USD/SEK | 96.1040 · 10.0264 | +0.16% · +0.65% | The krona was the weakest G10 leg and the largest single contributor to the DXY's residual problem |
| Commodities & digital assets | Last | 1d | Note |
|---|---|---|---|
| Brent (Nov-26 front) | ≈$102.34–102.70 | ≈−4.65% | ⚠⚠ LEVELS ARE WELL CORROBORATED; THE CHANGE DEPENDS ON A CONTESTED BASE. Every 29 Sep vendor computes its change off ~$105.3, not off the $107.71 Monday settle this desk published — which would give −2.75%. We use our own published prior for continuity and flag the conflict. The $107.71 stands on its own exact tie to Friday's $104.32 and on a Monday rally that happened in US hours, after the intraday prints the sceptical vendors appear to be anchored on |
| WTI (Nov-26) · Brent–WTI | ≈$89.32 · ≈$13.2–13.4 | ≈−7.18% | Three-way level cluster $89.22–89.77. Brent–WTI BLEW OUT from $11.48 — WTI underperformed hard, which is the SPR release showing up in the spread |
| Brent curve | Nov–Dec ≈$7.4+ | — | ⚠ No 29 Sep curve: the contract-by-contract page is frozen at a 28/09 09:09 GMT stamp across two fetches. The 28 Sep shape (Nov $104.13 / Dec $96.72 / Jan $93.62 / Mar $90.03) still says acute but finite. ⚠⚠ The contract-month trap is confirmed live: the quote page states Brent rolled to Dec-26 on 27 September and prints $96.84. Anyone quoting “Brent ~$96.8” is on the second contract and seven dollars light |
| Henry Hub · TTF | $3.05 · €69.20 | ≈−2.07% · ≈−2.97% | ⚠ Both vendor change columns compute off stale priors; figures recomputed. Gas fell with crude this time, unlike Monday |
| EU gas storage | 71.4% (28 Sep) | ≈808 of 1,132 TWh | ⭐ THE 2024-DATELINE DOUBT IS RESOLVED. The source benchmarks explicitly against 82.5% on the same date in 2025 (−11.1pp) and a 2021–25 average of 86.6%, which cannot be a recycled article. Below the five-year minimum of 73.8% for the period — the carried ~70.9% was the 26 Sep figure |
| Gold (spot) | $4,174.30 | ≈+1.37% | ⭐ Two sources, two currencies, same percentage — a 16:32 ET print and an independent CAD post-market summary both at +1.35–1.46%. Recomputed against the verified $4,117.70 prior gives +$56.60, +1.37%. ⚠ The vendor's own stated change implies a $4,114.10 prior, $3.60 off. Gold recovered roughly a third of Monday's 3.90% break but is still below $4,200 |
| Silver · Platinum · Gold/silver | $61.36 · $1,679.30 · 68.1 | +0.92% · −2.35% | Silver verified by triangulation through the CAD print. ⚠ Platinum is the weakest figure on the desk — its page contradicts itself across three fields and a second source dissents at ~$1,707. Published on internal arithmetic only. Gold outperformed silver on the bounce |
| Copper LME cash · 3M | $14,544.50 · $14,455.00 (28 Sep) | −1.33% | ⭐ YESTERDAY'S UNFILLED TARGET IS FILLED — the 28 Sep official settle posted on the usual one-day lag, and the 25 Sep row matches the carried figures exactly. Cash−3M narrowed to +$89.50/t from +$93; stocks fell only 150t to 251,350t. Falling price, stalled destocking and a narrowing backwardation is a coherently softer picture than the one carried |
| Iron ore | $96.92/t (28 Sep) | no new print | ⚠ No 29 September observation, so the streak stands at FIFTEEN consecutive sub-$100 prints — there is no sixteenth. With China shut 1–7 October there is no fresh demand signal for roughly a week and a half |
| Aluminium · Lithium · Uranium | $3,260.30/t · CNY 123,400/t · $89.30/lb | +0.27% · −2.33% · −0.22% | Lithium is the worst commodity on the desk: −22.15% on the month, deepening from −20.28%. Uranium refreshed after a three-session freeze (28 Sep figure) |
| BTC · ETH · SOL | $83,514 · $2,687.51 · $118.66 | −0.03% · +0.14% · −0.29% | ⭐ Crypto did not participate in the risk-off at all — the majors went nowhere while credit gapped and crude swung 5–7%. Total cap $2.94tn, BTC dominance 57.0%. ⚠⚠ The mandatory protocol had to be OVERRIDDEN: the two sources are ~14 hours apart, so pasting one's percentages onto the other's levels would have manufactured a false day. Changes derived from verified priors instead. A probable third sign inversion: the aggregator reports total cap +1.2% against flat-to-down majors |
| Spot ETF flows | BTC +$31.0m (28 Sep) | 25 Sep finalised +$134.5m | ⚠ 29 Sep shows the 0.0 placeholder across every issuer — that is NOT a zero. Note the 25 Sep row settled at the same value its incomplete version showed, so the 42× escalation precedent did not repeat. +$31.0m is a sharp deceleration from the +$2.4bn weekly pace to 25 September. ⚠ The 28 Sep ETH row could not be read |
| Crypto derivatives | BTC OI $26.5bn · ETH $18.2bn | liq. $47.5m · $51.2m | Funding mildly positive and rising on both. Under $100m of liquidations across both majors is very quiet for a session in which credit and crude moved hard. ETH liquidations exceed BTC's on 31% less open interest — the more levered book. ⚠ Per-asset, not market-wide; the page carries no timestamp |
| Counterparty (XCP) | $3.75 | — | ⚠ 24-hour volume of $231. Report the level, not the move — the −18.2% is meaningless at that liquidity |
Conventions: 1d = change on the Tuesday 29 September session, recomputed by this desk against a verified prior close; vendor change columns are discarded on principle after nine consecutive editions of failures. “≈” marks a derived or contested value and the reason is given in the row. Yields in per cent, changes in basis points; the negative sign is U+2212. Gold is spot; crude is the front-month Nov-26 contract on both Brent and WTI — read the contract label before trusting any Brent table. Australian spelling throughout.
What is driving markets
1. The RBA's statement and its press conference said different things, and the curve believed the press conference
This is the cleanest example of a communications split this cycle. The written statement lowered the bar to further tightening — “if needed” replacing “if upside risks materialise” — and removed the dated commitment on when inflation returns to target, both hawkish drafting choices. Thirty minutes later the Governor volunteered that a hold had been debated and used the word hope about whether 4.60% is restrictive enough. The instruments resolved the ambiguity immediately and in one direction: the 3-year rallied 2bp while the 10-year sold off 2bp, steepening 3s10s from 37bp to 41bp, and the ASX went from −6 points twenty minutes after the decision to +29.6 at the close. Rate pricing for 3 November barely moved, sitting at roughly 43%. What is left embedded is still substantial — the 3-year at 5.02% carries +42bp over the new 4.60% cash rate, about 1.7 further hikes — so the market has not concluded the cycle is over; it has concluded that the Bank is no longer leading it there. Read the divergence between commentary and pricing as the signal: houses are calling November the base case while only ANZ among the majors formally forecasts it and the physical curve prices less than half of one.
2. The energy shock is receding through the physical market, not through the headlines
Monday was a rejection headline and crude rose 3.25%. Tuesday was three pieces of physical and diplomatic substance and crude fell 4.7%, with WTI down 7.2%. The substance, in order of durability: the Saudi East–West pipeline is running at roughly 3.5 million b/d, about half its 7.0 mbd capacity, with Red Sea loadings resumed — which corrects this desk's own published claim that it was at 100% with zero redundancy, and that correction cuts against the bullish case. The United States ordered a further Strategic Petroleum Reserve release of up to 40 million barrels with the reserve already at its lowest since 1982. And Qatar is carrying new Iranian proposals to Washington with a US response expected today, on a mediation track that also runs through Pakistan and is described as having become more serious. Set against all of that: Iran has not softened its stated conditions, and the administration floated and then publicly disowned sanctions relief within three hours on Monday night, which is the cleanest available explanation for a violent two-way move. The forward curve has not capitulated — Nov–Dec backwardation of more than seven dollars in a single month still prices an acute but finite disruption — and the observed transit collapse is real on every provider's numbers, roughly 90–99% below normal.
3. The term premium is not an energy story, and Tuesday proved it
If the long end were pricing the Gulf, it should have rallied on a 4.7% fall in crude. Instead the 30-year set a fresh 24-year intraday high at 5.612%, the highest since June 2002, on the same session, while the front end rallied on a collapse in consumer confidence to a 12-year low. That is a bear steepener, and it is the shape that says the problem is supply, term premium and fiscal credibility rather than the near-term inflation path. Three pieces of corroboration sit alongside it. Treasury's long-end buybacks have been running at double size since 9 September and the long bond made a new high anyway. The last two coupon auctions tailed — the 5-year's was the second-largest tail on record with the weakest cover since December 2018 — and Tuesday's 52-week bill cleared 42 basis points above the prior auction. And the same thing is happening abroad on the same morning: the Italian 10-year auction cleared 48bp higher than its predecessor and the UK 10-year gilt auction 22bp higher. Meanwhile the options market has noticed in bonds and not in equities: MOVE at 101.82 tops 59 of the previous 60 readings while the VIX sits near 16 in its 120th consecutive day of contango.
4. The credit widening is beta, not decompression — and that changes how you should be short
Three trading sessions of missing data broke on Tuesday, and the answer is more interesting than the headline. CCC has gone 1,112 → 1,128 → 1,146bp and is four basis points from the level at which the house view confirms. But the view's trigger requires CCC through 1,150 with investment grade flat, and IG has gone 79 → 81 → 83bp. High yield has gone 280 → 293 → 302. Everything widened together, and the ratio of CCC to IG actually compressed, from 14.1× to 13.8×. That is the signature of a rates-driven repricing of the whole credit stack, not of low-quality issuers being singled out — which is what the underweight was designed to capture. The driver is legible: the 5-year Treasury auction cleared above 5% for the first time since 2007, municipal benchmarks in the 1–3 year sector rose as much as 48bp with liquidations at the fastest pace since the COVID era, and the long bond made a 24-year high. The one genuinely idiosyncratic datapoint runs the same way: SoftBank priced $11.1bn on 23 September, the largest high-yield corporate bond sale on record, clearing 7.5-year paper at 9.875% on a BB+ rating. A book of more than $20bn says the demand exists; the coupon says what it costs.
5. The crowded semiconductor trade stopped breaking in four tapes at once
Monday's edition led on a simultaneous break across Korea, Japan, mainland China and Hong Kong. Tuesday did not repeat it. Taiwan reopened after two closed sessions having priced none of that weakness and gave back only 0.82%, with its intraday high above the prior close and TSMC — more than 40% of market cap — closing unchanged; the damage was one name, MediaTek, at −7.10%. Korea's index recovered more than half its opening gap even with semiconductors down about 5%, so the rotation is being absorbed by the rest of the market rather than dragging it. Mainland AI-chip names rose outright. In Europe the sector was the day's clear leader — Infineon +4.8%, ASML +4.18%, BESI +4.42% on an Anthropic IPO headline — and in the US the Nasdaq-100 was the only major index up. So the break is now idiosyncratic — Korean memory and one Taiwanese designer — rather than systemic. The survey position has not changed: long global semiconductors remains the most crowded trade in the September fund-manager survey at 53% of respondents.
Central bank watch
Fed funds pricing — implied probabilities by meeting
| Bank | Policy rate | Last move / vote | Next decision (Sydney) | Market pricing | Bias |
|---|---|---|---|---|---|
| RBA | 4.60% | ⭐ +25bp 29 Sep, UNANIMOUS — verified from the Bank's own dated release. Fourth hike of 2026; 100bp this year; highest since 2011. A hold was debated before the vote, disclosed by Bullock at the presser | Tue 3 Nov · 14:30 AEDT · presser 15:30 · with a full SMP | ~43% priced immediately after the decision — minimal change from pre-announcement. ~40bp of further tightening priced in total. ANZ alone among the majors formally calls November (to 4.85%); Standard Chartered sees an extended hold; ING one more, contingent on oil. The 3y at 5.02% embeds +42bp over 4.60% | Tightening, data-dependent |
| Fed | 3.75–4.00% | +25bp 16 Sep, 12–0. Chair Warsh. ⭐ Governor Cook's position is settled on two fronts — the Supreme Court ruled against her removal on 29 June and she spoke publicly as a Governor on 28 September | Thu 29 Oct · 05:00 AEDT (28 Oct 14:00 EDT) · no SEP · 8–9 Dec carries one | ⚠⚠ A 09:35 ET PRE-DATA SNAPSHOT — the settled read is unseen for a SECOND consecutive session. Oct hike 71.2% (unchanged from Monday's post-close read); Dec two-hike modal 58.6%, one-hike 36.3%, no-hike 5.1%. Given the confidence collapse and a 2–4bp front-end rally after this stamp, the settled read should be lower — that is a hypothesis, not an observation | Hawkish, contested internally |
| ECB | DFR 2.50% | +25bp 10 Sep | Fri 30 Oct · 00:15 AEDT · presser 00:45 | ⚠ ~49% October, up from the carried 38.6% (25 Sep) — soft input, corroborated in direction only; the December figure could not be refreshed from 93.3%. The speakers moved it: Demarco “would not exclude a rate hike in October”; Escrivá “interest rates are not yet in restrictive territory”, against Lagarde's “measured” and no second-round wage signs. ⚠ The tone meter reads “markedly less hawkish” but is stamped 28 September — a day before those two spoke, so it is stale relative to the pricing move | Hawkish, split |
| BoJ | 1.25% | +25bp 18 Sep, 7–2 — a 31-year high | Meeting 29–30 Oct, decision Fri 30 Oct — resolved against the Bank's own schedule for a twelfth consecutive edition. Summary of Opinions Thu 1 Oct 08:50 JST | The 2y at 1.97% embeds ~72bp over policy, close to three hikes. Tuesday was the first session in three that did not add to October pricing. December 17–18 | Hiking; read the SoO for the two dissents |
| BoE | 3.75% | Held 17 Sep, 6–3 | Thu 5 Nov · 23:00 AEDT · with an MPR | ~80% November (single source). The 6–3 split argued in public on Tuesday: Mann — “inflation staying above 2% is an equality problem”; Taylor — policy “should not react mechanically to energy price movements if primarily relative-price shocks”. The gilt 2y gave back 3bp of Ramsden's Monday move. The 10y auction cleared 5.38%, the highest borrowing cost since 1999 territory | Hawkish; the doves are arguing back |
| RBNZ | 2.75% | +25bp 2 Sep | Wed 28 Oct · 12:00 AEDT — a Monetary Policy REVIEW, not a full MPS | ⚠ Not reproducible from any tracker. The Kiwi did the work in AUD/NZD on Tuesday, not the Aussie | Removing stimulus |
| BoC | 2.25% | Held 2 Sep — sixth or seventh consecutive depending on source | Thu 29 Oct · 00:45 AEDT | Unconfirmed. Canadian GDP flat m/m against +0.4% prior | On hold |
| Norges · Riksbank | 4.50% · 1.75% | Norges +25bp 24 Sep, all members supported; Riksbank unchanged since 17 June | Norges Thu 5 Nov 20:00 AEDT. ⚠⚠ Riksbank: UNCONFIRMED FOR A FOURTH EDITION — four primary routes failed and one vendor lists a “next” date already in the past | Norges ~40% for another hike within six months. A different source class is now needed for the Riksbank date | Hawkish hold · Tightening bias |
| SNB | 0.00% | Held 24 Sep, fifth consecutive | Thu 10 Dec · 19:30 AEDT | A live hike tail was ~40% on the 25 Sep read. KOF surprised high at 109.1 against 106.0 | Hold, hawkish tilt |
| PBoC | LPR 3.00% / 3.50% | Unchanged, sixteenth month | Tue 20 Oct · 12:15 AEDT | ⚠ The 29 Sep fix is single-source and internally implausible — see §03. Direction only: marginally weaker day-on-day, with spot closing stronger than the fix | Leaning against yuan strength |
| RBI | 5.25% | Held 5 Aug, fourth consecutive | Wed 7 Oct · ~15:30 AEDT | A hike to 5.50% is consensus — the first after a cutting cycle. ⚠ No rate level or consensus published from the calendar vendor proven column-shifted this edition. India is the most oil-sensitive major Asian market and traded like it for a ninth week | Turning hawkish |
| Korea (BoK) · Taiwan | 3.00% · 2.00% | Korea +25bp 27 Aug, second consecutive — the first back-to-back since early 2023 | Korea Thu 22 Oct 12:00 AEDT · Taiwan Thu 17 Dec | Forecasts point to 3.25%. Korea's 3-year yield topped 4.1% | Tightening · Hold |
| Banxico · Brazil · Czech | 6.50% · 13.75% · 3.75% | Banxico held 24 Sep; Brazil cut 25bp 16 Sep; CNB +25bp in June, its first hike since 2022 | Banxico Fri 6 Nov 06:00 AEDT · Brazil ~4–5 Nov (inferred) · CNB Fri 6 Nov 00:30 AEDT | Brazil remains the G20 outlier, easing into a hiking G10. CNB's Michl: the Board will “likely consider keeping rates steady or raising them” in November. ⚠ The peso's Monday −2.89% could not be confirmed as extended | Easing · Hawkish |
| SARB · Indonesia · Turkey | 7.25% · 5.75% · 37.00% | SARB +25bp 23 Sep unanimous | ⚠ All three dates unconfirmed, second consecutive edition | Indonesia stays the watch item — on hold while the JCI fell 1.51% on Monday to a verified 6,147.86, the region's worst tape | Mixed |
Fed detail — the committee is converging on “at least one more” and arguing about whether it stops there. Four speakers after Tuesday's close. Williams (New York, permanent voter) at 18:03 ET: “one more hike likely this year is likely enough”, with inflation expected to ease as the major shocks subside — the most market-relevant line of the day, consistent with the ~71% October pricing but arguing directly against the 58.6% two-hike December bucket. Goolsbee (Chicago, 2027 voter) at 17:58 ET: five and a half years above target is “playing with fire”, questioning looking through supply shocks and warning that expectations of AI productivity gains “could create a high danger of overheating now”. Musalem (St Louis) called the AI capex boom a source of demand pressure without visible productivity relief; Barr (Governor) said “further policy adjustments are likely needed”. Note that the AI-capex-as-inflation argument now appears in two Fed speeches and the RBA's own statement on the same day — that is a cross-bank narrative, not a house view. The 15–16 September minutes release Wednesday 7 October, confirmed from the Fed's own calendar. Core PCE for August lands tonight at 08:30 ET with consensus +0.3% m/m core, headline 3.8% y/y, and personal spending consensus running +0.6% to +0.9% against a +0.2% prior — a nominal-demand signal. The Cleveland nowcast for September core PCE is unchanged at 3.49% y/y, implying headline PCE pushes through 3.9% next month. Watch the revision line: downward revisions to certain non-market prices have been flagged as probable and would flatter the y/y.
RBA detail — what the statement actually cited. The drivers named were higher global energy prices, AI-related demand for technology goods, and domestic capacity pressures, with the Board noting “some of the upside risks to inflation are materialising” and that “growth in aggregate demand needs to remain subdued for a period”. The data trail behind it: August employment +39,500 against +20,000 expected with unemployment up to 4.6%, and August transport spending +2.3% with fuel costs +8.1% on the restored excise. A $750,000 mortgage adds roughly $114 a month from this hike and about $454 a month cumulatively across the four 2026 moves. ⚠ Two of the Bank's own web routes remain broken: the media-releases index now returns a redirect loop rather than the four-month-stale archive it served last week, and the cash-rate statistics page is sixteen months stale at 3.85%. The dated release path works and should be the primary route from here. The forward diary page remains reliable and confirmed the 2–3 November meeting with a 14:30 AEDT decision and a 15:30 press conference.
ECB detail — the pricing moved and the tone meter has not caught up. October pricing refreshed from the carried 38.6% to roughly 49% on an OIS-derived read, with about 12bp of expected move. Two named speakers on Tuesday are why: Demarco said above-expected underlying inflation “could be grounds to act” and that he “would not exclude a rate hike in October”, and Escrivá said rates “are not yet in restrictive territory”. Kažimír asked for “full flexibility”. Lagarde, speaking a day earlier, saw “no signs yet” of second-round wage effects and backed a measured approach. The Spanish flash CPI at 4.9% against 4.6% expected, up from 4.3%, is the datapoint that makes Friday's euro-area flash live — consensus there is 3.7% headline against a 3.2% prior and 2.5% core against 2.4%. The French and Italian flashes print today at 16:45 and 19:00 AEST and the German flash at roughly 22:00.
Regional briefs
United States
The consumer broke and the long bond did not care. Conference Board confidence at 81.9 was a 12-year low and a 7.1-point miss, with Expectations at 63.6 — a third consecutive decline, deep below the 80 recession-signal threshold — and JOLTS openings down to 7.079M on a −256k month. The Fed-relevant detail cuts the other way: 12-month inflation expectations rose to a 6.1% mean and 5.1% median, with the survey window capturing the 16 September hike. The front end rallied ~2bp; the 30-year set a 24-year high at 5.612% intraday. That combination — a demand shock that steepens rather than rallies the curve — is the cleanest statement available that the long end is a supply-and-term-premium problem. Supply corroborates: the 5-year auction's tail was the second-largest on record, the 7-year cleared the highest since April 1993, and Tuesday's 52-week bill cleared 42bp above the prior cycle, all with buybacks running at double size. Equities were small and split — mega-cap tech up, everything broader down, utilities leading on what reads as a defensive rather than a rate-sensitive bid. Government funding runs to 11 December, one day after the December FOMC; there is no September or October cliff. Today: core PCE 22:30 AEST, Q2 GDP third estimate, ADP, Chicago PMI, EIA inventories, four more Fed speakers, and quarter end.
Euro area
A semiconductor melt-up sold into a sovereign-supply shock. The Stoxx 600 traded +0.7% and closed −0.09%; the DAX held +0.10% while the CAC fell 0.53% to an 18-week low. The session's real information was in the primary market: the Italian 10-year cleared 4.58% against 4.10% previously — a 48bp jump — and the UK 10-year gilt 5.38% against 5.16%, both on the same morning with cover flat to lower, and BTP–Bund back through 100bp. Bunds rallied against all of it, which is the safe-haven leg of a European supply problem rather than a growth call. Spain's flash CPI at 4.9% against 4.6% expected, up from 4.3%, is a 30bp beat and a 60bp acceleration three days before the euro-area flash on Friday. On France, the position this desk published yesterday needs correcting: three sources now date the PLF 2027 Conseil des ministres to Wednesday 30 September — today — rather than Thursday 1 October, and the two sources for the later date have both gone offline. The AFT's long-term OAT auction on Thursday 1 October is verified, alongside a Spanish 10-year the same morning. So the “same-day collision” framing published yesterday is probably wrong — and what replaces it is arguably a cleaner setup: France sells long-dated paper into the first full session after the budget lands. Assembly deposit is 6 October by statute.
United Kingdom
The 6–3 split argued in public and the front end gave back Monday's move. Mann framed persistent above-target inflation as “an equality problem”; Taylor argued policy “should not react mechanically to energy price movements if primarily relative-price shocks” — a pointed dovish pushback given energy is the proximate driver at the BoE, the ECB and the RBA simultaneously. Gilt 2s richened 3bp, unwinding part of Ramsden's Monday +8bp. The auction is the number that matters: the 10-year cleared 5.38%, 22bp above its predecessor, into a Budget on 28 October with contested headroom — the OBR's last official figure was £22bn from the November 2025 Budget and the Resolution Foundation's estimate is that it may have shrunk to £8bn. PM Andy Burnham, Chancellor John Healey, both confirmed. August data were mixed: mortgage approvals 55K against 56K, net lending £6.9bn against £6.2bn expected, M4 +0.4%. The FPC statement and minutes land today at 19:30 AEST.
Japan
Read the TOPIX print carefully before concluding anything. The Nikkei fell 0.60% to 65,481.27 and the TOPIX 1.72% to 4,041.13 — a 112bp gap that is an ex-dividend artefact, not breadth deterioration: more than half of TOPIX constituents traded ex-dividend on Japan's fiscal-half record date, and the largest named decliners (the two megabanks, Toyota) are among them. Those are gross moves. JGBs went flat after two sessions of front-led bear flattening — 2y 1.97%, 10y 3.09%, 30y 4.18% — the first session in three that has not added to October pricing, though the 2-year still embeds roughly 72bp over a 1.25% policy rate. The exchange archive posted Monday's row and it is worse than carried: the 28th closed exactly at its low, 1,157.12 points off an intraday high that had been the first print through 67,000. USD/JPY at ~157.4 is about 2.6 yen below the 160 line, with no official commentary. The Summary of Opinions from the 17–18 September meeting publishes tomorrow at 08:50 JST and should be read for the two dissents behind the 7–2 vote; the Q3 Tankan lands the same morning, consensus +25 large manufacturing against +22. ⚠ Japanese single-stock prices remain unsourceable for an eighteenth consecutive edition, so no Japanese relative value is opened.
China & Hong Kong
The mainland rose, Hong Kong fell, and that reversal is the structure for the next week. Shenzhen +0.34% off a nine-month low, CSI 300 +0.10%, Shanghai +0.18%, against HSI −0.48% — an exact inversion of Monday. Mainland driver: accelerated Q4 government bond issuance plans and possible activation of unused local-government borrowing quotas, with Cambricon +2.55% on a day semiconductors fell everywhere else. Hong Kong's: higher oil, a US 10-year above 5.2%, continued Chinese-tech weakness and stated caution ahead of the holiday, with Tencent −1.8%, Xiaomi −2.6% and Meituan −2.0%. Today is the last mainland session before Golden Week: exchange-verified closed 1–7 October, resuming Thursday 8 October. HKEX is shut tomorrow, 1 October, only, and trades from Friday 2 October. That makes Hong Kong the sole venue pricing China for four sessions, with the mainland gapping to whatever has happened when it reopens. Today's PMIs at 11:30 and 11:45 AEST are therefore the last mainland input for a week and a half: the official manufacturing consensus is 50.1 against 49.8 prior — an expansion-threshold cross — with non-manufacturing expected to improve but stay contractionary at 49.3, and the private survey at 51.6 against 51.5. Note the ~170bp gap between the official and private manufacturing surveys, which sample different firm sizes.
Emerging Asia & LatAm
Korea stabilised at the index level while its largest sector kept falling: KOSPI −0.27% after opening −0.66%, with semiconductors “broadly plunged 5%”, Samsung ex-dividend, the won at 1,365, the 3-year yield above 4.1% and ~$3bn of foreign and institutional net selling. Taiwan reopened and gave back only 0.82% with TSMC unchanged — see §04. India is the problem market: Sensex −0.33% and Nifty −0.28% look benign until you note 13 of 16 sectoral indices lower, Nifty IT −1.48%, midcaps −1% and smallcaps −0.8%, more than US$2.17bn of FII selling in September and the Nifty down ~5.7% on the month — week nine after a seven-week losing streak, into an RBI decision on 7 October where a hike to 5.50% is consensus and commentary is starting to frame October as possibly too late given oil. Indonesia remains the watch item, with Monday's verified −1.51% and the exchange having just eliminated its Rp50 minimum share price. Brazil is still easing into a hiking G10 and Mexico is under real currency pressure, though neither could be marked for Tuesday.
Australia & New Zealand
The RBA: hiked, unanimous, and hedged
Cash rate 4.60% from 4.35%, unanimously, the fourth hike of 2026 and the highest since 2011. The decision was fully priced; the language was not. Hawkish in writing: “including increasing the cash rate target further if needed” replacing the conditional “if upside risks materialise”; the dated inflation-return guidance dropped, from “until late 2027” to “a reasonable period”; energy prices described as “much higher” than the August assumptions; “some of the upside risks to inflation are materialising”. Dovish at the microphone: Bullock disclosed that a hold had been debated, said the Board wants to see how four hikes “feed through”, argued the stance is already restrictive, pointed at weak productivity rather than wages, and used the word hope — “what is the hope here, is that this will be restrictive enough… then maybe there doesn't need to be any more interest rate rises.” Those two readings are compatible: the vote was unanimous and a hold was canvassed before it, and the disclosure is the only new information in the event. The market traded the disclosure. The street now splits four ways — ANZ to 4.85% in November; Standard Chartered an extended hold; ING one more contingent on oil; the rest not re-confirmed post-decision — against ~43% priced.
Today at 11:30 AEST the ABS publishes three releases in the same minute: the August monthly CPI indicator, August building approvals, and June-quarter engineering construction. ⚠ The CPI consensus is corroborated rather than verified and should be treated as indicative — roughly 0.4% m/m and 4.0% y/y — because the calendar vendor carrying it was demonstrably column-shifted on two other countries in the same fetch, inverting Previous against Consensus and Actual. Note that a flat 4.0% y/y consensus is itself striking: it implies no expected progress at all on a reading already a full percentage point above the top of the band. Private sector credit consensus 0.5%, building permits −0.9% against −2.0%. The RBA's Financial Stability Review lands tomorrow at 11:30.
The tape: a technology melt-up masking a bank sell-off
ASX 200 8,709.3, +29.6 points, +0.34% — but the composition inverts the headline. Information Technology +4.61% carried the index essentially single-handedly on Codan +23.9% to a record close of $62.73 on blowout guidance, Megaport +9.4% on roughly $1bn of new AI-infrastructure contracts, and Stakk +61.0%. Against that, Financials closed −0.13%, reversing Monday's +1.16%, with the big four down 0.3–0.9%; Utilities −0.82%, Energy −0.72%, Real Estate −0.20%. Materials +1.02% on BHP +0.6%, Rio +0.4% and South32 +2.9%. Breadth 115 advancers to 74 decliners with 11 unchanged. The intraday shape is the tell: the index was down 6 points at 14:50, twenty minutes after the decision, and closed up 29.6 — the entire day's gain was made across and after the press conference. Losers: Karoon −6.8%, Block −2.5%, Charter Hall Retail −2.5%, James Hardie −2.4%. ⚠ Volume, turnover and the A-VIX were not obtainable for a fifth consecutive edition; the A-VIX line should now be re-sourced or formally dropped. ⚠ The SPI quote remains stamped 26/09 for a third consecutive edition and is not a Tuesday mark; even if it were, its basis to cash is December carry, not a signal.
Rates, the currency, and the household channel
ACGB 2y 5.06% (+1bp), 3y 5.02% (−2bp), 10y 5.43% (+2bp) — 3s10s steepened 4bp to 41bp. A front end that rallies while the long end cheapens on the day a central bank hikes and keeps a tightening bias is a dovish-read curve, and it is what fired the house flattener. Premium over the new 4.60% cash rate: 2y +46bp, 3y +42bp, 10y +83bp — so the market still prices roughly 1.7 further hikes; the premium compressed, it did not collapse. AUD/USD ran ~0.7008 into the decision, spiked to 0.7029, broke 0.7000 after the press conference to a Sydney close of 0.6993, then recovered through the US session to ≈0.7018–0.7020. The household channel is where the story gets uncomfortable for the Bank. As at Tuesday's close, exactly one lender had announced a pass-through — Macquarie, the full 25bp, effective 15 October — and none of the big four had moved. Set that against the carried structural fact that 18 lenders repriced fixed rates in September, all four majors by up to 0.48pp: lenders front-ran the hike in fixed and are slow-walking it in variable. The household channel tightened ahead of the Bank, not behind it. A $750,000 loan adds ~$114 a month from this move and ~$454 a month across the four 2026 hikes.
Housing, the China link, and New Zealand
The Cotality release date is now settled — the September Home Value Index publishes around 1 October, closing a standing gap. August was −0.9% nationally, the fifth consecutive monthly fall, with Sydney −1.4%, Melbourne −1.1%, Canberra −1.1%, Brisbane −1.0%, Adelaide and Perth −0.8%, and Darwin the only capital still rising at +0.6%; national median dwelling value $912,885. That monthly run-rate is materially worse than the carried “−3.1% over three months” implies. ⚠ Auction clearance was not obtained this edition; the carried mark is a ten-week-low 50.3% on volumes −22.4% week-on-week, itself flattered by a Melbourne volume collapse. On the China link: iron ore has no 29 September print and sits at $96.92/t, a fifteenth consecutive sub-$100 reading, with the mainland shut 1–7 October — there is no fresh demand signal for roughly a week and a half. One genuine upside surprise is being under-reported: August household spending printed +1.1% m/m and +7.0% y/y against a +0.3% prior on the calendar record — ⚠ though a second source on the same vendor describes spending as “flat in August, ending a three-month growth streak”, and that contradiction is unresolved and flagged rather than adjudicated. New Zealand: the NZX 50 fell 1.06% to 13,683.63, its lowest since 16 September, erasing all of Monday's gain — attributed to global yields and oil rather than to the RBA. Hallenstein Glasson +5.7% on a 49.9% profit rise; energy led the decline. The RBNZ's next decision is 28 October, a Review rather than a full MPS.
| Australia — key data trail | Latest | Prior | Next release (Sydney) |
|---|---|---|---|
| Cash rate | 4.60% | 4.35% | Tue 3 Nov 14:30 AEDT, with a full SMP · ~43% priced |
| Monthly CPI indicator (Aug) | due 11:30 today | 4.0% y/y | ⚠ Consensus ~0.4% m/m / 4.0% y/y — corroborated only, vendor column-shift caveat |
| Quarterly CPI (Q3) | — | — | ⭐ Wed 28 Oct — the decisive release, six days before the meeting |
| Employment · unemployment (Aug) | +39,500 · 4.6% | cons. +20,000 | September labour force: late October |
| Household spending (Aug) | ⚠ +1.1% m/m / +7.0% y/y or flat | +0.3% m/m | ⚠ Two readings from the same vendor contradict each other — unresolved |
| Cotality home values (Aug) | −0.9% m/m, fifth fall | median $912,885 | ⭐ September HVI ~1 October — date now settled |
| Variable-rate pass-through | 1 lender announced | 18 repriced FIXED in Sep | Macquarie full 25bp from 15 Oct; no big four as at Tuesday's close |
| ACGB 3y · 10y · 3s10s | 5.02 · 5.43% · 41bp | 5.04 · 5.41 · 37bp | +42bp of 3y premium over a 4.60% cash rate |
| Iron ore | $96.92/t (28 Sep) | 15th sub-$100 | China shut 1–7 Oct — no demand signal for ~10 days |
| RBA Financial Stability Review | — | — | Thu 1 Oct 11:30 AEST |
House views & tactical framework
V028's confirm trigger fired, and the way it fired is the point. Yesterday this desk published that both legs of the level the view was opened on were false — the S&P's 200-day is approximately 7,209, not the 7,679.54 carried, and the dealer gamma flip is ~7,700, not 7,680 — and then deliberately did not move the trigger, on the stated principle that restating a trigger after a move, on the day you discover your own error, is indistinguishable from arguing away a bad outcome. The index closed Monday 3.26 points above the line. On Tuesday it closed 7,671.01, 8.99 points through it. The confirm has therefore happened on the line exactly as published, and conviction rises from Low to Med.
This is the second time this month a decision not to change something was vindicated within one session — the first was declining to re-enter the 5s30s steepener, which was proved right the following day. It is worth being precise about what it does and does not establish. It does not make the original level analysis correct; that analysis was wrong by 474 points and remains wrong. It establishes only that a pre-committed trigger left alone produces a clean, arguable-by-nobody outcome, while a trigger moved in the heat of a correction produces an outcome no one can score.
The independently-recomputed moving averages, with the plausibility check that was missed last week now run on both: 200-day 7,209.41, 50-day 7,640.56, as at 28 September. The 200-day moved +4.21 points over three sessions, implying entering values ~280 points above those dropping out — plausible. The 50-day moved +4.47, implying ~74.5 points — plausible for mid-July levels. Against Tuesday's close the index is +6.4% above its 200-day and +0.4% above its 50-day. The 50-day at ~7,640 is now the level that matters, not the 200-day — it sits ~30 points below spot, essentially on the ~7,670 put wall, and the 200-day is 6% away and irrelevant on a two-to-four-week horizon.
| Asset | Bias | Conv. | Horizon | Rationale | What changes the view |
|---|---|---|---|---|---|
| Equities | |||||
| S&P 500 (V028) | Own downside convexity, not delta | Med ↑ | 2–4 wk | ⭐ CONFIRMED. Closed 7,671.01 against a 7,680 trigger and a 7,743.41 reference — −0.94%. The mechanism strengthened rather than weakened: net GEX negative on all four sources carrying 29 September data, a second consecutive session in short gamma; and the structure has compressed hard — the call wall has COLLAPSED from the carried 7,800/8,000 dispute to ~7,700 on two sources, against a put wall clustered at ~7,650–7,675 and spot at ~7,671. The index is pinned inside a ~30-point band with hedging that amplifies. Breadth is the second leg and it is verified on both sides: 47.90% of members are above their own 200-day while the index is 6.4% above its, and lows beat highs roughly 10:1 — 450 to 43 across 4,737 US common stocks. Third leg: deep protection is expensive and near-money protection is not — SKEW 146.3 at its 94th all-time percentile against a put/call at the 8th percentile of its range, which is a crash-hedged, drift-complacent book | UNCHANGED AS PUBLISHED: two consecutive closes above 7,800 closes this. The 7,680 confirm has now fired and conviction is Med. ⚠ Three carried inputs did NOT verify and are corrected rather than relied on: the CTA asymmetry is ~18:1 on a 3 September vintage, not 21:1; “vol-control ~86%” is unverified and the sourced metric is a different one; and the buyback blackout at 61% of index weight with no reopening until 1 November is CARRIED AND UNVERIFIED — today is the date it supposedly peaks and no primary confirms it. Do not size the “no corporate bid” argument until it is sourced. ⭐ And the honest counterweight: at 19.2× forward against a 19.8× five-year average, with Q3 earnings growth estimated +29.1% and margins at 15.0%, this market is NOT expensive. Every bearish item here is positioning, flows, breadth or rates — not one is a valuation argument |
| ASX 200 (V006) | Underweight | Med | 2–4 wk | −3.29% from a 9,005.9 entry, and Tuesday's 0.34% gain against the view is less adverse than it looks. The index rose on Information Technology +4.61% — Codan +23.9%, Megaport +9.4% — while every sector this view is short fell: Financials −0.13% (reversing Monday's +1.16%, big four −0.3 to −0.9%), Utilities −0.82%, Real Estate −0.20%. Banks bought into the hike on Monday and sold it on the fact. Structural case intact and sharpened: only ONE lender has announced a variable pass-through and none of the big four, against 18 lenders repricing fixed in September by up to 0.48pp — the household channel tightened ahead of the Bank, not behind it — with home values −0.9% in August, a fifth consecutive fall | The trigger did NOT fire: it reads “an RBA hold 29 September” and the Bank hiked. What remains: a turn in consumer sentiment; banks stabilising as a trend. ⭐ The resolution date is restated and it is NOT the meeting — it is the quarterly CPI on Wednesday 28 October, six days before 3 November, which is what Bullock explicitly pointed at. ⚠ Volume, breadth and the A-VIX unobtainable for a fifth edition; the A-VIX line should be re-sourced or dropped |
| Rates & credit | |||||
| US credit (V017) | UW HY/CCC; prefer 3–5y IG | Med | 1–3 mo | ⭐ MARKABLE AGAIN AFTER THREE DARK SESSIONS — and the answer is that it does not confirm. 24 → 25 → 28 September: IG 79 → 81 → 83, HY 280 → 293 → 302, CCC 1,112 → 1,128 → 1,146bp. The 24 September values match the carried figures exactly on all three series, validating continuity. CCC is four basis points from the confirm level — and IG is four basis points WIDER, not flat, so the second condition fails outright. The CCC/IG ratio compressed from 14.1× to 13.8×: everything gapped together. This is rates beta, not the idiosyncratic low-quality decompression the view is built on — the short leg is paying and the long leg is bleeding. Corroborating driver: the 5-year auction cleared above 5% for the first time since 2007 and municipal 1–3y benchmarks rose up to 48bp on the fastest liquidations since COVID | UNCHANGED: CCC inside 1,050bp with IG unchanged or tighter closes this; through 1,150bp WITH IG FLAT confirms it and conviction goes to High. ⭐ The thing to watch is not whether CCC prints 1,150 — at the run rate it does on the next observation — but whether IG stops widening when it does. If both keep moving together this is a duration trade wearing a credit trade's clothes. ⭐ The SoftBank chase is formally abandoned after a seventh failure on secondary and ANSWERED from the primary side instead: $11.1bn priced 23 September, the largest high-yield corporate sale on record, 7.5-year at 9.75–9.875%, BB+, book >$20bn. A BB+ credit paying 9.875% into a widening tape is the mark |
| OAT–Bund (V025) | Widener | Low | 1–3 mo | ⚠⚠ UNMARKABLE FOR A SECOND CONSECUTIVE EDITION, AND THE ROUTE IS NOW DEAD RATHER THAN LATE — the dedicated same-page series 404s. Last established level 105.4bp (25 September), ≈11bp of widening from a ~94bp entry. Tuesday's legs imply ~6bp more widening (OAT +4.0bp against Bund −1.9bp), i.e. the low 110s and through the 24 September 52-week wide — published as direction only, because mixing legs across pages is the documented artefact that once produced 110.5bp against a true 97bp. ⭐ The substitute evidence is better than the mark would have been: the Italian 10-year auction cleared 48bp above its predecessor and the UK gilt 22bp, on the same morning, with cover flat to lower, and BTP–Bund went back through 100bp. European sovereign supply is being repriced hard into quarter-end | A compression inside 80bp; a credible French consolidation; a dovish ECB October (now ~49% priced, up from 38.6%). ⚠⚠ CORRECTION TO YESTERDAY'S CATALYST WORK: the PLF 2027 Conseil des ministres is now 3-to-2 for WEDNESDAY 30 SEPTEMBER — today — not Thursday 1 October, and both sources for the later date have gone offline. The AFT long-term OAT auction on Thursday 1 October is verified, with a Spanish 10-year the same morning. So the “same-day collision” published yesterday probably does not happen — France instead sells long paper into the first full session after the budget lands, which is arguably a cleaner setup because the auction prices a known budget. ⚠ And the censure price is harder than carried: the RN's stated conditions are four, not one — no tax increases, no “injustice sociale”, an immigration freeze and suspension of the energy programming law — while one source has the PS committed rather than leaning. Both single-sourced |
| FX & commodities | |||||
| AUD/NZD (V023) | Long | Low | 1–2 mo | ⭐⭐ THE FIRST CLEAN MARK ON THIS VIEW IN FOUR EDITIONS. 1.24211, ≈+0.86% from a 1.2315 entry, +0.31% on the session. Cross-computed from two legs of the same dated table per protocol — and the identical construction applied to 28 September gives 1.23824 against the carried 1.2384, a 1.6-pip match, which validates both legs and the method simultaneously. The Kiwi did the work, not the Aussie: NZD −0.27% against AUD +0.03%. The policy gap widened mechanically — the RBA is now at 4.60% against an RBNZ at 2.75% | The trigger did NOT fire: it reads “an RBA hold 29 September” and the Bank hiked. What remains: a hawkish RBNZ on 28 October (a Review, not a full MPS) and a China shock hitting Australia harder. ⚠⚠ Do not mark this off speculative positioning: AUD remains the single largest disagreement in the COT report — leveraged funds net LONG 58,726 against legacy non-commercials net SHORT 46,814, on open interest of 306,488 that matches to the contract across both pages, which is the strongest evidence yet that this is real trader classification and not a data error. GBP is now a second sign disagreement and was not previously flagged |
| Brent (V024) | Residual call spread only — no new risk | Low | 1–3 mo | ≈$102.34–102.70, only ~2% above the $100.60 reference — down from +7.1% yesterday, the largest single-session move toward the re-own band since the view was written. ⚠⚠ THE CARRIED PHYSICAL PICTURE WAS WRONG AND THE CORRECTION CUTS AGAINST US: the Saudi East–West pipeline is at ~3.5 mbd, roughly HALF its 7.0 mbd capacity — not “100% with zero redundancy” — and Red Sea loadings have resumed. “Zero redundancy” is dropped from the house language. Add an SPR release of up to 40 million barrels with the reserve at its lowest since 1982, and a Qatari-mediated proposal with a US response expected today. The 50% datapoint is also the tiebreaker on the restart timelines: a “within days” full restoration needs throughput to double inside a week, which favours the six-week sourcing. What still supports the view: Nov–Dec backwardation above $7 in one month, and observed transits 90–99% below normal on every provider | UNCHANGED AS A CONDITION: two consecutive settles in the $92–95 band re-owns this outright. That band is now ~7% away rather than ~12%. ⚠⚠ Asymmetry warning, restated because it has become more relevant: NYMEX WTI non-commercials are net LONG 141,106 contracts, 7.66% of open interest. De-escalation is a long-liquidation event into a thin book, not a short squeeze. ⚠ And this edition's crude changes rest on a contested base: every 29 September vendor differences off ~$105.3 rather than the $107.71 Monday settle we published. We carry our own prior and flag it |
V004 — ACGB 3s10s flattener — is CLOSED, and the scoring deserves to be argued in the open. The view was opened 7 September at a 43bp entry with the invalidation written as “a dovish RBA with a sticky 10y; a China stimulus impulse steepening the long end”. On Tuesday the RBA delivered a hawkish statement and a dovish press conference, the market traded the press conference, the 3-year rallied 2bp and the 10-year cheapened 2bp: 3s10s steepened 4bp to 41bp. That is literally the named condition, on the named catalyst date. The trigger fired and the view is closed.
It is scored SCRATCH, and here is the case against that. The ledger's rule says a view is wrong when “the trigger fired against the view”, and this trigger did fire. Against that, the ledger's rule for right is that “the bias paid before the trigger fired” — and it did: the spread went 43 → 37bp, six basis points in the money, before giving most of it back, and it is closed two basis points in the money. A two-basis-point move on a one-to-two-month curve view is not a result in either direction. Scratch is the defensible middle, and the reader is entitled to disagree and mark it wrong; the numbers to do so are all here. What is not arguable is that the view was closed on a condition written in advance, on the day that condition was met, rather than being carried because it happened to be marginally profitable. That is the sixth consecutive close fired on a pre-committed condition.
No new view was opened, and three candidates were declined — the eighth, ninth and tenth this month. (1) A 5s30s steepener, declined for a third consecutive edition — and this time the decline cost money. Tuesday was a clean bear steepener: 2s10s re-steepened ~4bp and 5s30s ~2bp on the confidence collapse. The previous two declines were vindicated within a session and this one was not, and that asymmetry should be recorded rather than buried; the reason for declining is unchanged and is the 28 September re-entry convention, but the convention has now been paid for. (2) An outright long ACGB 3-year on the argument that 5.02% embeds ~1.7 further hikes against a Governor signalling she hopes to be done — declined under the same re-entry convention, because it is an adjacent instrument, on the same event, one session after a related view was stopped. (3) A short-dated long-gamma trade around the ~30-point dealer pin — declined as a separate row because it is the expression of V028, which is already a convexity view rather than a delta view, and splitting it into two rows would double-count the same risk. It is folded into V028's rationale instead.
Portfolio-level read. Six views, one closed, none opened; the scorecard goes to 3 right / 12 wrong / 7 scratch across 22 decided. Better than yesterday: V028 is confirmed on a trigger this desk refused to move and its supporting structure has compressed into a genuinely unstable configuration; V023 has its first clean mark in four editions with the method validated to 1.6 pips; V017 is markable again and the answer, while not a confirmation, is diagnostically useful. Worse, and worth saying plainly: V004 was stopped on the event it was designed for; the crude view's physical premise was wrong in a way that helps the price and hurts the thesis; V025 has now lost its marking route entirely rather than merely lagging; and three inputs behind the equity view — the CTA ratio, the vol-control figure and the buyback blackout percentage — did not verify and one of them, the blackout, is dated today. A view with an unverified load-bearing input is a view carrying a risk it has not priced.
These are analytical framings for a professional reader, expressed in the vernacular of a macro desk. They are not personalised investment advice, consider no individual's objectives or circumstances, and carry no position sizing. The “what changes the view” column is the accountability mechanism: every view is logged, triggered and scored in the project's views ledger, including the ones that lose.
Positioning, flows & sentiment
| Indicator | Latest | Observation date | Read |
|---|---|---|---|
| ⚠ What did NOT print for 29 September — reported as “did not print”, never restated as current | |||
| VIX close · VIX3M · IVTS | 16.07 · 18.23 · 0.8815 | 28 Sep | ⚠ The 29 Sep VIX close was unobtainable on five routes; only two morning snapshots exist and neither is publishable. ⭐ But the 28 September vol stack is now fully resolved on four sources, closing last edition's gap: IVTS 0.8815, VIX3M/VIX 1.1344, and the curve is in its 120th CONSECUTIVE DAY of contango |
| SKEW · put/call · % above 200d · highs/lows | 146.3 · 0.52 eq · 47.90% · 43/450 | 28 Sep · 25 Sep · 28 Sep · 28 Sep | All advanced to 28 September; none printed for the 29th. Put/call is four sessions stale |
| CFTC COT · AAII · BofA Flow Show · FMS | — | 22 Sep · 23 Sep · 25 Sep · 15 Sep | All correct and current for their cadence. Next COT Friday 2 October (29 Sep data); AAII tomorrow; Flow Show Friday |
| Lipper weekly flows | ⛔ FORMALLY DROPPED | — | Sixth consecutive failure. Removed from the standing checklist, as the % above the 50-day was |
| CFTC positioning — 22 September data, released 25 September. No newer report exists; all five venue pages carry the same header | |||
| ⚠⚠ Gold — SERIES NAME CORRECTED | +225,853 · 54.71% of OI | 22 Sep | ⚠⚠ MATERIAL CORRECTION: this is the COMEX LEGACY NON-COMMERCIAL cut, not managed money. The numbers are right and reconcile exactly (253,982 − 28,129 on OI of 412,800); the series name this desk has published was wrong. That is precisely the conflation our own standing warning names. Still the most crowded long in the legacy report — and it was trimmed only 4,485 going into a −3.90% session |
| Silver — same cut | +25,444 · 23.90% of OI | 22 Sep | Same correction applies. Less crowded than gold, and it fell harder on Monday and rose less on Tuesday |
| AUD — the largest disagreement in the report | LF +58,726 vs legacy −46,814 | 22 Sep | ⭐ Open interest of 306,488 MATCHES TO THE CONTRACT across both pages — so a 105,540-contract sign disagreement is real trader classification, not a data error. Cross-page OI validation passed on all four FX majors (EUR 821,689, JPY 378,701, GBP 244,925, AUD 306,488) |
| GBP — a second sign disagreement | LF +13,239 vs legacy −82,568 | 22 Sep | Not previously flagged. EUR and JPY agree in sign; GBP and AUD do not. Never publish a positioning claim without naming the series and the venue |
| Leveraged-fund Treasury total | −6,517,822 contracts | 22 Sep | Carried as settled, not re-derived — and the raw file sums to it to the contract across all six CBOT contracts, which is a free validation. This is the squeeze fuel behind Friday's payrolls |
| NYMEX WTI · DXY | +141,106 (7.66% OI) · +10,330 (22.3% OI) | 22 Sep | Crude's downside is long-liquidation risk, not a short squeeze — the stops are below. ⚠ DXY exists only in the legacy ICE cut; it is absent from the TFF page entirely |
| Flows — three universes, labelled separately, never netted | |||
| ICI combined flows | Long-term −$10,098m · Equity −$13,297m · Bond +$3,244m | w/e 16 Sep | The composition is the story, not the total: mutual-fund outflows of $36.70bn against ETF net issuance of +$26.61bn is a substitution, not pure de-risking |
| ICI money market | $7.94tn, +$15.00bn | w/e 23 Sep | ⚠ Do not net this against the BofA cash weights. Money-fund assets are a LEVEL at a record; survey cash is a WEIGHT at a record low. Both are true. Institutional +$15.26bn against retail −$256m is the entire weekly move |
| BofA / EPFR Flow Show | Global equities −$10.2bn · US equities −$21.1bn · Bonds +$17.3bn | 25 Sep | Bonds have now taken inflows for 74 consecutive weeks. Hartnett's base case: “Debasement & Duration” — long commodities and EM, nibble peak-yield beneficiaries into the midterms |
| Sentiment — and the standing divergence has NARROWED on both sides | |||
| AAII | Bulls 32.7% · Bears 48.1% | 23 Sep | ⭐ The direction this desk has not been carrying: bears FELL 5.2pp from 53.3% and bulls ROSE 3.9pp. Retail is still bearish but materially less so than a week earlier |
| BofA Bull & Bear | 9.3 (sell) | 25 Sep | Path 9.7 (28 Aug) → 9.3 (25 Sep). De-escalating from the extreme while staying above the 8.0 sell threshold. Private clients: equities 66.1%, cash 9.4% — a record low |
| BofA Global FMS | Cash 3.9% · net 49% OW equities · net 48% UW bonds | 15 Sep | Confirmed nothing newer exists; October FMS due mid-month. Most crowded trade: long global semiconductors, 53%. Biggest tail risk: a disorderly rise in bond yields — which is the week the market is having |
| ⭐ The divergence, restated | — | — | The carried framing was “retail outright bearish while institutions hold record-low cash”. Both sides moved toward each other in the latest prints — AAII bears 53.3 → 48.1 and Bull & Bear 9.7 → 9.3. Allocation is still the side that has to unwind, because a record-low cash weight is a level rather than momentum, but report it as NARROWING, not as a standing extreme |
| Vol, options and dealer positioning | |||
| MOVE vs VIX | 101.82 vs 16.07 | 28 Sep | ⭐⭐ THE REGIME TAG IN OPTIONS FORM: the market is paying for rates risk and not for equity risk. MOVE tops 59 of the previous 60 readings while equity vol sits in the bottom third of its range in day 120 of contango. Backwardation has occurred on only 7.6% of trading days since 2010 and the last was a single day, 7 April 2026 |
| VIX curve shape | VIX1D 11.27 (−9.91%) · VIX9D 14.39 (+12.77%) | 28 Sep | The curve flattened from the front UP, not the back down — the 9-day is starting to price event risk (core PCE, payrolls) while the 1-day fades. That is the shape that precedes a repricing, not one that follows it |
| CBOE SKEW | 146.3 — 94th ALL-TIME percentile | 28 Sep | Advanced from the carried 144.9 (25 Sep); corroborated independently at 146.25. Quote the 94th all-time percentile, not the 67th one-year: deep downside protection got MORE expensive, not less |
| Put/call ratio | Equity 0.52 · total 0.75 | 25 Sep | ⚠ CORRECTION: the carried “0.55 at 24 September” matches neither series. 9-day average 0.81, 8th percentile, the page's own word for it “extreme complacency”. ⭐ Against SKEW's 94th percentile that is one book: short near-the-money puts, long far out-of-the-money ones. Crash-hedged and drift-complacent — exactly the configuration that makes a gap down worse than a grind down |
| Dealer gamma, 29 Sep | Flip ~7,695 · net GEX negative | 29 Sep | Four of five providers carry 29 September data. Three cluster 7,684–7,706 and all three pass the internal-consistency test (spot below flip with negative GEX); a fourth prints 7,855 with its own call wall 155 points BELOW its own flip, which is incoherent, and is cited but not believed. ⚠⚠ Net GEX magnitude is unreconcilable for a third edition (−$13.7bn / −$24.2bn / −47.9$bn per 1% / −$84K) — publish the SIGN, never the magnitude |
| ⭐ Call wall — the dispute resolved DOWNWARD | ~7,700 (was 7,800 vs 8,000) | 29 Sep | ⭐⭐ THE MOST TRADEABLE FINDING IN THIS SECTION. Two sources at 7,700, one at 7,660, one undated at 8,100. Against a put wall clustered 7,650–7,675 and spot at ~7,671, the index is pinned inside a ~30-point band. Max pain has also resolved — the 7,725 camp is gone, now ~7,695 on two sources. That structure is unstable: a break of either side removes the pin |
| ⚠ CTA · vol-control · blackout | ~18:1 · unverified · 61% UNVERIFIED | 3 Sep · — · — | ⚠⚠ THREE CORRECTIONS. The carried “21:1” cannot be reproduced — the best source gives $163bn down against ~$9bn up, i.e. ~18:1, on a 3 SEPTEMBER vintage. “Vol-control ~86%” is unverified; the sourced metric is a 100th-percentile allocation reading, which is a different object. And the buyback blackout at 61% of index weight with no reopening until 1 November has NO primary source — only a 31 August projection showing 60–70% by late September. ⚠ Today is the date that figure supposedly peaks, and it is the weakest load-bearing input in this note. A separate “88% in blackout” item was rejected on dateline: 30 June 2026. ⭐ And note the most recent sourceable CTA sell-trigger range, 7,349–7,558 (11 Sep), sits 1.5–4.2% BELOW spot — on the last published calibration CTAs are NOT near a mechanical sell trigger, which cuts against the carried framing |
| Breadth — and a material correction to this desk's own figures | |||
| ⚠⚠ New highs / lows — CORRECTED | 43 highs / 450 lows | 28 Sep | ⚠⚠ THE CARRIED FIGURES WERE OVERSTATED BY ROUGHLY 2×. Correct: 43 highs and 450 lows across 4,737 eligible US common stocks; Nasdaq 40/249; S&P 500 4/29. Two independent sources reconcile additively (40 + ~3 = 43; 249 + ~201 = 450). The carried “422/16 NYSE and 461/50 Nasdaq” cannot be reproduced. The direction is unchanged and still stark — lows beat highs roughly 10:1 — but the magnitude was about half what we published. ⚠ The eligible universe is 4,737 today, not 4,733 and not 4,751; it drifts daily and must be re-fetched |
| % of S&P 500 above its 200-day | 47.90% | 28 Sep | ⭐⭐ THE CLEANEST DIVERGENCE IN THE NOTE, verified on both legs: fewer than half the index's members are above their own 200-day while the index itself is 6.4% above its. Advanced from 49.10% — a break below 50% — and ~17pp below its own historical mean of 65.29 |
| McClellan · Hindenburg | −35.25 · 3 of 4 conditions | 28 Sep | The failing condition is “both extremes elevated”: new highs are only 0.91% against a 2.2% requirement. NYSE decliners led advancers 3.55:1 |
| S&P moving averages | 200d 7,209.41 · 50d 7,640.56 | 28 Sep | ⭐ Both pass the plausibility check on the arithmetic of their own window — the check that was missed last week, now run on both (see §08). The second source did not print and is reported as such, not as corroboration. ⚠ The demoted vendor was not consulted |
| Valuation and earnings — the counterweight to everything above | |||
| S&P 500 forward P/E | 19.2× | 25 Sep | ⭐⭐ BELOW its own 5-year average of 19.8× and barely above the 10-year 19.0×, despite the index sitting near a record. Not one bearish item in this section is a valuation argument — they are all positioning, flows, breadth and rates. That asymmetry belongs in the house view explicitly |
| Q3 2026 earnings growth · margin | +29.1% y/y · 15.0% | 25 Sep | Third consecutive quarter above 25%; the second-highest net margin since 2009. Bottom-up 12-month target 9,275.04, ~+21% — the analyst aggregate always runs above strategists, but the gap is worth labelling |
| ⭐ The newest strategist revision | Yardeni 8,400 → 7,900 | 16 Sep | ⭐⭐ THE SEPTEMBER YIELD MOVE DID PRODUCE A TARGET CUT — exactly one, and it came through the MULTIPLE, not earnings. In his own words: “Given the recent backup in bond yields, we are lowering our estimate for the forward P/E… from 19.8 to 18.6, which lowers our year-end target from 8,400 to 7,900.” Verified on two sources by dateline. DERIVED AND DIRECTLY TRADEABLE: his new 18.6× assumption sits BELOW the market's actual 19.2×. At constant forward EPS, de-rating from 19.2× to 18.6× is −3.1%, or ~7,430 from spot. The most recently-revised strategist on the street is implicitly saying the multiple is ~3% too high right now — a cleaner and better-attributed bear case than anything in the positioning stack. Other vintages: HSBC 8,100 (13 Sep), BofA 7,400 year-end / 7,800 12-month (14 Sep); four items rejected on dateline this edition |
The week ahead
| Day | Sydney | ET | Event | Cons. | Prior | Imp. |
|---|---|---|---|---|---|---|
| Wednesday 30 September — AEST = ET + 14h · QUARTER END · China's last session before Golden Week · Canada bank holiday | ||||||
| Wed | 11:30 | Tue 21:30 | AU monthly CPI indicator (Aug) m/m · y/y — ⚠ consensus corroborated only | ~0.4% · ~4.0% | 4.0% y/y | H |
| Wed | 11:30 | Tue 21:30 | AU building approvals m/m · private sector credit m/m | −0.9% · 0.5% | −2.0% · 0.5% | L |
| Wed | 11:30 | Tue 21:30 | CN NBS manufacturing PMI (Sep) — the expansion-threshold cross · non-manufacturing | 50.1 · 49.2–49.3 | 49.8 · 49.0 | H |
| Wed | 11:45 | Tue 21:45 | CN RatingDog (Caixin) manufacturing PMI | 51.6–51.7 | 51.5 | M |
| Wed | 16:00 | 02:00 | DE retail sales m/m · import prices · UK final Q2 GDP q/q · current account | 1.6% · 0.4% | −3.4% · 0.4% | M |
| Wed | 16:45 | 02:45 | FR flash CPI (Sep) m/m · FR consumer spending | −0.5% | +0.7% | H |
| Wed | 17:55 · 18:00 | 03:55 · 04:00 | DE unemployment change · CH UBS economic expectations | 0K | 4K · 12.1 | M |
| Wed | 19:00 | 05:00 | IT flash CPI (Sep) m/m | +0.2% | +0.5% | H |
| Wed | 19:30 | 05:30 | UK FPC statement and minutes · DE 10-year Bund auction (tentative) | 3.39%, b/c 1.5 | M | |
| Wed | ~22:00 | ~08:00 | DE flash CPI (Sep) y/y · m/m — ⚠ one calendar dates this to the 29th and is wrong; three say today | 3.2% · +0.5% | 2.9% · +0.2% | H |
| Wed | 22:15 | 08:15 | US ADP employment change | 73K | 38K | M |
| Wed | 22:30 | 08:30 | US CORE PCE (Aug) m/m · headline y/y · personal income · personal spending | +0.3% · 3.8% · +0.3–0.5% · +0.6–0.9% | +0.2% · 3.7% · +0.4% · +0.2% | H |
| Wed | 22:30 | 08:30 | US Q2 GDP third estimate q/q · GDP price index · goods trade balance | 1.5% · 6.4% | 1.5% · −$118.9bn | M |
| Wed | 23:45 | 09:45 | US Chicago PMI (Sep) — ⚠ two consensus figures circulate (46.9 and 51.2) | 46.9 / 51.2 | 47.1 | M |
| Wed | Thu 00:30 | 10:30 | EIA weekly petroleum (w/e 25 Sep) — crude, gasoline, distillate, Cushing, SPR | −1.9M | +1.786M | H |
| Wed | various | 13:30–18:00 | Fed: Barkin, Cook, Goolsbee, Kashkari · 2y/5y/7y notes SETTLE | M | ||
| Wed | — | — | ⚠ FR PLF 2027 to the Conseil des ministres — NOW 3-to-2 FOR TODAY rather than tomorrow | H | ||
| Thursday 1 October — CHINA SHUT (Golden Week, 1–7 Oct) · Hong Kong shut today only | ||||||
| Thu | 06:30 | Wed 16:30 | MICRON fiscal Q4 results and call (2:30pm Mountain Wed) — the crowded trade's earnings test, landing on the Sydney open | ⚠ not published | H | |
| Thu | 09:50 | Wed 19:50 | BoJ Summary of Opinions (17–18 Sep meeting) — read for the two dissents · JP Tankan Q3 large mfg / non-mfg | 25 / 36 | 22 / 37 | H |
| Thu | 11:30 | Wed 21:30 | AU RBA FINANCIAL STABILITY REVIEW · AU goods trade balance | A$2.00bn | A$1.92bn | M |
| Thu | ~day | — | AU Cotality September Home Value Index — date now settled | −0.9% m/m (Aug) | M | |
| Thu | 16:30–18:30 | 02:30–04:30 | CH CPI · ES / IT / FR / DE / EA / UK final manufacturing PMIs · EA unemployment (Aug) · BoE Bailey speaks | EA mfg 52.7 · UR 6.4% | 52.7 · 6.4% | M |
| Thu | tent. | tent. | ⭐ AFT LONG-TERM OAT AUCTION (French 10y) — verified from the issuer · Spanish 10-year auction the same morning | prior 4.23%, b/c 2.3 | ES prior 3.96% | H |
| Thu | 22:30 | 08:30 | US initial jobless claims · Challenger job cuts | 201K | 197K | M |
| Thu | Fri 00:00 | 10:00 | US ISM MANUFACTURING (Sep) · prices paid · construction spending | 54.8 · 72.9 | 54.6 · 71.1 | H |
| Thu | — | — | US Treasury ANNOUNCES the 3y / 10y / 30y (auctions 6–8 Oct, all settling 15 Oct) — sizes come with the announcement | M | ||
| Friday 2 October — payrolls and the euro-area flash on the same day | ||||||
| Fri | 09:30 | Thu 19:30 | JP Tokyo core CPI (Sep) y/y · unemployment rate | 2.4% | 1.8% | H |
| Fri | 19:00 | 05:00 | ⭐ EURO-AREA FLASH HICP (Sep) y/y · core y/y — after a Spanish flash that beat by 30bp | 3.7% · 2.5% | 3.2% · 2.4% | H |
| Fri | 22:30 | 08:30 | ⭐ US NON-FARM PAYROLLS (Sep) · unemployment rate · average hourly earnings m/m — the +90K vs +98K dispute is RESOLVED at +90K on two sources | +90K · 4.1% · 0.3% | +162K · 4.1% · 0.3% | H |
| Fri | Sat 00:00 | 10:00 | US factory orders · Fed Logan speaks · CFTC COT (29 Sep data) · BofA Flow Show · FactSet Earnings Insight | 0.1% | 0.9% | M |
| ⚠ SUNDAY 4 OCTOBER 03:00 — SYDNEY GOES TO AEDT (UTC+11). Every conversion below is ET + 15h. The US stays on EDT until 1 November. | ||||||
| Mon 5 | 18:15–19:30 · Tue 01:00 | 03:15–04:30 · 10:00 | EA / UK services and composite PMIs · EA PPI · US ISM SERVICES (Sep) | ⚠ withheld | ⚠ withheld | H |
| Tue 6 | 17:00–20:00 · Wed 04:00 | 02:00–05:00 · 13:00 | DE factory orders · EA retail sales · US and CA trade balance · US 3-YEAR NOTE AUCTION | ⚠ withheld | H | |
| Tue 6 | — | — | ⭐ FR PLF 2027 statutory deposit deadline at the Assemblée nationale (LOLF art. 39, first Tuesday of October) — verified three ways | H | ||
| Wed 7 | ~15:30 | ~00:30 | ⭐ RBI RATE DECISION — a hike to 5.50% is consensus, the first after a cutting cycle | 5.50% | 5.25% | H |
| Wed 7 | Thu 04:00 · Thu 05:00 | 13:00 · 14:00 | US 10-YEAR NOTE REOPENING · ⭐ FOMC MINUTES (15–16 September) — date confirmed from the Fed's own calendar | H | ||
| Thu 8 | Fri 04:00 | 13:00 | US 30-YEAR BOND REOPENING · ⭐ CHINESE MARKETS REOPEN after seven sessions shut | H | ||
| Wed 14 | Thu 00:30 | 08:30 | US CPI (September) — two weeks before the FOMC, three days before the 17 October blackout | H | ||
| Later | Wed 28 Oct: AU quarterly CPI — the decisive Australian release. Thu 29 Oct: Fed (05:00 AEDT, no SEP), BoC. Wed 28 Oct: RBNZ Review (12:00 AEDT). Fri 30 Oct: ECB (00:15) and BoJ. Tue 3 Nov: RBA with a full SMP. Thu 5 Nov: BoE with an MPR, Norges. 11 Dec: US government funding expires | H | ||||
⚠ Consensus and prior figures for 5–9 October are deliberately withheld. One calendar terminates at 3 October, and the only source covering that week is one this desk proved column-shifted this edition — it served an “Actual” column populated for events that have not yet occurred, and it printed Tuesday's consumer confidence at 89.4 and JOLTS at 7.271M against actuals of 81.9 and 7.079M. The events, dates and times in that block are confirmed; the numbers are not, and publishing them would import a known error. ⚠ The ≥$4bn Treasury buyback alongside the 8 October 30-year, carried for several editions, is absent from Treasury's own tentative schedule and is not published as fact. Auction times of 13:00 ET are the standard convention, not a verified figure. AEST = ET + 14h to 3 October; AEDT = ET + 15h from 4 October.
Risk radar
| # | Risk | Trigger / timing | Probability | Impact | Cheapest expression |
|---|---|---|---|---|---|
| 1 | Core PCE into a 30-year at a 24-year high, with equity vol in the bottom third of its range | Tonight 22:30 AEST, then payrolls Friday | Core +0.3% cons. | High. MOVE at 101.82 tops 59 of the last 60 readings while the VIX sits near 16 in day 120 of contango. The two markets disagree about how much risk is in the tape and rates has been the one that is right | Short-dated SPX put spreads or 9-day-tenor upside. The front of the vol curve has already started bidding (VIX9D +12.77% against VIX1D −9.91%) while the term structure is still in deep contango — you pay near-trough carry for the one window that matters |
| 2 | The dealer pin is ~30 points wide and it breaks | Continuous; any gap through either side | — | High. Call wall ~7,700, put wall ~7,650–7,675, spot ~7,671, flip ~7,695, max pain ~7,695 — all inside a ~45-point band with net GEX negative on four sources for a second session. In negative gamma, hedging amplifies, and compression makes the release fast whichever way it goes | Long short-dated SPX gamma around 7,670/7,700. This is a volatility trade, not a direction trade — the structure is symmetric and the pin is unstable |
| 3 | The index is being held up by a minority of its members | Continuous; confirmation on any close below ~7,640 | — | High. 47.90% of the S&P is above its own 200-day while the index is 6.4% above its; lows beat highs ~10:1 (450 to 43 across 4,737 issues); McClellan −35.25; 3 of 4 Hindenburg conditions. Narrow leadership breaks disorderly, not gradually | Equal-weight protection rather than cap-weight — it is cheaper and it is where the damage already is |
| 4 | The 50-day at ~7,640 is the real line and it is 30 points away | Continuous | — | Med-High. It coincides with the put wall and sits just under the gamma flip, so a break stacks a technical trigger onto a dealer trigger. The 200-day at 7,209 is 6% away and irrelevant on this horizon | Put spreads 7,640/7,450 — the lower strike is both the secondary max-pain camp and roughly Yardeni's implied de-rate level. Two independent methods land in the same place |
| 5 | Australia's resolution date is 28 October, not 3 November — and two views sit on it | Wed 28 Oct, quarterly CPI; meeting 3 Nov | ~43% priced for Nov | High for the home book. Commentators call November the base case; only ANZ among the majors formally forecasts it and the curve prices less than half of one. The 3y still embeds +42bp over 4.60%, so there is premium to unwind in either direction | Express through the CPI date rather than the meeting. The monthly indicator at 11:30 this morning is the first read into it |
| 6 | Credit's widening is beta, which means the long IG leg keeps bleeding | Any observation after 28 Sep | CCC 1,146bp, 4bp from confirm | Med-High. IG 79 → 83 while CCC 1,112 → 1,146: the ratio compressed. The question is not whether CCC prints 1,150 but whether IG stops when it does | If the two keep moving together, express it in duration rather than credit — it is a duration trade wearing a credit trade's clothes |
| 7 | The most-crowded trade meets its earnings test on the Sydney open | Micron, Thu 1 Oct ~06:30 AEST | Most-crowded 53% (FMS, 15 Sep) | High. The break stopped being simultaneous on Tuesday — Taiwan gave back only 0.82% with TSMC unchanged, Korea recovered half its gap, mainland AI chips rose. Postponed, not resolved | Convexity into the print rather than direction. The crowding is in the delta, so you are paid to take the other side of the consensus |
| 8 | Crude's downside is long-liquidation, and de-escalation is live today | US response to the Qatari proposal expected today | — | Med-High. NYMEX WTI non-commercials net long 141,106 (7.66% of OI). The pipeline at ~50% and an SPR release of up to 40mb have already taken $5 out of Brent; the stops are below | WTI put spreads, or short via options rather than futures. V024's re-own band at $92–95 is now ~7% away rather than ~12% |
| 9 | Payrolls into 6.5 million contracts of leveraged-fund Treasury shorts | Fri 2 Oct 22:30 AEST | Cons. +90K vs +162K prior | Med-High and two-sided. The asymmetry is inverted from normal: a strong print is the bearish equity outcome, a weak one is a violent bond rally into an enormous short | Own convexity, not direction — long strangles in TY/US options. The size on one side of the boat is the argument against picking a side |
| 10 | Europe's sovereign supply is repricing hard into quarter-end, and France taps the long end tomorrow | AFT long-term OAT auction Thu 1 Oct, plus a Spanish 10y | — | Med-High. The Italian 10y cleared 48bp above its predecessor and the gilt 22bp on the same morning, with cover flat to lower; BTP–Bund back through 100bp. ⚠ The budget date has moved to today, so the auction prices a KNOWN budget | The OAT–Bund widener is the existing expression. ⚠⚠ Its marking route is now dead, which is itself a risk to the position |
| 11 | Hong Kong is the only venue pricing China for four sessions | Mainland shut 1–7 Oct, reopens 8 Oct; HK trades from 2 Oct | Mfg PMI cons. 50.1 vs 49.8 | Med. Today's PMIs are the last mainland input for roughly ten days, and the official manufacturing survey is expected to cross 50 for the first time. Any China news between 1 and 7 October transmits through HSI alone, and the mainland gaps to it on the 8th | Express China through Hong Kong for the week, and size the 8 October gap risk explicitly. Iron ore also goes dark |
| 12 | Deep protection is expensive and near-money protection is not — the book is wrong-shaped for a gap | Continuous | SKEW 94th all-time pct | Med-High. SKEW 146.3 at its 94th all-time percentile against a put/call at the 8th percentile on a 9-day average of 0.81. The market is long crash insurance and short the first 5% | ⭐ The cleanest relative-value expression in the note: buy the 5–10% out-of-the-money strip, sell the 20%+ wing. You sell what is at its 94th percentile and buy what is at its 8th |
| 13 | A load-bearing input behind the equity view is unverified, and today is its date | Today, 30 Sep | — | Med, and it is a research risk rather than a market one. The buyback blackout at 61% of index weight and the 1 November reopening have no primary source; the CTA ratio is a 3 September vintage at ~18:1, not 21:1; “vol-control 86%” is unverified | No trade. Do not size the “no corporate bid” argument until it is sourced — and note the last published CTA sell-trigger range sits 1.5–4.2% below spot, which cuts against the forced-seller framing |
| 14 | Gold and duration are still the same trade and the crowd has barely reduced | Continuous; core PCE tonight | 54.71% of open interest | Med. The most crowded long in the legacy COT report was trimmed only 4,485 contracts going into Monday's −3.90% break, and Tuesday recovered only about a third of it | Gold put spreads or long gold vol — the metals complex's implied-vol rank is low in the assets that just moved most |
| 15 | The RBI hikes into an oil shock that may already be easing | Wed 7 Oct | 5.50% is consensus | Med. India is the most oil-sensitive major Asian market: >US$2.17bn of FII selling in September, the Nifty −5.7% on the month, week nine after a seven-week losing streak. Crude fell 5–7% on Tuesday, which changes the calculus late | The equity damage is already done; the risk now is a hawkish surprise into a de-escalating oil market |
Key levels
| Instrument | Last | Support | Resistance | Comment |
|---|---|---|---|---|
| S&P 500 | 7,671.01 | ~7,650–7,675 (put wall) · 7,640.56 (50d) · 7,450 · 7,209.41 (200d) | ~7,695 (flip / max pain) · ~7,700 (call wall) · 7,800 (V028 close) · 7,816.70 (record) | ⭐⭐ THE 7,680 CONFIRM TRIGGER FIRED — it closed 8.99 points through a line this desk refused to move the day its rationale collapsed. The structure is now a ~45-point band containing the put wall, spot, the flip, max pain and the call wall, with net GEX negative on four sources. The 50-day at 7,640.56 is the level that matters; the 200-day at 7,209.41 is 6.4% away and irrelevant on this horizon. Both MAs pass the window-arithmetic check |
| Nasdaq Comp · Russell 2000 | ≈26,796 · 2,809.54 | 26,500 · 2,800 | 26,820.38 · 2,817.91 | ⭐ The Russell is a published close for the first time in three editions and its chain ties exactly. The Composite is derived from the Composite ETF, not from QQQ — the Nasdaq-100 rose 0.20% and the two are not interchangeable |
| UST 30y · 10y | ≈5.59% · ≈5.26% | 5.56 (28 Sep official) · 5.24 | 5.612 (24-YEAR HIGH, intraday) · 5.30 | ⭐ The 30-year's 5.612% is the highest since June 2002 — and it printed with Treasury's long-end buybacks running at double size. The official 28 Sep row settles the previously published 5.53–5.58% range at 5.56% |
| UST 2y · 5y · 2s10s · 5s30s | ≈4.90 · ≈5.07% · ≈36 · ≈52bp | 4.81 · 4.98 · 32 · 50 | 4.92 · 5.10 · 40 · 55 | The front end RALLIED ~2bp on the confidence collapse — a clean bear steepener, exactly reversing Monday. Official 28 Sep: 2y 4.92, 5y 5.06, 2s10s 32bp, 5s30s 50bp. ⚠ The steepener was declined a third time and this time it cost money |
| ACGB 3y · 10y · 3s10s | 5.02 · 5.43% · 41bp | 4.95 · 5.30 · 37 | 5.10 · 5.50 · 43 (V004 entry) | ⭐ 3s10s steepened 4bp and fired V004's invalidation, which was written as “a dovish RBA with a sticky 10y”. The 3y at 5.02% still embeds +42bp over a 4.60% cash rate — roughly 1.7 further hikes. The premium compressed; it did not collapse |
| JGB 2y · 10y · 30y | 1.97 · 3.09 · 4.18% | 1.90 · 3.00 · 4.10 | 2.00 · 3.11 · 4.25 | Flat across the curve — the first session in three that did not add to October BoJ pricing. The 2y embeds ~72bp over policy |
| Bund 10y · OAT–Bund · BTP–Bund | 3.61% · 105.4bp (25 Sep) · ~100bp | 3.50 · 80 (V025 closes) | 3.70 · 109.9 (52-wk wide) | ⚠⚠ NO OAT–BUND MARK FOR A SECOND EDITION, and the dedicated route is now DEAD rather than late. Tuesday's legs imply ~6bp of widening to the low 110s — direction only. The substitute evidence is the auctions: BTP cleared +48bp, gilt +22bp |
| Gilt 2y · 10y · 30y | 4.74 · 5.36 · 5.88% | 4.69 · 5.30 · 5.80 | 4.85 · 5.38 (auction clear) · 6.00 | Gave back part of Ramsden's Monday move as Taylor pushed back publicly. The 10-year auction cleared 5.38%, 22bp above its predecessor, into a 28 October Budget with contested headroom |
| DXY · EUR/USD · USD/JPY | 101.3692 · 1.13619 · 157.39 | 101.20 · 1.1350 · 156.94 | 101.50 · 1.1450 · 160 (policy line) | ⭐ The DXY ties exactly to the verified prior. ⚠ But 39% of its stated move is unexplained by its own weight-summed legs — published, not reconciled away |
| AUD/USD · AUD/NZD | ≈0.7018 · 1.24211 | 0.7000 (broken intraday) · 1.2315 (entry) | 0.7029 (decision high) · 1.2450 | ⭐ The AUD broke 0.7000 on the press conference to a 0.6993 Sydney close and recovered it overnight. AUD/NZD is the cleanest mark in the edition — the same-table method validates to 1.6 pips on the prior session |
| Brent · WTI | ≈$102.70 · ≈$89.32 | 100 · 92–95 (V024 re-own) | 107.71 (Mon settle) · 96.23 | ⚠⚠ READ THE CONTRACT LABEL: Brent rolled to Dec-26 on 27 September and the quote page prints $96.84 — seven dollars below the front month. Nov–Dec backwardation above $7 still says acute but finite. The re-own band is now ~7% away |
| Gold · Silver | $4,174.30 · $61.36 | 4,110.20 (Mon low) · 60.00 | 4,200 · 4,300 (V014 stop) · 62.00 | Recovered about a third of Monday's 3.90% break but still below $4,200 and below its end-August level. Two sources in two currencies agree on the percentage. Gold/silver 68.1, up from 67.7 |
| Copper · Iron ore | $14,544.50/t · $96.92/t | 14,400 · 95 | 14,740 · 100 | ⭐ The 28 Sep official settle posted: cash −1.33%, backwardation narrowed to +$89.50/t and stocks fell only 150t — the destocking has stalled. Iron ore has no 29 Sep print; the streak is fifteen, not sixteen |
| S&P/ASX 200 | 8,709.3 | 8,679.70 · 8,600 | 8,750 · 9,005.9 (V006 entry) | Up 0.34% on IT +4.61% while Financials, Utilities and Real Estate — the sectors the underweight is short — all fell. The index was −6 points at 14:50 and closed +29.6: the whole gain came across and after the press conference. ⚠ Volume, breadth and A-VIX unobtainable, fifth edition; the SPI quote is stamped 26/09 for a third edition and is not a Tuesday mark |
| Nikkei · TOPIX · TAIEX | 65,481.27 · 4,041.13 · 47,631.96 | 65,000 · 4,000 · 47,573 (day low) | 65,877.62 · 4,112 · 48,045.13 (day high) | The TOPIX's 112bp underperformance is ex-dividend, not breadth. ⭐ Taiwan's day high was ABOVE the 24 Sep close — it did not gap down, and TSMC closed unchanged |
| IG / HY / CCC OAS | 83 / 302 / 1,146bp (28 Sep) | CCC 1,050 (closes) | CCC 1,150 (confirms, IF IG flat) | Four basis points from the confirm level on CCC — and IG is 4bp wider, so the second condition fails. Watch whether IG stops when CCC prints 1,150 |
| VIX · MOVE · SKEW | ≈16.03 · 101.82 · 146.3 | 15.50 · 95 · 140 | 16.43 · 105 · 150 | ⚠ No settled 29 Sep VIX on five routes; MOVE and SKEW are 28 Sep. ⭐ MOVE tops 59 of the last 60 readings while the VIX sits in day 120 of contango — the regime in one line. Quote SKEW's 94th all-time percentile, not its 67th one-year |
Data notes & sources
How this edition was built
Six research desks ran in parallel from 06:05 AEST; the verification pass and the US-close block were done last, between 06:30 and 07:10, per the standing rule that at a 06:00 Tuesday-to-Friday start the US bell is minutes old and the authoritative sources have not published. That rule paid this morning. Three of the four things it predicts would be missing were missing — AP's tabulation for 29 September is not indexed, Treasury's official par curve still ends 28 September, and the Fed Rate Monitor served a 09:35 ET pre-data snapshot when checked twice, an hour apart, either side of 17:00 ET. The fourth, the AP tabulation for the previous session, appeared at its dated URL roughly 24 hours late exactly as the method predicts, and it is what closed four carried gaps at once. Asia, Australia and Europe settled hours before filing and are the firm part of this edition; twenty-two index closes were reconstructed as level-minus-change against a verified prior and reported only where the chain tied.
⭐⭐ The 28 September US block is now fully verified, and it settles four carried uncertainties
AP's tabulation, retrieved a day late at its dated URL: S&P 500 7,683.69 (−59.72, −0.8%), Dow 51,481.51 (−347.11), Nasdaq Composite 26,820.38 (−248.34, −0.9%), Russell 2000 2,817.91 (−19.64), VIX 16.07 (+8.1%), corroborated four independent ways. Treasury's official par curve for 28 September: 2y 4.92, 3y 5.01, 5y 5.06, 7y 5.15, 10y 5.24, 20y 5.60, 30y 5.56; 2s10s 32bp, 5s30s 50bp. What that resolves: the 30-year, published yesterday as an unresolved 5.53–5.58% range, is 5.56%; the 2-year basis question resolves at 4.92%, the top of the carried band; the Nasdaq Composite figure this desk REJECTED yesterday as “a 12:16 ET midday snapshot” was in fact the close; and the Russell 2000, unpublished for two editions, is 2,817.91. Two method notes worth keeping: the AP tabulation reliably appears at its dated URL about 24 hours late, so retrieve yesterday's every morning rather than chasing today's; and a rejection on staleness grounds can itself be wrong.
Corrections — sixteen, seven material
(1) MATERIAL, and it explains a five-edition failure: three of the seven European 28 September “verified priors” carried in No. 017 were MIDDAY SNAPSHOTS, not closes. Actual 28 September closes: Stoxx 600 638.68, Euro Stoxx 50 6,301.28 (carried 6,308.85), DAX 25,374.42 (carried 25,513.30 — out by 138.9 points), CAC 40 8,078.48 (carried 8,118.50 — out by 40.0), FTSE 100 10,684.88 (carried 10,749 — out by 64.1), FTSE MIB ~51,760. IBEX and SMI were correct. This is why the Stoxx 600 has been “unobtainable” for four editions: the reconciliation test was working perfectly and the priors were bad, so nothing tied and everything was withheld. Note the pattern — the three wrong priors all carried positive changes and the two correct ones carried negative changes, because Monday's European tape was up ~0.4–0.5% at midday and faded to flat. Every prior must now carry a source time, not just a source. (2) MATERIAL: the 28 September new highs and lows were overstated by roughly 2×. Correct: 43 highs and 450 lows across 4,737 eligible US common stocks (Nasdaq 40/249, S&P 500 4/29), on two independent sources that reconcile additively. The carried “422/16 NYSE and 461/50 Nasdaq” cannot be reproduced. The direction — lows beating highs roughly 10:1 — is unchanged. (3) MATERIAL: the gold and silver crowding figures are the COMEX LEGACY NON-COMMERCIAL cut, not managed money. 225,853 / 54.71% and 23.90% are arithmetically correct on that series; the series name this desk has published is wrong, and it is exactly the conflation our own standing warning names. (4) MATERIAL: the Saudi East–West pipeline is at ~3.5 mbd, roughly 50% of its 7.0 mbd capacity — not “100% of capacity with zero redundancy”. Red Sea loadings have resumed. “Zero redundancy” is withdrawn from the house language, and the 50% datapoint favours the six-week restart sourcing over the “within days” claim, which could not be re-confirmed from any source this run. (5) MATERIAL: the US–China coal commitment is 10 million metric tons in EACH of 2027 and 2028, not “20 million tons” as a single figure. (6) MATERIAL: the French PLF 2027 Conseil des ministres is now 3-to-2 for Wednesday 30 September, not Thursday 1 October, and both sources for the later date have gone offline; the 30th is also the constitutionally normal day. The AFT long-term OAT auction on 1 October is verified. The “same-day collision” published yesterday probably does not happen. (7) MATERIAL: three inputs behind the equity view do not verify. The CTA asymmetry is ~18:1 on a 3 September vintage, not 21:1; “vol-control ~86%” is unverified and the sourced metric is a different one; and the buyback blackout at 61% of index weight with no reopening until 1 November has no primary source at all — only a 31 August projection showing 60–70% by late September. Today is the date it supposedly peaks. (8) The S&P 500's 28 September close is 7,683.69, not the 7,683.26 carried — which does not change the V028 analysis. (9) Jakarta's 28 September close is 6,147.86, −1.51%, not ≈6,151 / −1.45%. (10) The put/call ratio is equity 0.52 / total 0.75 at 25 September, not “0.55 at 24 September”. (11) The Hindenburg eligible universe is 4,737 today, not the 4,733 carried; it drifts daily. (12) Monday's five-source gamma cluster was contaminated — one provider was serving Friday's book under a live page, so the genuine cluster was four sources, not five. (13) The RN's censure price is four conditions, not one: no tax increases, no “injustice sociale”, an immigration freeze and suspension of the energy programming law; and one source has the PS committed rather than leaning. Both single-sourced. (14) Germany's “over €200bn” and €118.73bn are not in conflict — the first includes fully credit-financed Sondervermögen, the second is core federal net new borrowing. The desk should stop treating it as a discrepancy. (15) The FOMC minutes date is confirmed as Wednesday 7 October, not “~6–7 October”. (16) The Supreme Court ruled against Governor Cook's removal on 29 June 2026 — the question closed yesterday on membership pages is closed on the merits too.
Conflicts and how they were resolved
The S&P 500's 29 September close is RESOLVED at 7,671.01 on three convergent routes — a quote page stamped “Closed·15:59:58”, SPY's settled −0.18% implying 7,669.5, and two gamma vendors' near-bell spot at 7,670.04 and 7,671.59. A widely-circulating 7,683.73 / +0.00% comes from a 09:53 ET article presenting morning marks in close-like framing and is rejected; the same piece's Dow figure also misses the settled ETF by 6bp. This mattered more than usual because a house view's trigger sits nine points away. The Nasdaq Composite is resolved in favour of ONEQ's settled −0.09% over the QQQ chain's +0.20%, on internal coherence: QQQ tracks the Nasdaq-100, and a Composite at −0.09% sits correctly between the NDX at +0.20% and the Russell at −0.30% on a day the broad tape fell. The ASX 200 was resolved by arithmetic against its own vendor — the page printed a level of 8,689.70 while its own change (+29.61) and percentage (+0.34%) both imply 8,709.3, which is what three other sources carry. The Italian close is genuinely three-camped (+0.09% / −0.21% / +0.63%) and the published figure is the only one that both states a level and reconciles. The Indian close was resolved by chain against two self-consistent but wrong candidate pairs — a reminder that internal consistency is not verification. Left open, both sides published: the crude change base (every 29 September vendor differences off ~$105.3 rather than our published $107.71 Monday settle); platinum ($1,679.30 on internal arithmetic against ~$1,707 from a second source); net GEX magnitude for a third edition; the Chicago PMI consensus (46.9 against 51.2); and Australian August household spending, where the same vendor reports both “+1.1% m/m” and “flat, ending a three-month growth streak”.
Rejected outright
A 09:53 ET article's “29 September” index set, presented in close-like framing. A calendar vendor's entire Actual column — it was populated for events that have not yet occurred (30 September through 9 October) and printed Tuesday's consumer confidence at 89.4 and JOLTS at 7.271M against actuals of 81.9 and 7.079M; its first numeric column is the prior and its second is the consensus, and the same column-shift appears on the Australian and Chinese rows. A quote page serving Monday's Dow and Nasdaq under a “Closed” header. A Hang Seng Tech quote page serving a 28 September intraday against a 25 September prior. A German portal path that serves the STOXX Europe 600 Health Care sub-index rather than the headline index. An AUD/USD page presenting the post-decision intraday spike as a “previous close” under pre-decision commentary. A Euro Stoxx 50 quote stamped 15:45 CET on a day every index faded into the bell. An oil-chart page serving ~$43 crude — roughly 2020 levels — recommended for the accuracy blacklist. A crypto aggregator's 24-hour percentages, on a probable third sign inversion. Vendor change columns, ninth consecutive edition, including a sign error on the BTP 10-year and a 12bp error on the gilt 2-year. Four sell-side items on dateline grounds, and four geopolitical items dated 11 March 2026, 23 March 2026, 28 August 2026 and 8–9 September 2026 surfacing on searches for 29 September — one of which would have put a six-month-old 9% crude crash into this note.
Not published
The 29 September VIX close, VIX3M, IVTS and contango count (five routes). The Euro Stoxx 50 close. Any OAT–Bund mark, for a second edition — and the dedicated route is now dead, not merely lagging. Hang Seng Tech, on two routes. The 29 September Treasury par row (due ~08:00 AEST today). A settled Fed Rate Monitor read, for a second consecutive session. ASX 200 volume, turnover and the A-VIX, fifth consecutive edition — this line should now be re-sourced or formally dropped. Auction clearance rates. NZD crosses and any AUD cross other than AUD/NZD. The 29 September PBoC fix as a level. USD/MXN and USD/KRW for 29 September (both pages stamped the 28th). Canada's 10-year for the 29th. Iron ore and uranium for the 29th. The 29 September Brent forward curve (page frozen). Any 29 September credit observation. Consensus and prior figures for 5–9 October, deliberately. Indian breadth and the India VIX. European bank-sector performance for Tuesday. Japanese single-stock prices, an eighteenth consecutive edition. SoftBank secondary levels, a seventh consecutive failure — the chase is formally abandoned and answered from the primary side instead. Lipper weekly flows, a sixth failure — formally dropped.
Traps caught
Our own European priors, caught only because a desk questioned the brief it was given rather than the sources it was pointed at — the most valuable catch of the run. The contract-month trap, live and confirmed: Brent rolled to Dec-26 on 27 September and the quote page prints $96.84 against a front month seven dollars higher. The vendor calendar column-shift, caught because a carried figure existed to check the China PMI rows against — and flagged as unverifiable on the Australian CPI rows, where no carried figure exists. A level field contradicting its own change and percentage on the ASX. Two self-consistent but wrong Indian close pairs. The ex-dividend artefact behind a 112bp Nikkei/TOPIX gap that would otherwise read as a breadth break. The static-boilerplate trap on a Hormuz tracker whose header reads “~21 tankers/day” while its own chart shows 0–3. A gamma provider serving Friday's book under a live page, which retrospectively contaminated Monday's cluster. An intraday ETF capture masquerading as a settle, caught by reading each page's own stamp individually. And Maysan (Iraq) and Mayun (Yemen) held apart for a tenth consecutive edition.
Tomorrow's first verification targets
The Australian monthly CPI actual at 11:30 this morning and the 3s10s reaction — the consensus behind it is corroborated only, from a vendor proven column-shifted elsewhere in the same fetch. Then: Treasury's official 29 September par row, due ~08:00 AEST today; AP's 29 September tabulation at its dated URL, which should appear this morning and will confirm or correct the derived Dow and Composite; the 29 September VIX and VIX3M once the official series posts; any IG/HY/CCC observation after 28 September — the single highest-value datapoint of the week, because it tests whether IG stops widening when CCC crosses 1,150; a settled Fed Rate Monitor read, now unseen for two sessions; a replacement route for the OAT–Bund spread, which is a live gap on a day the house is long the widener; the buyback blackout percentage from a primary source, which is the weakest load-bearing input in this note; an A-VIX source or a formal decision to drop the line; the Nikkei's 29 September exchange row; Hang Seng Tech on any route; and the French PLF date against a government communiqué, since the balance of evidence moved by one day overnight.
United States, the Fed and positioning
- US Treasury — daily par yield curve, September 2026
- AP — how major US stock indexes fared, 28 September
- stockanalysis — SPY · DIA · QQQ · ONEQ (Nasdaq Composite) · VTWO
- Investing.com — S&P 500 · Russell 2000 · VIX
- Investing.com — Fed Rate Monitor
- Conference Board — consumer confidence, September
- BLS — JOLTS · BEA — release schedule (core PCE)
- Cleveland Fed — inflation nowcasting
- Federal Reserve — FOMC calendar · 2026 speeches
- US Treasury — tentative auction schedule · MarketScreener — 30-year at a 24-year high
- TFTC — the 5-year auction tail · investingLive · Newsquawk
- CFTC — Traders in Financial Futures · COMEX legacy · NYMEX · ICE (dollar index)
- AAII sentiment survey · ICI combined flows · ICI money market
- CBOE SKEW · VIX term structure · Hindenburg / breadth · % above the 200-day
- WallStreetNumbers — S&P moving averages · gexmetrix · zerogex · quantwheel
- FactSet — Earnings Insight, 25 September · Yardeni cuts to 7,900
Australia, New Zealand and Asia
- RBA — Monetary Policy Decision, 29 September · RBA — coming up
- Market Index — evening wrap, 29 September · ASX 200 live, 14:50 AEST
- ABC — markets live, 29 September · Canstar — RBA reaction
- ActionForex — Bullock's press conference · investingLive — RBA decision
- Finder — lender pass-through tracker · Cotality HVI tracker
- ABS — forward release calendar · Trading Economics — ACGB curve
- Nikkei Indexes — official archive · Trading Economics — Japan · BoJ — meeting schedule
- Trading Economics — Hong Kong · China · China calendar · Shenzhen — trading calendar
- Seoul Economic Daily — KOSPI close · Focus Taiwan — TAIEX close · The Federal — Sensex and Nifty
- Trading Economics — NZX 50 · interest.co.nz — NZX detail
Europe, geopolitics and the calendar
- Hargreaves Lansdown — Europe close, 28 September
- MarketScreener — European close, 29 September · invezz — European close
- dpa-AFX — XETRA close · AWP — Swiss close · ANSA — Milan close
- ForexFactory — economic calendar · Agence France Trésor · LegiWatch — PLF 2027 calendar
- Bundestag — the 2027 budget timetable · House of Lords Library — UK fiscal outlook
- Econostream — ECB speak · Bank of England — MPC dates
- Windward — Hormuz vessel transits · TankerMap — Hormuz tanker transits
- Rigzone — Saudi pipeline restored to half capacity · anews — Qatar to present new ideas
- ABC News — Iran live updates · Ukrinform — latest news · Axios — the US–China tariff deal
Commodities, credit and digital assets
- OilPrice — Brent futures curve · Trading Economics — Brent · Buckhead Energy — WTI and Henry Hub · Investing.com — Brent (contract label)
- Kitco — spot gold · Gold Stock Canada — CAD cross-check · Trading Economics — silver
- Westmetall — LME copper official settlements · Trading Economics — iron ore · lithium
- FRED — IG OAS · HY OAS · CCC OAS · Convex Trade — HY corroboration
- SoftBank — $11.1bn bond sale · EU gas storage tracker
- CoinGecko · CoinDesk · Farside — BTC ETF flows · Coinalyze — derivatives · Polymarket — CLARITY Act