Closing Briefing — Tuesday, September 8, 2026

1. A failed trial in Basel repriced an American index sector. Novartis's pelacarsen missed the primary composite endpoint — cardiovascular death, non-fatal myocardial infarction, non-fatal stroke and urgent revascularisation — in an 8,323-patient Phase III run with Ionis, having lowered Lp(a) as designed. The market did not treat it as a single-asset failure. Novartis fell 14% to $137.63 and Amgen 10.08%, on the read-across to olpasiran, with BMO cutting Amgen to Market Perform the same morning. Citi: "the Lp(a) hypothesis has been weakened" — the open question being mechanism, trial design or the underlying biology. The tell that this was a class event rather than a company event is that Amgen published positive Phase 3 DeLLphi-305 survival data in small-cell lung cancer the same morning and still had the worst session of the three.

2. The Swiss market is the cleanest cross-check available. SMI -1.55% against DAX 0.00%, CAC +0.14%, Euro Stoxx 50 +0.14%, FTSE 100 -0.10%. One index in Europe fell more than a tenth of a percent, and it is the one with the largest pharmaceutical weight. Europe closed the Lp(a) trade before New York opened, and U.S. healthcare then delivered -2.54%, four times the next-worst Finviz group.

3. Nvidia's non-participation, now with a sign change. NVDA -2.01% on a day the SOX rose 1.30%, after Friday's +0.83% against a +3.38% index. Two consecutive sessions in which the semiconductor complex has moved without its largest member, and Tuesday's version is stronger evidence because the two moved in opposite directions. Broadcom +2.98% did participate, so the dispersion is inside the megacap pair as well. This matters directly to the position in Section 12 item 5, whose invalidation requires two consecutive up sessions of index outperformance — the Nasdaq 100 fell, so the clock did not advance.

4. Intel, and a pricing decision doing the work of a product cycle. INTC +9.05% to $104.47 on 135.81m shares, the heaviest volume on the board. Three things landed together: a reported 10% increase in PC CPU prices from early October, Northland's upgrade to Outperform with a $120 target citing server-CPU shortage and the Terafab collaboration, and an Intel Foundry-ASML milestone of more than one million wafers processed on High-NA EUV. A supply-constrained incumbent raising price is a margin event that needs no unit growth, which is why the move came without a revenue revision.

5. Medical devices went with pharmaceuticals, and that is a separate problem. Stryker -8.81%, Boston Scientific -5.90%, Intuitive Surgical -4.51%, Edwards Lifesciences -3.54%, Zimmer Biomet -3.93%, Baxter -3.76%, Becton Dickinson -3.05%, Abbott -2.59%, Medtronic -1.89%. The Lp(a) failure has no device read-across; what these names share is a cyber overhang — Boston Scientific disclosed a breach on 25 August that disrupted global order processing, and Stryker's own March incident cost roughly 5% of a quarter's revenue and took about six months to normalise. A sector that sells off together on unrelated causes is a positioning unwind, and it is the more tradable half of the healthcare move.

6. Crude at a six-week high, and the equity finally followed. WTI +1.89% to $94.27 and Brent +2.34% to $99.27 on Houthi attacks on Saudi facilities, with energy the best Finviz group at +1.18% and the year-to-date lead extended to +40.58%. For most of the past fortnight this report has flagged energy equities refusing to follow the barrel; on Tuesday they did, and the barrel is now $3.00 higher than Friday's WTI settle, a 3.05% two-session move.

7. The belly has a hump the vendor tape does not show. On the official par curve the three-year richened 1 bp to 4.44% while the five-year and seven-year each cheapened 3 bp and the 10-year 2 bp. WSJ's 5:00 p.m. bond board marks the three-year the other way, 4.463% and 1.0 bp cheaper. Both cannot be right about direction, and the gap is the ninety minutes between Treasury's 3:30 p.m. bid-side snapshot and the 5:00 p.m. quotes. Section 6 publishes the official row and names the conflict rather than smoothing it.

8. The travel complex broke without a data point. Expedia -7.88%, Booking -6.72%, Marriott -2.29%, United Airlines -2.83%, Delta -1.51%, Carnival -1.32%. No earnings, no guidance and no macro release; what changed is the price of jet fuel's input, with crude at a six-week high on Middle East supply risk. Consumer cyclical closed -0.65% with the travel names doing most of the damage.

9. Financials fell 1.22% on a session the front end did nothing. The two-year cheapened 2 bp and the three-year richened 1, so this was not a rate move. JPMorgan -1.43%, Wells Fargo -2.23%, Visa -1.71%, Mastercard -1.44%, with the insurance and data complex worse — Verisk -5.54%, Brown & Brown -4.54%, Arthur J. Gallagher -4.28%, Fidelity National Information -5.90%. Financials are the second-largest contributor to the Dow's 626 points after healthcare, and the group is now +7.94% on the year against a Dow that has spent the window making highs.

10. The yen made its move while New York was shut. USD/JPY 153.729, a seven-month high for the currency, 1.60% stronger than Friday's close but only 0.41% stronger on Tuesday's own session — roughly three-quarters of the move was made in Monday's holiday session. On Friday this report wrote that Asia had bought a softer American rate path the payroll destroyed and that the correction could not happen in New York until Tuesday. The correction happened; it went the other way, driven by Japanese data and Bank of Japan expectations rather than by American rates, and 10-year JGBs richened 2 bp to 2.88% while it did.

Source: Finviz group screener, Performance table view (g=sector&v=140&o=name), read after the close. 1-Day is the Change % column, 1-Week Perf Week, YTD Perf YTD. Finviz classification, not GICS.

YTD reconciliation, and it is the cleanest of the reporting window by a clear margin. Compounding each group's 4 September YTD by Tuesday's one-day move reproduces the published YTD to within 0.03 percentage points for all eleven groups, with seven at 0.01 or better. Worked examples: healthcare 1.1021 × 0.9746 = 1.07411, or +7.41% against a published +7.41%, deviation 0.000 pp; energy 1.3896 × 1.0118 = 1.40600 → +40.60% against +40.58%, deviation 0.02; technology 1.2589 × 1.0020 = 1.26142 → +26.14% against +26.13%, deviation 0.01. The single widest case is basic materials at 0.03 pp — 1.1999 × 1.0001 implies +20.00% against a published +20.03% — on a group that moved one basis point, so the deviation is rounding rather than drift. No group requires a carry-forward flag, and the three that carried flags earlier in the window have now reconciled for two consecutive sessions.

The rotation is defensive-into-cyclical, which is the wrong way round for a down day. Energy, utilities, industrials, technology and basic materials all closed green; the four weakest were healthcare, financials, consumer cyclical and consumer defensive. A tape in which the two classic defensive groups are among the four worst while the index falls is not risk-off — it is a sector-specific liquidation with the rest of the market carrying on. Healthcare's -2.54% is four times communication services' -0.31%, the next-worst non-financial group, and it is the largest single-group gap this report has recorded.

The composition traps to name, and two of them invert Friday's. Technology at +0.20% again understates the semiconductor complex, where SOX rose 1.30% and Intel added 9.05%, because the software leg was sold for a second consecutive session — ServiceNow -4.99%, Salesforce -3.86%, Intuit -4.14%, Adobe -3.47%, Accenture -4.12%, PTC -5.51%, Paycom -5.40%, Tyler Technologies -4.45%, Cadence -2.93%. Three consecutive sessions in which the Finviz technology print points the wrong way about chips is now a pattern, not an artefact. Financial at -1.22% conceals a split: the money-centre banks were merely soft — JPMorgan -1.43%, Wells Fargo -2.23%, Bank of America -0.46%, Goldman -0.20%, Morgan Stanley -0.68% — while the insurance-brokerage and data cohort was routed, with Verisk -5.54%, Brown & Brown -4.54% and Arthur J. Gallagher -4.28%. Consumer defensive at -0.61% hides a packaged-food bid, where the nine-name basket of Section 12 item 4 averaged +0.29% on strength in Bunge, Campbell's and Tyson — the first session of the window in which the sub-industry has outperformed its own group, by 0.90 percentage points.

Higher — semiconductors, power and the barrel.

• Intel (INTC) +9.05% to $104.47 on 135.81m shares, the heaviest volume on the board. Reported 10% PC CPU price increase from early October, Northland upgrade to Outperform, target $120, and an Intel Foundry-ASML milestone of over one million High-NA EUV wafers processed.

• Hewlett Packard Enterprise (HPE) +7.78% to $56.05, Corning (GLW) +7.56%, Seagate (STX) +6.47% to $904.22, AMD +5.90% to $505.74, Teradyne +4.21%, Lam Research +4.15%, Applied Materials +3.98%, Qualcomm +3.18%, Broadcom +2.98%, Western Digital +2.10%, KLA +1.82%, Super Micro +1.69%.

• Qualcomm (QCOM) +3.18% to $174.11 on an announced AWS collaboration for custom AI chips — the day's only megacap-scale design win, and the reason the semiconductor complex led on a session its largest member fell.

• Enphase +6.79%, First Solar +4.30%, Freeport-McMoRan +5.35% to $76.62 on a copper session that added 0.60%, Edison International +4.51%, PG&E +3.60% on 65.55m shares, GE Vernova +3.12% to $971.31, Valero +3.27%, Xylem +2.95%, Eaton +2.75%, Mosaic +2.63%, Albemarle +2.61%.

• Tesla (TSLA) +3.98% to $368.16 on 49.72m shares, recovering most of Friday's 5.96% Cybercab decline. Lululemon (LULU) +2.56% to $103.19, a partial bounce off Friday's 17.39% guidance-cut rout. Oracle (ORCL) +2.36% to $162.53 into Thursday's results.

• Bunge +4.36%, Campbell's +1.78%, Tyson +1.67% — the packaged-food complex bid for the first time in the window. Caterpillar +1.05%, Lockheed Martin +2.07%, UnitedHealth +0.93% as the sole healthcare megacap higher.

Lower — pharmaceuticals, devices, travel and the insurance-data cohort.

• Howmet Aerospace (HWM) -10.70% to $231.53 on 10.43m shares, the worst S&P 500 performer, extending the de-rating that began when SpaceX said it would produce turbine blades and vanes in-house for data-centre gas turbines. Howmet holds over half the industrial-gas-turbine blade casting market, supply agreements running to 2030 and a planned 38% capacity expansion; Citigroup keeps a Buy at $329 and Bernstein an Outperform at $328. The CEO presents at the Jefferies Global Industrials conference on Wednesday.

• Amgen (AMGN) -10.08% to $393.17 — full detail in Section 2 item 1. Stryker (SYK) -8.81%, Boston Scientific -5.90%, Intuitive Surgical -4.51%, Zimmer Biomet -3.93%, Baxter -3.76%, Edwards Lifesciences -3.54%, Becton Dickinson -3.05%, Abbott -2.59%, Medtronic -1.89%.

• Bristol-Myers -3.19%, Vertex -3.15%, Gilead -2.88%, Pfizer -2.32%, Eli Lilly -2.21%, Merck -1.24%, Thermo Fisher -1.75%, Danaher -1.17% — the read-across ran through the whole group, not only the Lp(a) names.

• GoDaddy -8.32%, Expedia -7.88% to $274.55, Gartner -7.42%, Shopify -7.57%, Booking -6.72% to $180.30 on 10.18m shares.

• Fidelity National Information -5.90%, Verisk -5.54%, PTC -5.51%, Paycom -5.40%, DoorDash -5.33%, CDW -5.18%, ServiceNow -4.99% to $134.21, Builders FirstSource -4.88%, Aptiv -4.63%, Brown & Brown -4.54%, HP Inc -4.50%, Tyler Technologies -4.45%, ON Semiconductor -4.44%, CarMax -4.30%, FactSet -4.29%, Arthur J. Gallagher -4.28%, Intuit -4.14%, Accenture -4.12%.

• Constellation Brands -5.62% and Dollar Tree -5.62% are the consumer names that did not participate in the defensive bid.

• Nvidia (NVDA) -2.01% to $225.73, Apple -1.17%, Microsoft -1.15%, Amazon -0.60%, Meta -0.53%, Alphabet -0.03%, Netflix -1.89% — five of the six largest weights fell, which is how a 1.30% semiconductor rally sits alongside a 0.58% index decline.

• General Motors -2.24%, Home Depot -2.29%, Starbucks -2.36%, Kroger -2.37% into Friday's results, Lowe's -1.79%, Nucor -1.79%, Walmart -1.02%.

Analyst actions, 8 September.

• Upgrades. Northland took Intel (INTC) to Outperform from Market Perform, target $120, the day's most consequential call and the one the tape paid. Citigroup upgraded Old Dominion (ODFL) and C.H. Robinson (CHRW) to Buy, framing the recent pullback as "an opportunity to buy quality at a reasonable price." JPMorgan raised Cenovus (CVE) and Magnolia Oil & Gas (MGY) to Overweight on valuation and asset quality, into a session crude added 1.89%.

• Downgrades. BMO Capital cut Amgen (AMGN) to Market Perform on risk-reward, arriving the same morning as the Novartis read-across and amplifying a 10.08% decline. BMO also moved BioNTech to Market Perform. JPMorgan cut Gouverneur/Gulfport (GPOR) to Underweight on natural-gas fundamentals, on a day the October contract fell 2.95%.

• Standing calls restated. Citigroup maintains Buy on Howmet at $329 and Bernstein Outperform at $328 — roughly 34% and 33% above Tuesday's $231.53 close — against a stock that has now fallen 10.70% in a session and roughly 13% in a month on a competitive threat neither firm regards as material.

• Percentage upside is asserted only where a same-session closing price was independently captured. For the non-S&P-500 names the target changes are stated as published. See Data Notes.

Sourcing, disclosed. The Earnings Whispers day pages remain behind a cookie-and-usage-agreement consent banner, which this unattended session did not accept. The rosters below are captured from the Nasdaq earnings calendar API for each date and screened name by name against an S&P 500 constituent list. Nasdaq publishes a before-open or after-close bucket rather than a clock time, so no clock times are asserted this session; confirm every time against company investor relations before trading a date.

Wed 9/9. AMC: Cooper Companies (COO).

Thu 9/10. AMC: Oracle (ORCL), Adobe (ADBE), Copart (CPRT).

Fri 9/11. BMO: Kroger (KR).

Mon 9/14. No S&P 500 reporter on either bucket.

Tue 9/15. No S&P 500 reporter on either bucket.

Wed 9/16. AMC: Lennar (LEN).

Thu 9/17. No S&P 500 reporter on either bucket.

Fri 9/18. No S&P 500 reporter on either bucket.

Changes vs. the prior calendar (9/4 report):

• Oracle's move to Thursday 10 September is now confirmed by a second agreeing capture. The 9/4 report placed Oracle on 9/10 after the close for the first time, having carried it on 9/8 for four consecutive captures, and held the 9/8 slot open under the two-capture rule. This capture repeats 9/10, after-close, and no S&P 500 name reported on 9/8. The slot is now recorded as vacated and the re-dating as a publisher correction rather than a company re-schedule.

• Adobe (ADBE) and Copart (CPRT) unchanged on 9/10 after the close for a fourth consecutive capture; Cooper Companies (COO) unchanged on 9/9 after the close; Kroger (KR) unchanged on 9/11 before the open.

• One addition, and it is next week's only name: Lennar (LEN) on 9/16 after the close. The prior calendar's horizon ended at 9/11, so this is a first appearance rather than a change. LEN.B appears on the same date and is deduped as a dual listing.

• Four of next week's five sessions carry no S&P 500 reporter at all. That is the emptiest forward week of the reporting window.

• No removals.

• Non-members on the covered dates, listed so nobody mistakes their absence for an omission: Chewy (CHWY), SailPoint (SAIL), Core & Main (CNM), AeroVironment (AVAV), Korn Ferry (KFY), Signet (SIG), American Eagle (AEO), Academy Sports (ASO) and Caleres (CAL) on 9/9; Descartes (DSGX), Macy's (M), RH, National Beverage (FIZZ), Hub Group (HUBG), Zumiez (ZUMZ), Designer Brands (DBI), Lovesac (LOVE) and 1-800-Flowers (FLWS) on 9/10; Hooker Furnishings (HOFT), Rent the Runway (RENT), Children's Place (PLCE) and MoneyHero (MNY) on 9/11; Grifols (GRFS), Apartment Investment (AIV), Dave & Buster's (PLAY) and Hain Celestial (HAIN) on 9/14; Trip.com (TCOM) and Vera Bradley (VRA) on 9/15; Manchester United (MANU) and Cracker Barrel (CBRL) on 9/16. Borderline membership cases are listed in Data Notes and conservatively excluded.

• What the forward calendar hands the desk. Everything the next fortnight has is in the next seventy-two hours, and then there is nothing. Thursday 10 September after the close carries Oracle, Adobe and Copart into one overnight window — the month's most important artificial-intelligence capital-expenditure disclosure, a software franchise ten days into a new chief executive, and an auction operator — with Kroger the following morning as the only consumer read. After that the calendar hands the desk four blank sessions out of five next week and a single homebuilder on the Wednesday the FOMC begins. Two consequences follow. The first is that the Thursday-evening reaction has no single-name follow-through to trade into: whatever the three prints do to the artificial-intelligence complex has to be expressed against macro and positioning rather than against the next reporter. The second is that Lennar lands on 9/16 alongside the first day of the meeting and the retail-sales print, so the one earnings event of next week is also the least likely to be read on its own terms. The reaction function to carry is Section 4's: guidance was punished 17.39% on Friday and bought back 2.56% on Tuesday, which is a market with no settled view on how to price a cut.

Source: U.S. Department of the Treasury Daily Treasury Par Yield Curve Rates, Text View for September 2026, read after publication on the evening of 8 September. Changes are versus the 4 September official row (1-day, the prior completed session) and the 1 September official row (1-week).

The read: a humped belly with the wings pinned, and the diagnostic is that this was not a policy-repricing session at all. The three-year richened a basis point to 4.44% while the five-year and seven-year each cheapened three — a four-basis-point spread between adjacent points that ordinarily move together, and it happened on a day the two-year moved 2 bp and the September FOMC probability moved zero. The classic explanations do not fit. It is not a policy-timing move, because the tenors that span the September and October meetings barely moved and the strip did not reprice. It is not a term-premium move, because the 20-year and 30-year each cheapened only 1 bp and the curve's long end is flatter on the week. What is left is supply and positioning in the five-to-seven-year sector, on the session before Treasury's expanded buyback programme begins, and the buyback's own maturity buckets are the reason to take that seriously — see below.

2s10s was unchanged at +41 bp for the first time in the window, 2s30s flattened 1 bp to +86 bp and 3M10Y flattened 1 bp to +86 bp. On a one-week view the shape is the mirror of Friday's: the bills and the front richened — the 1-month 4 bp and the 1-year 3 bp — while the belly cheapened 2 bp and the long end richened 1 to 2 bp, so the week's curve move is a belly-led cheapening bracketed by two rallying wings. That is an unusual configuration and it is worth naming precisely: it is what a market does when it has stopped trading the policy path and started trading paper.

The 20s30s inversion held at exactly -1 bp, with the 20-year at 5.26% and the 30-year at 5.25%, both a basis point cheaper. The spread has now printed -1 bp for two consecutive sessions after seven at exactly zero. For the position in Section 12 item 2 that is a flat session, and the invalidation at -3 bp is unchanged at two basis points away. What is not unchanged is the catalyst: Treasury confirmed that from 9 September its nominal long-end liquidity-support buybacks rise from a maximum of $2bn to at least $4bn per operation, in the 10-to-20-year and 20-to-30-year sectors, running through the 4 November refunding. The operation the trade was built to wait for is tomorrow, and its buckets straddle the exact spread being expressed.

The vendor cross-check does not reconcile at the three-year, and the disagreement is about direction rather than level. WSJ's bond board at the 5:00 p.m. ET close marks the 3-year at 4.463%, up 1.0 bp, against the official par row's 4.44%, down 1 bp — a 2.3 basis-point level gap and opposite signs. Elsewhere the two agree: WSJ has the 10-year at 4.794% (+0.5 bp), the 30-year at 5.248% (+0.1 bp), the 2-year at 4.404% (+3.2 bp), the 5-year at 4.570% (+1.9 bp) and the 7-year at 4.673% (+1.4 bp), every one within two basis points of the official row and every direction agreeing. Bloomberg marks the 10-year at 4.79%, +1 bp. This report publishes the official par row and states the conflict rather than choosing the vendor that makes the belly story tidier: the ninety minutes between Treasury's 3:30 p.m. bid-side snapshot and the 5:00 p.m. quotes is the most likely explanation, and if it is, the hump will be gone from Wednesday's official row.

The bill curve cheapened with the belly, which is the opposite of Friday. The 1-month rose 2 bp to 3.81% after falling 4 bp on Friday, and the 3-month rose 3 bp to 3.94%, taking the 3-month-minus-1-month spread to 13 bp from 12 bp. Off the published table, the 1.5-month rose 5 bp to 3.88% — the largest single move anywhere on the fourteen-point curve — the 2-month rose 1 bp to 3.91%, the 4-month rose 2 bp to 4.02% and the 6-month rose 2 bp to 4.00%. Two consequences. The 1.5-month-to-2-month gap collapsed to 3 bp from 7 bp, and it has now moved 9, 7 and 3 across three observations, so the anomaly this report has tracked since late August is closing. The four-month-above-six-month inversion held at 2 bp. All four tenors are off-table under the trimmed set and are cited here because they carry the financing story; the funding read sits with the money-market data in Section 9 block b.

Source: Federal Reserve Bank of New York Economic Indicators Calendar for September 2026 (all times Eastern), cross-checked against the Wall Street Journal and Investing.com economic calendars. Consensus figures below are the WSJ market forecasts, independently captured this session.

Current week (Sep 7 - Sep 11) — still to come

Next week (Sep 14 - Sep 18)

Current target range: 3.50%-3.75%. Three days, a seven-month yen high and a six-week crude high moved the September meeting by exactly nothing.

CME FedWatch headline — 16 September 2026 meeting. Data as of 8 Sep 2026, 05:04:51 p.m. CT (6:04 p.m. ET), read from the FedWatch probability table. Contract ZQU6.

Provenance of every column, stated — and Friday's live read corrected by 0.8 of a point. The footer timestamp reads 05:04:51 CT with no meridian; the page's own "Last Updated" line reads 08 Sep 2026 05:19:57 PM CT, fifteen minutes later, so it resolves as p.m. and is an indicative read taken about an hour after the 4:00 p.m. CT ZQ session close rather than a settlement snapshot. The 1 DAY column carries the legend date 4 September and prints 59.4%, against the 58.6% this report published from CME's live column on Friday evening — a +0.8 pp correction, recorded here rather than left standing, and the same size of correction as the previous session's. 1 WEEK (1 September) at 67.2% and 1 MONTH (7 August) at 44.4% carry genuine reference dates and are used below. The Investing.com matrix underneath is timestamped 8 Sep 2026, 05:55 p.m. EDT and is the primary source for parts (a), (b) and (c).

The CME-versus-Investing.com gap, quantified, and it widened back out. CME puts the September hike at 59.4% at 6:04 p.m. ET; Investing.com at 58.4% at 5:55 p.m. ET — a 1.0 percentage-point difference across nine minutes, against 0.2 points on each of the two previous sessions. Investing.com publishes the September future at 96.303, unchanged from Friday to the third decimal. Because the 16 September meeting sits mid-month, only about 47% of the contract's averaging period is affected, so one basis point of ZQ price is worth roughly ten percentage points of hike probability — which means the 1.0-point vendor gap is about a tenth of a basis point of price and is not a disagreement about the market so much as about the moment it was sampled.

One-day, one-week and multi-day momentum, and the striking fact is the absence of one. The September hike is unchanged on CME's own columns, 59.4% against 59.4%, and up 2.0 points on Investing.com's own columns, 58.4% from 56.4%. The two vendors disagree about whether anything happened, which is what a zero-change session looks like when it is sampled twice. The ZQU6 price is identical to Friday's at 96.303 and ZQZ6 is identical at 96.060, so the front of the strip did not move at all across three calendar days. The multi-day read is the discipline, and it is unflattering to the narrative of the past fortnight: the meeting sat at 67.2% a week ago, fell to 49.4% on Waller's Thursday, rebounded to 59.4% on the payroll and has stayed there — so it is 7.8 points less hawkish than a week ago and 15.0 points more hawkish than a month ago. Further out, the 2026 horizons barely moved: October's cumulative-above to 69.9% from 68.5%, December's to 86.2% from 84.4% on the vendor's own prior-day columns. The probability of a cut at any 2026 meeting remains 0.0%.

(a) Current-year meeting distributions

Investing.com Fed Rate Monitor, updated 8 Sep 2026 05:55 p.m. EDT. Format: current [prior day] [prior week]. Modal range in bold.

Sums are 100.0% at September and October and 100.1% at December. Three observations. First, the September distribution is identical to Friday's published card, 41.6% and 58.4%, so the vendor's own current column reproduces the prior edition exactly for a second consecutive session — the cleanest run of the reporting window. Second, the December contest keeps narrowing and the mode keeps losing ground: one hike leads at 41.1% against 36.4% for two hikes, a 4.7-point gap where Friday's was 5.6 and Thursday's 9.9, and the direction is unbroken across three sessions as mass moves from +25 into +50. Third, the tails were bought for a second session — October's +50 bucket to 16.1% from 15.6% and December's +75 to 8.7% from 7.9% — so the 2026 distribution has added roughly 1.3 points of tail mass on top of Friday's 3.4. A meeting probability that does not move while the tails around it are being bought is a market hedging the outcome it cannot price.

(b) Next-year meeting path

Modal range, its probability, and the cumulative probability above and below the current 3.50%-3.75% range, with the contract price that draws it.

The whole session's rate content is in this table, and it is a repeat of Friday's shape at a quarter of the amplitude. The eight contracts print 96.010, 95.910, 95.855, 95.780, 95.760, 95.745, 95.740, 95.755 against Friday's 96.015, 95.915, 95.865, 95.795, 95.775, 95.760, 95.755, 95.775 — 0.5 to 2.0 basis points cheaper, with the cheapening rising monotonically along the curve. Set that against the front, where September 2026 and December 2026 are unchanged to the third decimal. So for a second consecutive session the market added to the terminal rate while leaving the schedule alone, and it did so on a day with no American data. Cumulative-above at December 2027 rose to 94.5% from 93.6%, January 2027's modal probability firmed to 37.6% from 37.1%, and the first non-trivial cut probability fell to 0.4% at 3.25-3.50% from 0.5%. Two sessions of back-end-only cheapening with a pinned front is the cleanest expression available of a market that believes in the destination and has stopped arguing about the route.

(c) Year-end probability ladders

Year-end 2026 — the 9 December meeting.

Cumulative above the current range: 86.2%. Cumulative below: 0.0%. Sum: 100.1%.

Year-end 2027 — the 8 December meeting.

Cumulative above the current range: 94.5%. Cumulative below: 0.4%. Sum: 100.0%.

Rounding, transparently. Every figure is reproduced at the vendor's own one-decimal precision. Column sums of 99.9% or 100.1% are rounding artefacts of that precision, not missing probability mass; no cell has been rescaled, and cells shown as 0.0% are ranges the vendor's own card either omits entirely or publishes as zero, which under CME methodology means a probability below the rounding floor. The 2026 year-end ladder sums to 100.1% for exactly this reason; the 2027 year-end ladder and the September and October 2026 rows sum to exactly 100.0%.

(a) IG and HY credit spreads

ICE BofA option-adjusted spreads via FRED. FRED publishes with a one-business-day lag: the levels below carry the 7 September 2026 as-of date, a federal holiday on which the index rolls forward, not the 8 September close. Same-day direction is cross-checked against the cash-market proxies underneath and against Bloomberg and WSJ credit coverage. 1-Week is versus the 31 August row.

CDX — the six-step ladder was worked again in the local Chrome browser and is reported so the gap stays auditable. (1) Bloomberg in Chrome: /markets and /markets/rates-bonds both rendered fully, and a full-text scan of the rates-and-bonds page returns zero occurrences of the string; the fixed-income index tables carry Bloomberg's own aggregate indices, not credit-default swap levels. (2) WSJ Market Data bonds page in Chrome: rendered with its Treasury, consumer-rate, government-bond and economic-calendar tables populated, and a full-text scan returns zero occurrences. (3) Cbonds: the dedicated CDX.NA.IG 5Y index page that carried a subscription-masked quote on Friday now returns "no such page," and the site's index search is behind the same wall. (4) FT Markets Data: markets.ft.com is refused by the Chrome extension's domain policy ("Navigation to this domain is not allowed"), so the step could not be executed at all this session; that is a tooling failure, not an absence of data, and it is recorded as such. (5) Barchart is refused by the same domain policy; TradingView symbol searches for CDX continue to resolve to unrelated instruments, and CME's credit-index product pages carry specifications rather than levels. (6) Cash-market proxies, labelled as proxies: HYG closed $79.13, -0.04%, and LQD $105.50, +0.02%, against Friday's $79.16 and $105.48 — both reconciling to the cent against the prior edition. No CDX level is published here.

The tail resumed widening, and the index did nothing at all. CCC widened 1 bp to 1,055 on the 7 September stamp, so the one-basis-point relief this report recorded on Friday lasted a single update and the four-update trend is intact — the series is 13 bp wider on the week and 167 bp wider on the year. Against that, IG was unchanged at 81 bp and HY unchanged at 268 bp, taking the CCC-minus-HY differential to 787 bp from 786. That is the pattern in one line: the aggregate indices have not moved for two consecutive updates while the tail has widened 14 bp in a week. The cash market went quiet again for a fourth session — HYG -0.04% and LQD +0.02%, a cumulative 0.12% and 0.26% across four sessions — and the reconciliation to the cent against Friday's published closes is worth stating, because it means the silence is real rather than a capture artefact.

(b) Money-market & funding plumbing

New York Fed reference rates, published at approximately 8:00 a.m. ET for the prior business day. No 4 September row had published to the reference-rates endpoint at capture, so the table below repeats the 3 September 2026 effective date carried in the prior edition; the new information this session is in the operations table beneath it, which is dated 8 September. Rate up = red.

The plumbing tightened again and the reserve number is the one that changed character. Reserve balances fell $30.4bn to $2.8945tn in the week ended 2 September — a fifth consecutive weekly decline and three times the previous week's $10.4bn, the largest single-week drop of the run. Against that, the standing repo facility took $21m on 8 September, smaller than Friday's $33m but non-zero for a second consecutive operation, and reverse repo take-up fell to $626m from $702m across three counterparties rather than four. Read together: cash is leaving both sides of the Fed's balance sheet at the margin while a facility designed for inconvenient repo keeps being used. SOFR at 3.66% remains a basis point above the 3.65% IORB on the stale 3 September basis, with a 99th percentile at 3.74%, nine basis points above the rate itself. The September quarter-end is now two weeks away rather than three, and the reserve drain has accelerated into it.

The off-table bill tenors belong here, and this session they say supply rather than policy. Every bill on the fourteen-point curve cheapened: the 1-month 2 bp to 3.81%, the 1.5-month 5 bp to 3.88% — the largest move anywhere on the curve — the 2-month 1 bp to 3.91%, the 3-month 3 bp to 3.94%, the 4-month 2 bp to 4.02% and the 6-month 2 bp to 4.00%. A uniform bill cheapening on a day the September meeting probability did not move is a supply signal, and it collapsed the 1.5-month-to-2-month gap to 3 bp from 7 bp — 9, then 7, then 3 across three observations. The anomaly that has run since late August is closing from the front, on the same session the reserve drain tripled.

(c) Rates volatility & swap spreads

The MOVE vintage is disclosed rather than smoothed, and this session it has moved on retrospectively. The Investing.com series now carries a 04/09 date stamp at 73.10 with an open of 74.68 and a day range of 73.10 to 74.68 — so the 4 September value that had not published when this report went out on Friday has since appeared, and it is 1.58 points below the 74.68 the report carried with a 3 September vintage. That correction is recorded here. What has not appeared is any 8 September value: the card's change field reads 0.00, and it fails its own internal consistency check for a ninth consecutive session, with a "previous close" of 95.74 outside its own 73.10-74.68 day range. The level is published with its vintage; the change is withheld, and the MOVE-versus-VIX ratio is computed on a stale numerator and flagged. What can be said without the series: VIX rose 8.19% against the 4 September close on a session the index fell 0.58%, while the rate market moved 3 bp at its widest point and the September meeting probability moved zero. That combination — equity volatility bid, rate volatility unobservable, rate levels inert — is the least informative configuration this section can print, and it will not resolve until the series republishes or PPI lands on Thursday. Swap spreads at the 2-year, 10-year and 30-year were not obtainable from a primary source this session and are not asserted; the substitute evidence is the 20s30s holding at -1 bp in Section 6 into an expanded buyback that starts tomorrow.

(d) Issuance, leveraged loans & private credit

The September stampede did not show up, and that is the most important credit datapoint of the session. Post-Labor-Day investment-grade issuance is running at its weakest since 2020, after an August that set a record at roughly $130-145bn against a post-2019 August average near $95bn. Year-to-date supply through August exceeded $1.68tn, up 27% on 2025, against a full-year projection of $2tn or more that the current pace makes "a stretch rather than a sure thing." The constraint named is the level rather than the spread: the 10-year at 4.80% is the highest close of this reporting window, average IG yields sit near 5%, and credit spreads have been stable in a 70-80 bp band all year. This is the first hard evidence against the $215bn September expectation this report has carried since the start of the month, and it cuts both ways — borrowers sidelined by yield are borrowers who still have to come, so the calendar is being deferred into a narrower window rather than cancelled, with only 9 September and the two data mornings before the FOMC. JPMorgan's 24 August syndicate view that the market can absorb the stampede has not been tested, because the stampede has not started. On the loan side no updated Morningstar LSTA print was obtainable this session and none is asserted. The named private-credit watch item is unchanged and unresolved: Broadcom's contingent residual-value guarantees to two AI laboratories are vendor financing that no published spread series captures, and Broadcom closed +2.98% on a session its index rose 1.30%.

Levels from the TradingEconomics currency board read after the U.S. close. Every major row carries a Sep/08 date stamp, so the vendor's own %Chg column measures the 8 September session — which began at 17:00 ET on Monday and therefore excludes the Labor Day holiday session in which most of the yen's move was made. Both bases are published below: the Session column is the vendor's own 8 September change, and the vs 4 Sep column is this report's own two-session calculation against the prior edition's published levels for the same vendor. Worked examples are in Data Notes. A handful of Asian crosses carry a live 06:23-06:24 clock stamp instead of a date and are flagged in-line. Corroborated against the Investing.com majors board. Quote basis: USD per unit for EUR, GBP, AUD and NZD; units per USD for JPY, CHF, CAD, KRW, TWD, CNY, MXN, SGD and INR.

The take: the dollar gave back the payroll and the market that took it was Japan, in a session New York could not trade. DXY at 98.822 is 0.34% below Friday's 99.155 and 0.86% lower on the week, which erases the 0.25% payroll gain and then some. The concentration is what makes it interesting. USD/JPY fell 1.60% across the two sessions but only 0.41% during Tuesday's own, so roughly three-quarters of the yen's advance to a seven-month high was made on Monday 7 September, while U.S. cash equity and Treasury markets were shut. The same is true, in miniature, of the Taiwan dollar: 0.37% stronger versus Friday and exactly unchanged on Tuesday's session. On Friday this report wrote that Asia had bought a softer American rate path which the payroll destroyed, and that the correction could not be arbitraged in New York until Tuesday. The correction arrived on schedule and in the opposite direction to the one that argument implied — the region did not sell its currencies back, it bought them harder, and it did so on domestic grounds.

The yen's bond leg refuses to corroborate the obvious story, and that is the most useful thing on this page. The regional coverage attributes the move to Japanese data and Bank of Japan tightening expectations into the 18 September decision. If that were the whole of it, 10-year JGBs should have cheapened; they richened 2 bp to 2.88%. A currency at a seven-month high while its own government curve rallies is not trading the policy differential — it is trading flow, and Bloomberg's framing on Tuesday was repatriation. That distinction matters for how the move behaves from here: a rate-differential move mean-reverts when the differential does, and a repatriation move does not. The U.S. leg gives no help either, since the 2-year cheapened 2 bp and the 3-year richened 1, so the American side of the pair was close to inert while the currency made a seven-month high.

The European currencies did nothing at all, for a second consecutive session, on a day their equity markets diverged violently. EUR/USD +0.03%, GBP/USD 0.00% to four decimal places, USD/CHF -0.03% — three of the four largest majors inside four basis points, on a session when 10-year Bunds, OATs, BTPs and Bonos all richened 2 bp against a 2 bp U.S. cheapening at the 10-year. That is a four-basis-point differential move producing no currency at all. The Swiss franc is the sharpest version: USD/CHF was flat on the day the Swiss equity market fell 1.55% on a single stock. A currency that will not move on a four-basis-point rate move or a 1.55% domestic equity decline is a currency waiting for Friday's CPI, which is the same conclusion Section 8's frozen meeting probability reached from the other direction.

Settlement basis, stated, and reconciled to the prior edition. This edition switches to the Investing.com per-contract historical board as the settle series of record for every row, which the playbook designates as the primary basis; the prior edition took its daily settles from each contract's quote-page fields, and the two differ by up to $0.43 on Brent and $0.26 on WTI. The full reconciliation and the restated 4 September figures are in Data Notes; daily changes below are computed within the new basis and are internally consistent. WTI, RBOB, heating oil and natural gas are on the October contract; gold, silver and copper on December; Brent on the November contract, front since the 23 August rollover. No front-month roll occurred this session. Because Monday 7 September was a U.S. holiday with a live electronic session, the vendor's daily change is measured against the 7 September row; the two-session move against Friday is given in the driver column where it differs materially. Weekly and year-to-date columns are TradingEconomics spot returns on the vendor's Sep/08 stamp.

\YTD figures marked with an asterisk are TradingEconomics spot year-to-date returns, not futures returns on the contracts quoted above. Weekly columns are TradingEconomics one-week changes on the same caveat. Mixing the two would be a basis error; they are presented in separate columns for exactly that reason.*

Gold's session is the one to study, and the shape is the opposite of Friday's. On Friday the contract gapped down and never traded above the prior close; on Tuesday it opened at $4,469.01, ran to $4,488.54 — a level it had not seen since the 3 September settle — then fell $98.09 to $4,390.45, closing at $4,397.61, down $71.48 or 1.60%. That is a $90.93 give-back from the high, 2.03% off the top, and it is a failure at resistance rather than a gap that would not fill. The mechanism is not the one that worked on Friday: the two-year cheapened only 2 bp, the three-year richened, and the dollar fell 0.36%, so gold declined 1.60% into a weaker dollar and an essentially unchanged real-rate path. That is the first session of the window in which the metal has fallen without a dollar or rate explanation, and the residual is positioning. Silver fell 1.37% to $66.285, marginally less than gold, so the gold-silver ratio slipped to 66.34 from 67.07 and silver has now outperformed for a fourth consecutive session including three down ones. Copper closed up 0.60% at $6.7725, refusing the precious selloff for a second session and confirming again that the move is monetary rather than industrial.

Energy is where the session's conviction was, and the split is crude against distillate for a third time. Brent settled $99.27, up $2.27, and WTI $94.27, up $1.75, so Brent outperformed and the Brent-WTI differential widened to $5.00 from $4.80. Underneath, RBOB rose 1.85% and heating oil fell 0.75% — the third session in four in which the distillate leg has lost money on a rising barrel. The catalyst is supply risk rather than demand: Houthi attacks on Saudi facilities took WTI to a six-week high, on top of the U.S.-Iran escalation and the Hormuz vessel blacklist this report has tracked for a fortnight. Natural gas fell 2.95% to $2.899, the worst move on the board and a clean giveback of Monday's 2.02% gain, on the same morning JPMorgan cut Gulfport to Underweight on gas fundamentals — and the contract remains -20.28% on the year, the only major commodity still deeply negative.

The crack spreads, on a consistent October basis against October WTI at $94.27, and with Friday restated onto the same historical-board basis:

• Distillate crack: $4.6319 × 42 - $94.27 = $100.27, up $1.06 from a restated $99.21.

• Gasoline crack: $3.2823 × 42 - $94.27 = $43.59, up $0.06 from a restated $43.53.

• The differential widened $1.00 to $56.68 from $55.68 — the first widening in three sessions.

That $1.00 widening does not undo the two-session, $9.59 collapse the prior edition recorded, and it is not evidence that the collapse was wrong. What it is evidence of is that the physical signal and the paper signal have started to converge: Reuters has U.S. diesel at record highs, the October crack fell nine dollars into that, and it has now taken one dollar back on a session crude rose 1.89%. The structural case for distillate is unchanged on the spot basis — heating oil +118.19% year to date against gasoline's +91.91% — so this remains a timing question. The observation to carry into Thursday's EIA petroleum status report at 12:00 is that a crack which widens on a supply-driven crude rally is behaving normally again, and one more such session would make the two-day collapse look like a positioning flush rather than a thesis break.

Desk-style ideas for institutional investors. Each carries an explicit expression, catalyst and invalidation. These are not personalized investment advice; verify independently and size to your own mandate before acting.

1. The rates trade — long ZQZ6 against short ZQZ7 paid a third time; hold the quarter

Mark first. Long ZQZ6 (December 2026) against short ZQZ7 (December 2027), DV01-matched one-for-one at $41.67 per basis point per contract, entered on 3 September at 96.085 / 95.820 for a spread of 26.5 bp, quarter size. Tuesday's mark: ZQZ6 96.060, ZQZ7 95.755 — a spread of 30.5 bp. That is +2.0 bp on the session, worth +$83.33 per contract pair before costs on a quarter, and the position is +4.0 bp from entry, or +$166.67 per pair.

Why it paid, and it is the same mechanism as Friday. ZQZ6 was unchanged to the third decimal at 96.060 and ZQZ7 cheapened 2.0 bp, so the spread widened entirely because the back leg moved. That is now two consecutive sessions in which the front of the strip has been frozen while the 2027 contracts cheapen monotonically — 0.5 bp at January rising to 2.0 bp at December — and on Tuesday it happened with no American data at all. A back-end cheapening on an empty calendar is the cleanest version of the thesis this trade was built on, and it is worth being precise about the reversal: the entry note argued the back would be the stable leg and it has been the moving one on both paying sessions. The trade is right for a reason it did not predict, and that is recorded rather than reinterpreted.

The modal path, base case and tails. Modal path: a 25 bp hike on 16 September at 59.4% on CME and 58.4% on Investing.com, with ease at 0.0%; one hike is modal at October (53.8%) and at December (41.1%); the 2027 strip modes at 4.00-4.25% at all eight meetings. Base case: one hike in September, a second priced by December at 86.2% cumulative-above, and a terminal rate that has risen on each of the last two sessions while the schedule has not moved at all. Tail one, and it is the whole trade: CPI on 11 September at 08:30, consensus +0.4%. A hot print takes September through 70% and cheapens the front leg, which costs this spread; the offset is that it would cheapen the back harder. Tail two: a soft CPI that reinstates a parallel richening of the back, which is the configuration that narrows this spread and the only one that has hurt it. Practical implication: at roughly ten percentage points of probability per basis point of ZQ price for a mid-month meeting, and with ZQU6 and ZQZ6 both literally unchanged across three calendar days, the front leg is now carrying no information — the entire position is a 2027 terminal-rate view expressed through a calendar spread, and it should be sized as one. Catalyst: the 9 September buyback at double size; PPI 9/10; CPI 9/11; the 16-17 September FOMC. Invalidation, unchanged: the spread through 22.0 bp; or the September cumulative hike back below 45% on either vendor; or the December 2026 mode reverting to two hikes — which is now 4.7 points away, from 5.6 on Friday, and is the invalidation closest to firing. Sizing: a quarter, at $41.67 per basis point per pair.

2. Long the 20-year against the 30-year, into a buyback that starts tomorrow — flat on the session; hold the half

Expression: long the 20-year bond against short the 30-year, DV01-neutral, half size. Mark: 20s30s at -1 bp, with the 20-year at 5.26% and the 30-year at 5.25%, both a basis point cheaper — unchanged on the day and 1 bp against the position from entry. What changed, and it is the reason to keep holding. Treasury has now confirmed the operation this trade was built to wait for: from 9 September, nominal long-end liquidity-support buybacks rise from a maximum of $2bn to at least $4bn per operation, in the 10-to-20-year and 20-to-30-year sectors, running through the 4 November refunding. Treasury's own rationale is "consistent strong sponsorship" in those sectors. That is a doubling of a bid that straddles both legs of this spread, and the 20-to-30-year bucket is the one that matters: if the operation concentrates there, the 30-year richens against the 20-year and the position loses; if it spreads across both buckets, the 20-year's greater cheapness attracts the marginal bid. The trade is a bet on the second, and tomorrow it stops being a thesis and becomes a printed result. Catalyst: the 9 September operation and its maturity buckets; Treasury's 4 November quarterly refunding and the buyback guidance that comes with it. Invalidation, unchanged: 20s30s through -3 bp, or an explicit Treasury statement ruling out changes to long-end auction sizes. Sizing: a half, unchanged. Do not add ahead of the operation.

3. Protection on the CCC cohort funded in IG — back to paying; hold the half

Expression: long CCC-exposed credit protection (or short a levered-loan / CCC-heavy vehicle) against long IG cash. Mark: CCC 1,055 bp, +1 bp; HY 268 bp, unchanged; IG 81 bp, unchanged on the 7 September FRED update, taking the CCC-minus-HY differential to 787 bp from 786 — a 1 bp gain, reversing Friday's 1 bp loss, for a cumulative +21 bp across five updates. The honest reading: the position has now been paid on four of five updates and the one loss lasted a single day. More useful than the mark is the shape: the aggregate indices have been unchanged for two consecutive updates while the tail has widened 13 bp on the week, which is exactly the decoupling the pair was constructed to isolate. The cash proxies remain silent for a fourth session — HYG -0.04% and LQD +0.02% — so the pair is still neither confirmed nor contradicted by anything tradeable, and that is the standing caveat. Action: hold the half; do not add. Catalyst: the September IG calendar, which is running at its weakest post-Labor-Day pace since 2020 and has to clear somewhere; PPI 9/10 and CPI 9/11; quarter-end funding from the 15th, with reserves down $30.4bn in a week. Invalidation, unchanged: the differential back through 750 bp, or IG widening beyond 90 bp, which would mean the aggregate has joined the tail and the pair no longer isolates anything. Sizing: a half, unchanged.

4. Short the packaged-food cohort against long consumer defensive — the first losing session; hold the half into its last day

Mark. The nine-name basket closed Bunge +4.36%, Campbell's +1.78%, Tyson +1.67%, Kraft Heinz +0.20%, McCormick -0.35%, Lamb Weston -0.94%, JM Smucker -1.13%, Conagra -1.42%, General Mills -1.57% for an average of +0.29%, against consumer defensive at -0.61% — a 0.90-point loss on the pair, against day one's 0.57-point gain, for a cumulative -0.33 points. The honest reading. This is the first session in which the basket has outperformed its own group, and it did so on a broad bid rather than a single name — four of the nine closed higher. The invalidation was written as "the basket recovering more than half of Thursday's underperformance within three sessions." Thursday's dispersion was 3.77 points; half is 1.885; the cumulative recovery after two of the three sessions is 0.33 points. The trigger is not close, but the direction has turned and the window has one session left. Action: hold the half through Wednesday and mark it out on Thursday whatever happens, rather than extending the window because the trigger did not fire. Catalyst: Kroger on 11 September before the open (Section 5) as the channel read; any payout cut in the cohort. Invalidation, unchanged: the basket recovering more than half of Thursday's underperformance within three sessions; or a takeover approach for any basket constituent. Sizing: a half, dollar-neutral. One session of the three-session window remains.

5. Long volatility on the semiconductor complex — the invalidation clock did not advance; keep the quarter

Mark. October volatility on the semiconductor proxy, a quarter, expressed in premium rather than delta. The clock, precisely. The invalidation is "SOX outperforming the Nasdaq 100 on two consecutive up sessions, which would say the non-participation was positioning rather than doubt." Friday was the first of those. Tuesday was not a second, because the Nasdaq 100 fell 0.12% — SOX outperformed by 1.42 points, but not on an up session for the index, so the counter does not advance and by the invalidation's own wording it resets. That is the literal reading and this report takes it rather than the convenient one. The internals strengthened the position. The complex rose 1.30% while Nvidia fell 2.01% — its largest member moving the opposite way, where Friday it merely lagged — and the day's leadership was Intel +9.05% on a price increase and Qualcomm +3.18% on an AWS design win, neither of which is a cycle datapoint. A sector index that rallies on its two most troubled members while its bellwether falls 2% is the definition of dispersion, and it is now doing it in both directions inside three sessions. Action: keep the quarter, do not add. Mark it out on the next genuine up session for the Nasdaq 100; if SOX outperforms on that one and again on the next, close it and record the loss rather than re-reading the invalidation a third time. Catalyst: Oracle, Adobe and Copart on 10 September after the close (Section 5); CPI 9/11. Invalidation: as written. Sizing: a quarter.

6. Short the credit-bureau complex against long the S&P 500 financials — flat on a violent day for financials; hold the half

Mark. Fair Isaac +0.11% to $933.30 and Equifax -3.07% to $171.62; TransUnion did not appear in the 494-line component capture and no independently sourced close is asserted for it, so the basket is marked on the two names that were captured, at -1.48%, against S&P 500 financials at -1.22% — a 0.26-point gain on the pair. The honest reading, and the caveat is larger than the mark. A two-name mark on a three-name basket is a weak measurement, and it is disclosed as such rather than being completed with an estimate. What the session does say is that FICO stopped falling — it closed effectively unchanged after Friday's 16.68% rout — while Equifax kept going, adding 3.07% to Friday's 6.37%. That is the read-across name underperforming the directly-affected name on day two, which is the opposite of a sympathy-move unwind and is consistent with the thesis: the bi-merge risk sits with the bureaus that could lose the tie-break, not with the scoring monopoly whose loss has already been announced. Catalyst: sell-side mortgage-channel revisions, which have now had a full weekend and a holiday; any FHFA follow-through on bi-merge; the bureaus' own investor communications. Invalidation, unchanged: the basket recovering more than half of Friday's decline within five sessions — one session used, and the basket is 1.48% lower rather than higher; or an explicit FHFA statement retreating from bi-merge. Sizing: a half, dollar-neutral. The risk to name honestly: this remains a crowded short after a 17% day, borrow will be expensive, and the mark above rests on two of three legs.

Prior closes, marked forward. The AI-halo basket against long Nvidia, closed on 1 September, would have lost on Tuesday for the first time since being closed: Palo Alto +1.12%, ServiceNow -4.99%, Fortinet +0.76% and Adobe -3.47% for a four-name average of -1.65% against Nvidia -2.01%, a 0.37-point loss on the pair, against Friday's 3.17-point gain. The October distillate-versus-gasoline crack pair, closed on 3 September at a $61.48 differential, would have lost a further $1.00 on Tuesday as the differential widened to $56.68 — the first session since it was closed in which the position would have given money back, and a reminder that the $9.59 the decision saved was never going to be a straight line. The short-debasement basket against long dollar, closed on 3 September, would have been mixed: gold -1.38% then -1.60% paid the short leg, but DXY fell 0.36% on Tuesday, so the two legs split for the first time since the close.

The vol note. VIX closed 15.72, up 1.19 points or 8.19% against the 4 September close, with a session range of 15.22 to 15.94, on a day the index fell 0.58% — a 14-to-1 ratio of volatility gain to index decline, against Friday's 3.9-to-1. The four-observation path is 14.32 → 14.53 → 15.72, so the window low is now two sessions behind and the surface has repriced without the index doing much. A 15.72 handle asks for roughly a 0.99% daily move, against a market that has just delivered 0.38% and 0.58% at the index level while producing a 10% single-name decline in a megacap pharmaceutical and a 9% rise in a megacap semiconductor. That is the argument for owning dispersion rather than index vol here, and it is reinforced by the fact that the rates side of the comparison remains unavailable: MOVE has still not published an 8 September value, and its 73.10 carries a 4 September vintage (Section 9 block c). The instruction is unchanged in direction and cheaper in one leg only: own the 10-11 September inflation block outright, finance it in the 17-18 September meeting expiry where a 59.4% meeting still carries the richest premium on the curve, and note that the front weekly is no longer the giveaway it was on Friday — VIX has already added a point and a fifth, and the calendar it is pricing has four blank earnings sessions behind it.

Crowded consensuses to stress-test, with the numbers.

1. "The September meeting is settled at just under 60%." It is priced there, and it has not moved. CME's NOW column reads 59.4% against a 1 DAY column of 59.4%, and ZQU6 is unchanged at 96.303 across three calendar days that contained a seven-month yen high, a six-week crude high, a 10% decline in a megacap pharmaceutical and the worst breadth of the window. Stress test: the 1 WEEK column reads 67.2% and the 1 MONTH column 44.4%, so the meeting has travelled 15 points in a month and 8 points in a week and has now simply stopped. At roughly ten points of probability per basis point of ZQ price, the whole distribution sits inside three-quarters of a basis point of Friday's CPI print. Nothing is settled; the market has run out of information, which is not the same thing.

2. "The terminal rate is anchored." It has now moved on two consecutive sessions, in the same direction, with no data. Every 2027 contract cheapened 0.5 to 2.0 bp with the move rising monotonically along the curve, cumulative-above at December 2027 rose to 94.5% from 93.6%, January 2027's modal probability firmed to 37.6%, and the first non-trivial cut probability fell to 0.4%. Stress test: a back-end-only cheapening on an empty calendar cannot be explained by the reaction function, because there was nothing to react to. It is either supply — the 2027 strip absorbing hedging from a deferred corporate calendar — or a slow re-rating of the destination. The first would reverse when the calendar clears; the second would not, and the two are indistinguishable until the primary market opens.

3. "The long end is the problem." Still not, and the belly took its place. The 20-year and 30-year each cheapened only 1 bp and are 1 to 2 bp richer on the week, while the five-year and seven-year cheapened 3 bp and the three-year richened 1 bp on the official par row. Stress test: this is the first session of the window in which the curve's largest disagreement was between adjacent belly points rather than between the wings, and the vendor tape does not corroborate it — WSJ marks the three-year 1.0 bp cheaper, not 1 bp richer. Either the official 3:30 p.m. snapshot has a stale print in it, in which case Wednesday's row erases the hump, or the five-to-seven-year sector is absorbing something the 5:00 p.m. quotes have already smoothed. Treasury's doubled buyback starts tomorrow in the 10-to-20 and 20-to-30-year buckets, which is adjacent to but not on top of the disagreement.

4. "Credit is fine because credit is quiet." IG at 81 bp and HY at 268 bp are unchanged for two consecutive updates, and CCC widened 1 bp to 1,055, taking the tail 13 bp wider on the week and 167 bp wider on the year. Stress test: the aggregates are quiet because post-Labor-Day investment-grade issuance is running at its weakest since 2020, against an August that set a record near $130-145bn and a year-to-date total above $1.68tn. A spread series with no new paper behind it is not evidence of health; it is an absence of evidence, and the deferred calendar now has to clear into a 4.80% ten-year with three usable sessions before the FOMC. The nearer risk is the plumbing: reserves fell $30.4bn in a week, three times the prior week's drain, the standing repo facility was used on two consecutive operations, and quarter-end is two weeks out.

5. "Volatility is right to be calm." It is no longer calm and it is still not expensive. VIX at 15.72 is 8.19% above the 4 September close and asks for roughly a 0.99% daily move into PPI and CPI on 10 and 11 September and a meeting priced at 59.4%. Stress test: the index delivered 0.58% while a megacap pharmaceutical fell 10.08%, a megacap semiconductor rose 9.05% and the worst S&P 500 name fell 10.70% — realised single-name dispersion far above what a 15 handle implies at the index level. And the cross-check is still missing: MOVE has not published an 8 September value and has now failed its own internal consistency check for nine consecutive sessions. An equity surface being repriced with no functioning rates-volatility comparison is a surface being judged on half the evidence, for the second consecutive edition.

The two-sided geopolitical tape. Escalation moved from the Gulf to the Red Sea and the barrel answered immediately: Houthi attacks on Saudi facilities took WTI to a six-week high at $94.27 and Brent to $99.27, on top of the U.S.-Iran clashes Reuters has described as the fiercest since July, Israeli Defence Minister Israel Katz's threat to "cripple" Iranian energy facilities, and Iran's expanded Hormuz vessel blacklist. U.S. diesel remains at record highs, with Ukrainian strikes on Russian refineries compounding the physical tightness. Against that sit the same three deflators: Iraq's August exports at 2.34m barrels a day against 1.35m in July, President Putin's signalled openness to Ukraine negotiations, and ANZ's observation that elevated inventories absorbed the initial supply crisis while it raised its short-term Brent forecast to $95 — a forecast the market is now $4.27 above. Canada's retaliation on $20bn of U.S. goods took effect on 8 September and did not visibly move the tape.

Structural watch items. Reserve balances at $2.8945tn, down $30.4bn on the week and falling for a fifth consecutive week; a standing repo facility take-up of $21m on 8 September after $33m on 3 September, two consecutive non-zero prints; reverse repo at $626m across three counterparties; SOFR at 3.66%, a basis point above the 3.65% IORB on a stale 3 September basis, with no 4 September reference-rate row published at capture; a 1.5-month bill 5 bp cheaper at 3.88%, the largest move on the fourteen-point curve, collapsing the 1.5-month-to-2-month gap to 3 bp from 7; post-Labor-Day IG issuance at a six-year low into a $215bn monthly expectation and a 4.80% ten-year; Treasury's buyback doubling to at least $4bn per operation from 9 September in the 10-to-20 and 20-to-30-year sectors through 4 November; the Lp(a) class re-rating and whether Amgen's olpasiran is read as mechanism or trial design; Boston Scientific's 25 August cyber incident and a device complex that sold off together on unrelated causes; Japan's 10-year 2 bp richer at 2.88% with the yen at a seven-month high and a Bank of Japan decision on 18 September; and Howmet at $231.53 against standing Citigroup and Bernstein targets of $329 and $328.

Companion files: the canonical Markdown report of record US_CrossAsset_Daily_2026-09-08.md, which additionally carries the Overnight / Asia & Europe Read-Through and the Source Links appendix, and US_CrossAsset_Daily_2026-09-08_DataNotes.txt, which carries the full Data Notes & Conflicts, the source links and the quality-gate results.