Closing Briefing — Tuesday, August 11, 2026

Sources: CNBC daily market live blog for 11 August and the 11 August futures page; Bloomberg /markets and /markets/rates-bonds boards, “US Stocks End Down as Hormuz Deadlock Sends Oil Up Ahead of CPI” and “Nvidia’s Show of Financial Force Soothes Jittery Credit Markets”; WSJ Market Data Bonds & Rates (5:03–5:05 PM ET); Investing.com Major Indices, NDX, SOX, Russell 2000, U.S.-500-component, Dow-component, world-indices, index-futures and real-time-futures boards (equity boards stamped 15:59:59); U.S. Treasury Text View; CME FedWatch (four-column table, “Data as of 11 Aug 2026 05:03:29 CT”) and Investing.com Fed Rate Monitor (Aug 11, 2026 05:55 PM EDT); FRED; NY Fed; TradingEconomics; Finviz Groups; Earnings Whispers day pages 8/12–8/21.

1. Alphabet is now a single-stock index risk, and it cost the S&P 500 more than the whole rest of the tape gained. Alphabet A −3.84% to $343.80 and Alphabet C −3.60% to $343.04, the fourth losing session in five, dating to Google’s announcement last week that it is reshuffling its artificial-intelligence divisions (CNBC). The Finviz Communication Services group fell 2.06% — three and a half times the next-worst group and the only sector move greater than 1% in either direction — and is now −1.15% year to date, one of only two negative groups on the board (§3). AppLovin −6.01% to $318.64 compounded it after Bank of America downgraded to Neutral and cut the target to $400 from $430, saying risks to the revenue-growth forecast have increased. The structural point for a long/short book: with 270 names up and 221 down, an S&P that fell 0.32% is arithmetically impossible without megacap concentration, and the concentration is currently expressed through one name whose problem is organisational rather than cyclical. A re-org-driven de-rating does not mean-revert on a CPI print. Forward catalyst: any Alphabet leadership or segment-reporting clarification.

2. The semiconductor complex reversed a close-on-the-low, and the reversal was led by equipment rather than logic. The SOX rose 0.87% to 12,098.5 after Monday’s 2.94% decline that ended one tenth of a point off the session low. The leadership order matters: KLA +4.02% to $200.48, Teradyne +3.96% to $379.56, Arista Networks +3.31% to $197.85, Jabil +5.94% to $356.61, Lam Research +1.64%, Micron +0.84%, Applied Materials +0.67%, against Nvidia −0.03% and Broadcom −1.50%. Semi-cap equipment and networking outperformed logic and merchant silicon by 3–4 points on the day, which is a capex signal rather than a demand signal — and the specific catalyst was Nvidia’s $500bn third-party financing coalition, which is a statement about who funds the build-out rather than about who sells into it. UBS upgraded Jabil to Buy on “a multiyear growth cycle fueled by AI investment from Amazon, Meta, and Google.” The fade is the caveat: the SOX high was 12,204.5 and the close 12,098.5, so 0.87% of the bounce was given back into the bell. Forward catalyst: Applied Materials 8/13 AMC (§5); Coherent 8/12 AMC.

3. Nvidia’s credit risk fell about 4 bp, and that is the most important number of the session. Bloomberg reports the cost of protecting Nvidia’s debt for five years dropped to roughly $73,000 a year per $10m of principal — 73 bp — down about 4 bp on the day, after climbing to as high as $82,000 in late July from about half that level for much of the year. The company’s bonds rallied and risk premiums over Treasuries returned to where they sat last week. The mechanism is explicit: Jensen Huang enlisted Apollo, Blackstone, BlackRock, Brookfield, Goldman Sachs and KKR to line up more than $500bn, with the group independently judging individual deals and Nvidia’s own contribution “relatively limited.” GW&K’s Brett Kozlowski called it “a positive development to take out some of the uncertainty about both infrastructure build-out and customers’ future spending.” What was being priced before Monday was circular financing — vendor loans that book revenue now and losses later — and a near-doubling of a single-name CDS in under three weeks in the world’s largest company by market value is a systemic-adjacent move, not an idiosyncratic one (§9). The equity market paid the intermediaries rather than the issuer: KKR +6.91%, Apollo +6%, Blackstone +3.89%, Nvidia −0.03%. Forward catalyst: whether the CDS holds inside 75 bp through the 26 August print.

4. The Dow gave back a 431-point round trip and one name did a third of it. The Dow printed 54,222.85 intraday, 246.87 above Monday’s close, and finished 53,791.85, down 184.13 — a 0.79% high-to-close fade and the widest of the four majors. Honeywell −5.27% to $230.12 removed $12.80 of index price on its own; the reviewed sources attribute the move to updated 2026 guidance — EPS $8.05–$8.35 on revenue of $19.8–20.0bn — landing below consensus alongside cautious commentary on uneven regional demand after the aerospace separation. It did not appear on CNBC’s midday movers list, so treat the attribution as indicative and verify against company IR. UnitedHealth −$6.54, Amazon −$5.84, Apple −$3.36, Amgen −$2.92 and Microsoft −$2.23 completed the drag; Caterpillar +$5.78, Home Depot +$3.68, JPMorgan +$2.27 and IBM +$2.13 were the offset. Six of the ten largest positive contributors were cyclicals and financials — the Dow’s decline was a defensive-and-megacap-tech problem, not an economic one.

5. Crude made a fifth consecutive high and could not hold it, and the airlines told you first. WTI September traded $84.61 in Asian hours — the highest of the month — and settled $83.20, +1.30%, with Brent $88.91, +1.36%. The intraday give-back was 1.63%, and the trigger was Pakistani Defence Minister Khawaja Asif telling Bloomberg that “things are shaping up again in favor of a peace arrangement or a deal,” against the Iranian SNSC secretary’s reiteration that the strait stays shut until conditions are met. The second-order tell is the consumer-travel complex, which was Monday’s cleanest short-crude proxy: United Airlines +2.06% to $126.31, Southwest +1.38% to $45.52, Delta +1.35% to $90.41 — all higher on a day the barrel rose. Equity is trading the reopening headline; the futures curve is trading the closure. One of those is wrong and the resolution is binary. Forward catalyst: the EIA weekly at 10:30 a.m. Wednesday; any SNSC statement.

6. Volatility fell into the most important print of the fortnight. The VIX closed 15.28, −1.16%, with a session range of 15.23–15.61, on a day the S&P fell 0.32% and twelve hours before July CPI. September VIX futures mark 16.78. A spot index declining on a down day, into a scheduled event that both vendors’ Fed cards describe as a coin flip, means the option market is not paying for the event. Cross-check the rates side: the MOVE index is 75.46 on a 10 August vintage and did not print an update, and the implied MOVE/VIX ratio at 4.94× is the highest of the fortnight — rates vol is being carried and equity vol is being sold (§9). That is the cheapest identifiable asymmetry on the board tonight and §12 expresses it.

7. Small caps outperformed on a bull-steepening curve, and it was the third day in four that they moved with the front end rather than with beta. The Russell 2000 rose 0.36% to 3,028.38 against the S&P’s −0.32%, a 68 bp spread, and closed 0.67% below its 52-week high of 3,048.85 — closer to its own high than the S&P 500, the Nasdaq or the Dow are to theirs. CNBC noted the advance pulled the index into positive territory on the quarter. The mechanism is the 3-year at 4.27%, down 4 bp and the biggest mover on the curve: the Russell’s floating-rate and refinancing exposure sits at exactly that tenor. Watch this as the cleanest CPI expression in equity — a soft core print steepens the front further and small caps have the highest beta to it in either direction.

8. The healthcare REITs broke without a catalyst, and nobody has explained it. Ventas −5.43% to $86.96 and Welltower −3.84% to $226.01 were the second- and fifth-worst S&P 500 performers on a session when the 10-year fell 2 bp and Real Estate as a group fell only 0.67%. A 5.4% decline in a rate-sensitive REIT on a day rates rallied is not a rates move, and the reviewed sources — CNBC’s movers lists, the Bloomberg and WSJ market boards and a targeted search — produced no company-specific catalyst. Two senior-housing names moving together at that magnitude with no news is either a sell-side note this report could not source or a large single seller. Flagged rather than explained: this is the kind of unattributed pair move that resolves into a disclosed catalyst within forty-eight hours, and until it does it is a position risk rather than an opportunity.

Four green, seven red, and a 3.04-point dispersion — narrower than Monday’s 4.93 points but concentrated in exactly the same way, at the opposite end. Communication Services −2.06% is 1.39 points worse than the next-weakest group and is the entire index story: strip it out and the median of the other ten is −0.07% and the mean is +0.01%. Two sessions, two single-sector tapes — Monday was Energy up alone, Tuesday was Communication Services down alone — and in both cases the other ten groups moved less than a tenth of a percent at the median. That is a market with no macro view, waiting for a number.

Utilities +0.98% was the best group and the choice is diagnostic: Duke Energy +1.65% to $123.19, NextEra +1.23% to $85.74, Edison International +3.04% to $69.89, Constellation Energy +2.89% to $278.24. Utilities were Monday’s second-worst group at −1.29% on a 7 bp rise in the 10-year and Tuesday’s best on a 2 bp fall — a 2.27-point swing on a 9 bp round trip in the long end, which is the purest duration beta on the board and confirms the sector is currently trading as a bond substitute rather than as a power-demand story. Note the year-to-date: Utilities +1.94% against Energy +33.94% is still a 32-point spread, and one good day does not change it. Energy +0.53% was second on a 1.30% crude settle, and the internal ranking has flipped from Monday: Marathon Petroleum +5.01%, Phillips 66 +4.10% and Valero +2.85% again led, but APA −0.98% and Occidental was flat — the refiners kept the crack and the producers gave back the barrel. Energy’s year-to-date +33.94% now leads Technology’s +23.62% by 10.32 points, the widest gap of the year.

Technology +0.13% is a nothing number that conceals a 10-point internal range: KLA +4.02%, Teradyne +3.96%, Jabil +5.94%, EPAM +2.53%, Keysight +2.47% against Oracle −3.69%, Dell −3.69%, Adobe −3.39%, HP Inc −2.45% and Gartner −3.05%. Hardware and semi-cap up, enterprise software and PCs down — the same split as Monday, running in the opposite direction, which is a rotation with a two-day period rather than a trend. Real Estate −0.67% on a day the 10-year fell 2 bp is the group that did not behave: Ventas −5.43% and Welltower −3.84% dragged it while Prologis +0.53% and American Tower +0.26% rose. Financial −0.15% is a fourth consecutive session in which the banks have declined to trade the curve — JPMorgan +0.63%, Citigroup +0.41%, Bank of America +0.21% against Wells Fargo −0.09%, Morgan Stanley −0.12% and Schwab −0.27%, on a 2 bp fall in the 3M10Y spread. The alternative managers are the exception and they were the best-performing financials in the index: KKR +6.91%, Blackstone +3.89%, on the Nvidia mandate rather than on the curve.

Reconciliation. All eleven groups were checked against Monday’s published YTD compounded by Tuesday’s 1-day move: Energy 1.3323 × 1.0053 = +33.94% (exact); Basic Materials +16.91% (exact); Healthcare +8.33% (exact); Communication Services −1.15% (exact); Financial +8.23% (exact); Consumer Defensive +7.54% (exact); Technology +23.62% (exact); Consumer Cyclical −1.10% vs −1.09%; Industrials +16.00% vs 15.99%; Real Estate +8.79% vs 8.78%; Utilities +1.93% vs 1.94%. Maximum deviation 0.01 pt; seven groups reconcile exactly. This is the cleanest reconciliation of the month and it retires the Financial/Industrials/Utilities drift flagged on Monday. Source: Finviz Groups (Performance table view). Finviz buckets are not GICS/S&P sector indices — Alphabet and Meta sit in Communication Services and Amazon in Consumer Cyclical.

Up

KKR & Co (KKR) +6.91% to $111.00 — the best S&P 500 performer, one of the six firms Nvidia named to mobilise more than $500bn of third-party capital. Apollo rose more than 6% and Blackstone (BX) +3.89% to $147.17 on the same mandate (CNBC). The trade the market chose was the fee-earner, not the chipmaker — Nvidia itself closed −0.03%.

Axon Enterprise (AXON) +6.69% to $636.20 — the second-best name in the index and a continuation of Monday’s +4.43%; two sessions, +11.4%, with no new company catalyst in the reviewed sources.

Jabil (JBL) +5.94% to $356.61 — UBS upgraded to Buy; David Vogt: “We upgrade Jabil to Buy on a multiyear growth cycle fueled by AI investment from Amazon, Meta, and Google, rising healthcare demand as capacity comes online, and scaling automation and robotics markets.”

Marathon Petroleum (MPC) +5.01% to $336.37 — the best refiner for a second consecutive session, +12.4% over two days. Phillips 66 (PSX) +4.10% to $224.36 and Valero (VLO) +2.85% to $323.92 followed; the crack-spread trade survived a day in which the producers did not (§11).

Generac (GNRC) +4.74% to $215.86 — reversing Monday’s −2.89% in full and then some. Builders FirstSource (BLDR) +4.33% to $75.10, DR Horton (DHI) +2.82% to $150.79, Lennar (LEN) +2.28% to $87.55 — the three worst housing names of Monday were three of the best on Tuesday, on a 4 bp fall in the 3-year. Housing is currently a pure duration trade and it round-tripped in 48 hours.

KLA Corp (KLAC) +4.02% to $200.48 and Teradyne (TER) +3.96% to $379.56 — Monday’s −2.71% and −3.75% reversed. Semi-cap equipment led the SOX bounce.

Emerson (EMR) +3.57%, Eaton (ETN) +3.22%, Hubbell (HUBB) +2.92%, Rockwell Automation (ROK) +2.72%, Johnson Controls (JCI) +2.69% — the electrical-equipment cohort moved as a bloc, which is the second-derivative data-centre trade and the natural beneficiary of a $500bn financing announcement.

Enphase +3.37%, Arista Networks +3.31% to $197.85, Edison International +3.04%, Constellation Energy +2.89%, Charter +2.86%, Cencora +2.73%, Domino’s +2.73%, Hasbro +2.70%, EPAM +2.53%, McKesson +2.49% to $901.77, Keysight +2.47% (reports 8/18 AMC), CBRE +2.45%, Seagate +2.44%, ConocoPhillips +2.35%, Expedia +2.25%.

United Airlines (UAL) +2.06% to $126.31, Southwest (LUV) +1.38% to $45.52, Delta (DAL) +1.35% to $90.41 — the airlines rose on a day crude settled up 1.30%.

Duke Energy +1.65%, Lam Research +1.64%, Diamondback +1.38%, ONEOK +1.30%, Cardinal Health +1.30% to $240.26 (reported 8/11 BMO), NextEra +1.23%, Baker Hughes +1.17%, Vertex +1.10%, Home Depot +1.05% to $354.48 (reports 8/18 BMO), Chevron +0.90%, Corning +0.90%, Micron +0.84% to $868.25, EOG +0.83%, Meta +0.70% to $599.11, Super Micro +0.70%, Caterpillar +0.69%, Applied Materials +0.67% to $525.61, First Solar +0.66%, Uber +0.65%, JPMorgan +0.63%, Tesla +0.58%, Walmart +0.53%, Halliburton +0.51%.

After the close: Super Micro Computer (SMCI) +9.8% in postmarket trading (Bloomberg) / “more than 6%” (CNBC) on a first-quarter net-sales forecast above analyst estimates. CoreWeave (CRWV) +12% (Bloomberg) / +14% (CNBC) (not an S&P 500 constituent) on second-quarter adjusted operating margin of 5% against expectations. Two AI-infrastructure names beating on the same evening the Nvidia financing coalition was digested is a coherent read-across, and it is why futures were bid overnight.

Riot Platforms (RIOT) +~20% (not an S&P 500 constituent) — Q2 revenue $174.2m against a $154.3m FactSet consensus, plus a 191-megawatt data-centre lease with “a Leading Frontier AI Lab.” Note the reversal versus Monday’s after-hours −5.46%: the market has now decided the Anthropic contract is a positive.

Analyst actions elsewhere: Argus upgraded Boeing (BA) to Buy from Hold, $265 target, ~14% upside, on production ramp approvals, two flight certifications and the first positive cash flow since 2023; Kristina Ruggeri: “high demand and order backlogs, increased production, and a pivot to positive cash flow should support future earnings.” Boeing closed +0.19% at $233.24 — a 14%-upside upgrade that moved the stock 19 basis points. Truist upgraded Best Buy (BBY) to Buy from Hold, target to $95 from $81, ~15% upside, Scot Ciccarelli citing “continued replacement demand, internal changes (like appliance delivery) and emerging mini-product cycles (like AI wearables).”

Down

Amentum (AMTM) −8.18% to $22.44 — the worst S&P 500 performer, on its own 8:00 a.m. print (§5).

AppLovin (APP) −6.01% to $318.64 — Bank of America downgraded to Neutral and cut the price target to $400 from $430, saying risks to the revenue-growth forecast have increased.

Ventas (VTR) −5.43% to $86.96 and Welltower (WELL) −3.84% to $226.01 — no company catalyst identified in the reviewed sources (§2 item 8).

Honeywell (HON) −5.27% to $230.12 — the largest single drag on the Dow at −$12.80 of index price; attributed by the reviewed sources to updated 2026 guidance of $8.05–$8.35 EPS on $19.8–20.0bn of revenue, below consensus. Confirm with company IR before acting.

Alphabet A (GOOGL) −3.84% to $343.80 and Alphabet C (GOOG) −3.60% to $343.04 — the fourth losing session in five since the AI-division reshuffle was announced.

Deckers Outdoor (DECK) −3.72% to $93.82 — on a day Barclays cut Under Armour to Underweight and named Deckers among the brands it prefers. Being cited as a preferred alternative in a competitor’s downgrade is normally a positive; the stock fell 3.7%, which says the sector read-through outweighed the relative call.

Oracle (ORCL) −3.69% to $145.48 and Dell Technologies (DELL) −3.69% to $440.97 — the enterprise-hardware-and-software pair, down together on a day semi-cap equipment rose 4%.

Adobe −3.39%, Humana −3.28%, Gartner −3.05% (reversing Monday’s +4.08%), CF Industries −3.03% (reversing Monday’s +6.24%), DaVita −2.91%, Allstate −2.76%, Tapestry −2.57% (reports 8/13 BMO), Centene −2.54%.

Berkshire Hathaway B (BRK.B) −2.46% to $516.38 — giving back Monday’s +1.46% and more, two sessions after the operating-earnings beat. A post-print fade of this size in the largest non-tech name in the index is worth noting: the buyback and the first net equity purchase in fifteen quarters bought one session of outperformance.

HP Inc −2.45%, Universal Health Services −2.33%, Freeport-McMoRan −2.33% (on a day copper rose 0.19% — the miner did not trade the metal), CVS Health −2.26%, Ralph Lauren −2.25%, Molina −2.16%, CH Robinson −2.15%.

Amazon (AMZN) −2.10% to $272.26 — the third-largest Dow drag at −$5.84 and the reason Consumer Cyclical finished −0.59%.

Fox Corp A −2.02%, Assurant −2.02%, Netflix −1.97%, Nike −1.96%, VeriSign −1.95%, TJX −1.94% (reports 8/19 BMO), Dollar General −1.92%, Bristol-Myers −1.90%, Cisco −1.72% to $120.46 (reports 8/12 AMC), UnitedHealth −1.60%, Broadcom −1.50% to $416.08, CrowdStrike −1.44% (reversing Monday’s +5.01%), Eli Lilly −1.37%, Apple −1.09% to $304.91, Costco −0.88%.

On Holding (ONON) −20%+ (not an S&P 500 constituent) — the worst day on record. Q2 revenue CHF 850.3m against a CHF 878.4m consensus; full-year revenue growth guided to the low-20% range against a prior “at least 23%.” The stock is down nearly 34% in 2026.

Intel (INTC) +0.19% to $97.71 — listed here for the catalyst rather than the direction: the common stock offering announced Monday was upsized to $20bn from $15bn. The stock fell 4.06% on the $15bn announcement and rose 0.19% on the $20bn upsize. A 33% larger raise absorbed with a green close is the clearest evidence in this report that the equity market’s appetite to fund AI capex has not cracked (§9).

Airbnb (ABNB) +0.15% — PhillipCapital downgraded on valuation while raising the target to $158, roughly 14% below Monday’s close; Paul Chew: “Recent rally has pushed ABNB to a premium valuation, trading at 30.9x PE versus its 2-year historical +1 SD of 29.6x.” A downgrade with a target 14% below spot moved the stock 15 basis points — the second analyst call of the day the tape ignored.

Times are ET; S&P 500 components only. Every day page 8/12–8/21 was pulled from Earnings Whispers this session and screened against the Investing.com U.S.-500 constituent board. Re-verify times and membership against company IR before trading any date.

Mon 8/10 — completed. BMO: Berkshire Hathaway B (BRK.B) 8:00 — closed +1.46% at $529.42 on the day, then −2.46% to $516.38 on Tuesday, giving the move back with interest. AMC: Simon Property Group (SPG) 4:05 — closed −0.46% at $219.53 the following session.

Tue 8/11 — completed. BMO: Cardinal Health (CAH) 6:45 — closed +1.30% at $240.26; Amentum (AMTM) 8:00 — closed −8.18% at $22.44, the worst performer in the index. AMC: Lumentum (LITE) 4:00; Super Micro Computer (SMCI) 4:05 — closed +0.70% at $31.68 in the regular session, then +9.8% in postmarket trading (Bloomberg) on a first-quarter net-sales forecast above estimates; CNBC marked the move at “more than 6%.”

Wed 8/12. BMO: Amcor (AMCR) 6:00, Trimble (TRMB) 6:55. AMC: Cisco (CSCO) 4:05, Coherent (COHR) 4:05.

Thu 8/13. BMO: Tapestry (TPR) 6:45. AMC: Applied Materials (AMAT) 4:00.

Fri 8/14. Neither page lists an S&P 500 reporter — the before-open page is entirely micro-cap, biotech and broadcasting, and none of the reviewed names is carried by the constituent board.

Mon 8/17. Neither page lists an S&P 500 reporter (ten names screened across both buckets).

Tue 8/18. BMO: Home Depot (HD) 6:00. AMC: Keysight Technologies (KEYS) 4:05, Jack Henry & Associates (JKHY) 4:15.

Wed 8/19. BMO: Lowe’s (LOW) 6:00, Estée Lauder (EL) 6:00, Target (TGT) 6:30, Analog Devices (ADI) 7:00, TJX Companies (TJX) 7:30. AMC: Nordson (NDSN) 4:30.

Thu 8/20. BMO: Deere & Company (DE) 6:20, Walmart (WMT) 7:00. AMC: Ross Stores (ROST) 4:00.

Fri 8/21. Neither page lists an S&P 500 reporter (four names screened before the open; the after-close page returned “NONE”).

Changes vs. the prior calendar (8/10 report)

No additions and no removals. Every 8/12–8/21 name carried by the prior edition re-appeared on this session’s pull with an identical timestamp: AMCR 6:00, TRMB 6:55, CSCO 4:05, COHR 4:05, TPR 6:45, AMAT 4:00, HD 6:00, KEYS 4:05, JKHY 4:15, LOW 6:00, EL 6:00, TGT 6:30, ADI 7:00, TJX 7:30, NDSN 4:30, DE 6:20, WMT 7:00, ROST 4:00. CAH, AMTM, LITE and SMCI all reported as scheduled.

Count correction, disclosed rather than buried. The prior edition described next week as “thirteen S&P 500 reporters across four sessions”; the names it actually listed number twelve across three sessions, and this session’s independent re-pull confirms twelve across three (8/18, 8/19, 8/20). The corrected count is used above.

Membership caveat, restated rather than buried: the constituent board used as this report’s screening proxy does not carry Lumentum (LITE) or Coherent (COHR). Both have been carried by the prior four editions and are retained for continuity; they remain the two least certain names in this week’s list — confirm with company IR before trading either date.

Conservative exclusions on this pull: Toll Brothers (TOL) and Mercury Systems (MRCY) on 8/18; Coty (COTY), BILL Holdings (BILL) and Wolfspeed (WOLF) on 8/19; Flowers Foods (FLO), Advance Auto Parts (AAP) and OSI Systems (OSIS) on 8/20; BJ’s Wholesale (BJ) on 8/21; Fabrinet (FN) on 8/17; Brinker (EAT) on 8/12; Applied Industrial Technologies (AIT) on 8/13. Toll Brothers remains the most borderline. The full excluded list is in the companion Data Notes file.

Timing bucket still unpublished: none. Every name above carries a specific clock time.

What the forward calendar hands the desk. Cisco (8/12 AMC) is the first large-cap networking read since the Nvidia financing announcement and the direct check on the semi-cap-versus-enterprise split described in §3 — the stock fell 1.72% into it. Coherent (8/12 AMC) follows Lumentum by one night and completes the photonics pair. Applied Materials (8/13 AMC) is the single most important scheduled equity event in the window: it rose 0.67% on a day the SOX gained 0.87%, having fallen 3.16% on Monday. And next week concentrates six U.S. consumer reporters — Home Depot, Lowe’s, Target, TJX, Walmart and Ross Stores — into three sessions, the densest consumer read of the quarter.

Cross-checks. Bloomberg marked the U.S. 10-year at 4.69%, −2 bp, at 4:59 p.m. ET; WSJ’s real-time page at 5:05 p.m. showed 10-year 4.696% (−1.7 bp), 30-year 5.242%, 7-year 4.533%, 5-year 4.391%, 3-year 4.293%, 2-year 4.226%, 1-year 4.024%, 3-month bill 3.813% (+0.2 bp). The one visible disagreement is at the 3-year, where the official par curve shows −4 bp and WSJ −2.0 bp; that is a baseline artefact — WSJ strikes its change off its own Monday real-time mark and Treasury off Monday’s 3:30 p.m. par fixing — and the two levels agree to 2.3 bp. WSJ’s 3-month bill at 3.813% against the par curve’s 3.89% is the usual bank-discount-versus-coupon-equivalent gap on a bill quote.

The read — a front-led bull steepener with the long end anchored, and that is a policy-path repricing, the exact mirror image of Monday. The largest move on the curve is the 3-year at −4 bp, then the 2-year at −3 bp, then 5s/7s/10s at −2 bp, the 30-year at −1 bp and the 20-year and the 3-month bill at zero. Monday’s shape was the inverse: +7 bp at 7s and 10s, +6 bp at 2s/3s/5s and the 30-year, +2 bp at the bill. Two sessions, two opposite shapes, and the diagnostic is in the spreads: 3M10Y narrowed 2 bp to 81 bp while 2s30s widened 2 bp to 102 bp. A curve where the point of maximum sensitivity is three years out — the tenor that prices the next four to six FOMC meetings and nothing else — is repricing the reaction function, not the inflation premium. That is consistent with everything else on the board: CME’s September hold went back to modal at 52.0% (§8), the crude rally faded 1.63% off its high (§11) and the front end took back a third of Monday’s move while the 20-year took back none of it.

Two further observations. First, the week-over-week picture is the honest one and it is unambiguously bearish. Against 4 August the 10-year is +7 bp, the 20-year +7 bp, the 7-year +7 bp, the 30-year +6 bp and the 5-year +6 bp, while the 2-year and 3-year are only +2 bp and the 3-month bill is exactly unchanged. One session of front-end relief has not dented a week in which the curve added 6–7 bp of term premium beyond three years. 2s10s at 48 bp is 5 bp steeper on the week and the steepest since this report began tracking it; 2s30s at 102 bp is the first triple-digit print of the month. The market is not pricing a Fed that eases — it is pricing a Fed that may hike and a long end that has to fund it. Second, the bill is now the most informative point on the curve precisely because it will not move. The 3-month has printed 3.89% for three consecutive sessions and is unchanged on the week while the 10-year is 7 bp higher. That is a financing signal and it belongs with the funding data in §9: cash that used to sit at the RRP is buying bills, which pins the front of the bill curve regardless of what the coupon curve does. The configuration to watch is the one where the bill finally moves — that would mean the bid has been exhausted, and it is the earliest available warning on the September quarter-end.

Source: New York Fed Economic Indicators Calendar, August 2026 (all times ET), with consensus figures from the WSJ U.S. Economic Calendar and CNBC (Dow Jones poll), and actuals from the primary agency or the wire summary where released. Sensitivity is this report’s own rating and is stated explicitly for every line, because the “Very high” tag is what drives the §1 tape rule.

Current week (Aug 10–14)

Next week (Aug 17–21)

Current target range: 3.50–3.75%. EFFR 3.63% (NY Fed, 10 August).

CME FedWatch — next meeting (16 September 2026), four-column headline

Provenance of every column, stated. CME published a complete numeric four-column table under the September chart on this pull, so all four columns are read live from CME rather than reconstructed. The table’s own footnote reads “Data as of 11 Aug 2026 05:03:29 CT”, and the page’s related-futures block is stamped “Last Updated 11 Aug 2026 05:14 PM CT”; the historical columns are CME’s own end-of-day settlement snapshots for the dates shown.

All four columns sum to exactly 100.0%.

Correction to the prior edition, disclosed rather than buried. The 10 August report published CME’s September card as 48.8% hold / 51.2% hike, read live at approximately 10:00 p.m. ET on the Monday. CME’s own settled 1-DAY column now shows 47.8% / 52.2% for 10 August. The prior edition’s number was 1.0 point too dovish, for the same reason as the 7 August correction: a live read taken after 5:00 p.m. ET is indicative, not settled. The pattern is now established across two consecutive editions and the fix is procedural — every historical comparison in this section uses CME’s settled columns, and the live NOW figure is labelled as such.

The move, and the momentum. The hold is modal again: 52.0% against a settled 47.8% on Monday, a 4.2-point swing back toward the hold and the first time the hold has been modal since Friday’s settlement. On the week the picture is larger and it runs the same way: 41.6% on 4 August to 52.0% now, +10.4 points. On the month it is larger still: 30.4% on 10 July to 52.0%, +21.6 points, and the +50 bp bucket has gone from 18.4% to exactly zero. Set the three horizons side by side and the arc is unambiguous — a month ago the market priced a hiking cycle with a second increment; a week ago it priced one hike as the base case at 58.4%; today it is a coin flip that leans to no hike at all. The three-session sequence inside that is 41.6% → 47.8% → 52.0%, a monotone recovery in the hold that has survived both the crude spike and the crude fade. What did the damage to the hike case was Friday’s payroll contraction; what has kept the recovery going since is the absence of any inflation datapoint to contradict it. That absence ends at 8:30 tomorrow.

The CME-versus-Investing.com gap, reconciled and quantified

Investing.com’s Fed Rate Monitor, timestamped Aug 11, 2026 05:55 PM EDT, shows the September card at 52.6% hold / 47.4% hike against CME’s 52.0% / 48.0% — a 0.6-point gap, down from 2.1 points on Monday, and it is entirely a contract-price artefact. Work it through: Investing.com quotes the September contract at 96.310, an implied average September effective rate of 3.690%. With EFFR at 3.63% and the meeting on the 16th, 15 of September’s 30 days carry the new rate, so a certain hike lifts the month’s average by 0.25 × 15/30 = 12.5 bp and the implied hike probability is (3.690 − 3.630) / 0.125 = 48.0% — exactly CME’s published figure, from Investing.com’s own quoted price. Investing.com’s published 47.4% corresponds instead to a price of 96.3108, eight hundredths of a basis point higher. The entire vendor disagreement is 0.08 bp of contract price. The amplifier is the mid-month meeting date: because only half the contract month is affected, one basis point of ZQU6 is worth about eight points of headline probability. That is the single most useful number in this section — it is why a 4-point “swing” and a 0.6-point vendor gap can both be economically trivial, and why a 1.5 bp move in the contract on Wednesday morning is a 12-point move in the headline. Use one vendor’s columns consistently: CME’s four-column table for the headline, Investing.com’s cards for the per-meeting matrix.

(a) 2026 meeting distributions — Investing.com matrix, current / [prev day] / [prev week]

Highlighted cell = modal range. Rows sum to 100.0% (September), 100.1% (October, December).

The multi-day momentum read, which is the part a single day-on-day figure hides. Take the three sessions together and the shape is a V that has now more than closed at the front and only half closed at the back. September’s hold was 42.3% a week ago, 50.9% on Monday and 52.6% now — it is 10.3 points more dovish on the week and has taken back Monday’s entire crude-driven hawkish move and then some. October’s cumulative hike probability is 62.0% against 64.4% on Monday and 68.4% a week ago — 2.4 points of relief on the day against 6.4 on the week. December’s is 78.5% against 80.4% and 81.6% — 1.9 points on the day and 3.1 on the week. The gradient is the whole message: the front of the strip has recovered 10.3 points of dovishness on the week, the middle 6.4 and the back 3.1. A crude shock that lifted the whole strip on Monday has been unwound in inverse proportion to maturity, which means the market treated it as a timing event, not a level event — the same conclusion the curve reached independently in §6, where the 3-year fell 4 bp and the 20-year did not move at all. Two markets, two instruments, one diagnosis. The one number that has still not moved in either direction is the 2026 cut probability, which has printed 0.0% for the seventeenth consecutive session — a payroll contraction, a 5% crude spike and a crude fade have all left it untouched, and it remains the most crowded consensus in this report (§13).

(b) 2027 meeting path — modal range, cumulative above/below, and the contract prices that draw the terminal

The terminal came back in, and the March meeting changed its mind. The strip’s low price is 95.925 at the 28 July 2027 meeting, an implied 4.075%, against 95.900 and 4.100% in the prior edition — the market removed 2.5 bp of terminal in one session, giving back well over half of Monday’s 4 bp addition, while leaving the peak meeting where it was. The second change is at March 2027, where the modal range flipped back from 4.00–4.25% to 3.75–4.00% (34.8% against 33.5%) — the closest call anywhere on the strip, decided by 1.3 points. The peak-to-December-2027 give-back is 5.0 bp (95.925 to 95.975), so the strip still prices a shallow easing cycle beginning in the fourth quarter of 2027 and nothing before it. The cumulative-below column is the honest measure of how little the market believes in cuts: exactly 0.0% at every 2026 meeting and at the first four meetings of 2027, and it does not exceed 1.5% until October 2027. Note also that the cumulative-above column peaks in June–July 2027 at 88.8%/88.1% and then declines to 80.4% by December — the market’s conviction that rates end 2027 above today’s range is 8 points lower than its conviction that they pass above it on the way.

(c) Year-end probability ladders

Year-end 2026 (9 December 2026 meeting), from a 3.50–3.75% starting range:

Year-end 2027 (8 December 2027 meeting):

Rounding, stated transparently. The September 2026 current column and the January, March and April 2027 current columns sum to exactly 100.0%; October 2026, December 2026, June 2027, July 2027 and October 2027 sum to 100.1%; the September 2027 and December 2027 current columns sum to 99.8%. These are vendor rounding artefacts at one decimal place and no cell has been adjusted. The 28 July 2027 and 15 September 2027 cards show an em-dash in the prev-day column for the lowest populated range, which is the vendor’s convention for a bucket that was not populated in the prior snapshot rather than a zero. The end-2027 cut tail is 3.3% against 3.1% on Monday and 5.2% a week ago — it recovered 0.2 points on the session and remains a third below where it started the week.

(a) IG and HY credit spreads

FRED publishes the ICE BofA OAS series with a one-business-day lag. The table below is as of Monday 10 August 2026 — it is not a same-day mark, and Tuesday’s direction is addressed beneath it from single-name CDS and index evidence.

CDX retrieval note — the six-step ladder was worked again and the index level remains publicly unobtainable, but step 1 produced a genuine credit datapoint for the first time this month. (1) Bloomberg in Chrome: /markets/rates-bonds publishes the Bloomberg Fixed Income Indices and a full sovereign 10-year grid but no CDX quote; however, the credit story “Nvidia’s Show of Financial Force Soothes Jittery Credit Markets” carries an explicit five-year single-name CDS level and daily change, reproduced in the table. (2) WSJ Market Data Bonds & Rates: Treasurys, money rates, a sovereign grid and the story “Treasury Yields Cool Down Ahead of July CPI” — no CDX. (3) ICE / S&P Global / Cbonds index pages: the CDX.NA.IG 5Y page was rendered directly this session and confirms the index description and methodology but publishes no level without entitlement. (4) FT Markets Data / Reuters credit wraps: no CDX level in the reviewed 11 August material. (5) TradingView / Barchart / CME CDS product pages: symbols carried, prices masked. (6) Cash-market proxies, labelled as such: the Bloomberg U.S. Aggregate rose 2.04 points on a session in which Treasury yields fell only 1–2 bp across the coupon curve; a duration-only explanation would produce a smaller gain, so the residual is a modest spread tightening. All six steps are named so the gap is auditable. Quoting convention reminder: CDX IG 5y is quoted in basis points of spread and CDX HY 5y in price points, where a rising price means tightening credit spreads.

The direction cross-check, and the one number that actually moved. The index-level credit spread did nothing for a second session: IG 78 bp for the sixth consecutive print, HY 270 bp for the second, CCC 1,014 bp against 1,013 bp. On a weekly basis the tail is still leading — CCC −14 bp against HY −8 bp and IG unchanged — and the CCC-minus-HY differential is 744 bp against 750 bp on 3 August and 604 bp at the start of the year. That differential has narrowed in three of the last five sessions and widened in two — 750 → 746 → 748 → 746 → 743 → 744 across 3–10 August — so the trend is intact but it is not the clean four-session run the prior edition implied. But the tradable credit event of the day was not in any index. Nvidia’s five-year CDS at roughly 73 bp, down about 4 bp, matters more than a static 78 bp IG OAS for three reasons: it is the largest company in the world by market value, its debt is a large and growing IG index constituent, and the thing being priced — vendor financing to customers who may not be able to pay — is the single most concentrated credit risk in the U.S. corporate market. A near-doubling from about 41 bp to 82 bp in under three weeks and then a 9 bp retracement in two sessions on a financing announcement is not idiosyncratic noise; it is the credit market repricing the AI capex chain in real time, invisibly to an index that sits at 78 bp because 500 other issuers have not moved.

(b) Money-market and funding plumbing

NY Fed publishes reference rates at ~8:00 a.m. ET for the prior business day; SOFR and EFFR below are as of Monday 10 August 2026. ON RRP is same-day (11 August).

The distribution is the story, and it points the other way from Friday. Friday’s secured market translated cleanly downward — every percentile fell together, which is a supply-of-cash effect. Monday’s did not: the mean rose 1 bp to 3.63%, the 25th percentile rose 1 bp, the 75th and 99th were unchanged and the 1st percentile fell 1 bp, so the 99th-minus-1st span widened from 12 bp to 13 bp. A widening distribution with a rising mean and a falling floor is the early signature of dispersion between cash-rich and cash-poor counterparties, and it is the first such print since the last week of July. It is small — 1 bp — and it would not be worth a paragraph in a normal month. It is worth one now because the buffer is gone: ON RRP printed $0.975bn on Monday, the first sub-$1bn reading the series has ever produced, and recovered only to $1.250bn on Tuesday. With the facility empty and reserves $150bn below the July peak, dispersion is the variable to watch rather than the level, because the level will look fine right up until it does not. EFFR volume falling to $108bn on a day the unsecured percentiles did not move at all says the same thing from the other side: the unsecured market is not clearing less because it is stressed, it is clearing less because there is nothing to arbitrage. The one scheduled release that speaks directly to this is the NY Fed’s Reserve Demand Elasticity print on 20 August (§7).

(c) Rates volatility and swap spreads

Read the vol complex honestly, because one of the two series did not update. The clean fact is the VIX: 15.28, down 1.16%, on a session the S&P fell 0.32%, twelve hours before July CPI, with the September future at 16.78 — a 1.50-point contango. A spot index that declines on a down day into a scheduled binary event is an option market that is not paying for the event; the contango says the term structure is, which means the demand is for September rather than for tomorrow morning. That is a very specific mispricing and §12 expresses it. The unclean fact is the MOVE, which shows 0.00% change on a 10 August stamp — the delayed series simply did not publish an 11 August value, so the 4.94× ratio pairs a Tuesday numerator with a Monday denominator and should not be traded off. The honest gap in this table remains swap spreads, and it matters more than usual this week: a term-premium argument (§6, where the 20-year did not rally at all) and a reserve-dispersion argument are exactly the two forces that show up first in the 30-year swap spread, and this report cannot source it from public data.

(d) Issuance, leveraged loans and private credit

The credit event of the session was a $500bn financing coalition, and it worked. Nvidia enlisted Apollo, Blackstone, BlackRock, Brookfield, Goldman Sachs and KKR to mobilise more than $500bn of third-party capital for the AI build-out, with the group independently judging individual deals and Nvidia’s own contribution “relatively limited, and only factor[ing] into some deals.” The credit market’s verdict was immediate: five-year protection on Nvidia fell about 4 bp to roughly 73 bp and its bonds rallied back to last week’s risk premiums (Bloomberg). The structural read: this is the AI capex chain moving from vendor balance sheets to third-party capital, which is bullish for the IG index because it keeps the debt out of it, and it is the single most important thing that has happened to the credit-versus-equity divergence in a month.

Intel upsized its common stock offering to $20bn from $15bn and the shares closed green. On Monday the $15bn announcement cost the stock 4.06%; on Tuesday a 33% larger raise was absorbed at +0.19%. Equity funding at that scale, twice in two days, is the second data point in the same direction as Nvidia’s coalition — AI capex is being funded outside the bond market by choice. That caps how much of the build-out ends up in the IG index and it is a real, if slow-burning, technical support for IG credit spreads.

The private-credit warning is now on WSJ’s own bonds page. “Private Credit Is Under Growing Strain, Despite Industry’s Upbeat Tone” (WSJ, 10 August) sits directly above the Treasury table, and it is the third consecutive week in which a mainstream outlet has led its fixed-income coverage with private-credit stress rather than with rates. Keep it on the watch list alongside data-centre project finance: Riot Platforms’ 191-megawatt, twenty-year lease to “a Leading Frontier AI Lab” was marked down 5.46% after hours on Monday and up roughly 20% on Tuesday — the same contract, two opposite verdicts in 24 hours, which is what an unpriced asset class looks like.

The after-close AI-infrastructure prints are the credit read-across nobody indexes. Super Micro +9.8% postmarket on a first-quarter sales forecast above estimates and CoreWeave +12% on a 5% adjusted operating margin. CoreWeave in particular is the leveraged expression of the whole chain, and its cost of capital is the transmission line between Nvidia’s CDS and the physical build-out.

Morningstar LSTA leveraged loan index and bank CDS: not obtainable this session. Sources attempted and named so the gap is auditable: Bloomberg /markets and /markets/rates-bonds, the Bloomberg Markets Wrap and credit story for 11 August, WSJ Market Data Bonds & Rates, Cbonds’ CDX.NA.IG 5Y page, and the Investing.com and TradingEconomics boards. None published a loan-index level or a bank CDS quote for 11 August. The Finviz Financial group at −0.15% on a 2 bp bull-steepening day is the only bank-risk proxy this report can source tonight, and it is an equity proxy, not a credit one.

Quote basis. All pairs are spot in the market convention shown; a positive move on a USD/XXX pair means the dollar strengthened. The TradingEconomics pull landed at approximately 18:20 ET on the 11th, after that vendor’s daily boundary had rolled, and its %Chg column printed between −0.38% and +0.10% for the entire board — that column is NOT reproduced. The “24h” column is computed against the same vendor’s levels published in the 10 August edition at approximately the same hour; worked examples are in the companion Data Notes file. Investing.com’s real-time board, pulled at 18:23–18:24 ET, is the independent cross-check.

The take — the dollar did nothing, and on this particular Tuesday that is the trade. Start with the arithmetic. The 3-year Treasury yield fell 4 bp, the 2-year 3 bp, and CME’s September hold probability rose 4.2 points to become modal again. Every one of those is a dollar-negative input, and the dollar index closed unchanged to the second decimal on two independent vendors with an intraday range of 0.03 points. Compare Monday, when a 7 bp rise in the 10-year bought a 0.21% dollar gain, and Friday, when a 6 bp fall in the 2-year bought a 0.33% loss. Two sessions of clean rate-beta and then nothing. The reading is not that the relationship broke; it is that the FX market has stopped taking a view before an 8:30 print — it is flat, and flat is a position. The practical consequence: whatever CPI does, the dollar’s move will be larger than the rate move implies, because there is no inventory on either side to absorb it.

Two crosses did something and both are second-order. The first is the won. USD/KRW fell 0.26% — the won strengthened — on the day the Kospi opened −0.76% and closed +0.73%, having traded a 3.1% intraday range. On Monday the won weakened 0.66% on a day the Kosdaq rose 7%. Take the two together: the currency strengthens when the large-cap index reverses on foreign-led semiconductor flow, and weakens when the small-cap index melts up on domestic leverage. That is a clean discriminator between the two kinds of Korean equity rally, available in real time in a market that trades around the clock.

The second is the franc, and it is now a pattern rather than an observation. USD/CHF rose 0.07%, so the franc weakened for a second consecutive session — this time with no oil shock, no VIX spike and an equity decline of 0.32%. A haven that will not bid on a geopolitical tape, will not bid on an equity decline, and will not bid twelve hours before the most important inflation print of the quarter is not being treated as a haven at all. Set it against gold, which rose for a fifth consecutive session (§11): the market is buying the non-yielding real asset and selling the low-yielding currency, which is the inflation-hedge configuration rather than the risk-off configuration. That distinction has now held for three sessions and it is the most durable cross-asset signal in this report. And the one that should have moved and did not: USD/CAD at −0.10% on a 1.30% crude settle — the magnitude is one thirteenth of the commodity move. Anyone still using CAD as a crude proxy is running an unhedged rates position; the Bank of Canada’s own 10-year fell 1 bp against the U.S. 2 bp, so the differential moved 1 bp and the currency expressed exactly that.

Basis note. All futures rows are 11 August settlements derived from the Investing.com real-time futures board pulled at 18:20–18:25 ET (last price at the 17:00 stamp, with the board’s own change column), with the contract month shown. Control checks: WTI September derives to $83.23 against CNBC’s published $83.20 settlement, and the prior-day back-out reproduces Monday’s $82.13 exactly; Comex December gold back-solves Monday to $4,419.70, reproducing the prior edition to the cent. Percentage changes are struck against the 10 August settlements, so the basis is consistent front-month-to-front-month.

Withheld figures, with the reason stated in-line. The TradingEconomics %Chg column is not reproduced anywhere in this table: its gold row shows an absolute change of +17.52 alongside a percentage change of −0.40% in the same row with an “06:15” stamp, which is internally inconsistent, and its silver row does the same. The vendor’s YTD column is reproduced only where the level agrees with the settlement derivation — crude +45.34%, Brent +46.71%, gasoline +84.14%, heating oil +101.66%, gold +1.28%, silver −9.05%, copper +16.43%, natural gas −25.37%, platinum −15.14% — and note that TradingEconomics’ crude level of $83.456 agrees with the $83.20 settlement to within 0.31%, inside this report’s ~1% tolerance for a third consecutive session.

The take — the barrel could not hold its high, the product complex kept the crack, and gold is now the crowded trade. Positioning first, on energy, because the order of the move inverted. Monday: heating oil +7.28%, gasoline +5.89%, crude +5.05% — products led. Tuesday: heating oil +2.04%, crude +1.30%, gasoline +0.41% — distillate led and gasoline lagged by 89 bp. That split is informative. Distillate is the transit-exposed product and it kept outperforming; gasoline is the demand-exposed product and it stopped, on the day the airlines rallied and a peace headline hit the tape. The 3-2-1 crack is therefore now a distillate crack rather than a barrel crack, and the equity confirmed it: Marathon Petroleum +5.01% and Phillips 66 +4.10% against APA −0.98% and Occidental flat. The refiners have outperformed the producers by roughly 5 points in two sessions. The positioning risk is unchanged and violent: a confirmed reopening compresses the distillate crack faster than it compresses flat price, because the logistics premium unwinds instantly. The session’s own price action is the warning — $84.61 to $83.23 on a single sentence from a Pakistani defence minister.

On metals, the complex fractured after one day of moving together. Gold +0.18%, copper +0.19%, platinum +0.04%, silver −0.63%, palladium −0.99%. Compare Monday: silver +2.23%, copper +0.99%, gold +0.42%, platinum +0.37%, palladium +0.33% — all five green and silver leading. Silver has now outrun gold on four of the last six sessions and underperformed it on the other two, which is what a high-beta expression looks like when the underlying trend pauses. The gold-silver ratio widened back to 68.3× from 67.7×. The more important input is positioning: Bespoke’s finding that gold closed a full standard deviation above its 50-day moving average for the first time in 103 trading days, and entered overbought territory for the first time since 10 March, is the first genuine positioning warning on the metal this cycle. Gold is up roughly 8% in August alone. This report has been constructive on the inflation-hedge bid since Friday and that view is unchanged, but the entry is now materially worse than it was and §12 sizes accordingly.

The basis caveats, explicitly. Gold is quoted on two bases that differ by roughly $54, or 1.23%: the Comex December futures settlement of $4,427.80 and the TradingEconomics spot mark of ~$4,373.82. That gap is carry — four months of financing at roughly 3.6% on a $4,374 metal is approximately $52 — so the December basis is consistent with the curve and is not a data conflict. Silver is quoted on the September contract; copper on the September Comex contract in USD/lb, and the LME three-month grade-A contract on the same board marks $14,154/t, a different instrument on a different exchange that should not be compared directly. Dutch TTF is the September contract and its 16:41 stamp shows a 0.00% change, so the −2.6% quoted above is computed against the prior edition’s live mark and is directional only. The one open discrepancy is copper: this session’s board back-solves Monday’s September settlement to $6.6160 while the prior edition published $6.6358, a 0.30% gap; the 1-day change of +0.19% is quoted from the board’s own change column and is internally consistent, and the discrepancy is recorded rather than smoothed.

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

1. The rates trade — hold the long-back/short-front fed funds spread through CPI, and mark the recovery honestly

Honest mark on the prior edition’s rates idea. The book is long ZQZ7 against short ZQU6, DV01-matched one-for-one, carried at two-thirds size since Monday. It made money on Tuesday and it has recovered 1.5 bp of the 3.5 bp it lost on Monday. ZQZ7 rose from 95.955 to 95.975 (+2.0 bp) while ZQU6 rose only from 96.305 to 96.310 (+0.5 bp), so the DV01-matched spread gained 1.5 bp and is now approximately 2.0 bp offside since inception at 96.000 / 96.315. The stated invalidations — July core CPI at or above +0.4% m/m, or ZQN7 below 95.850 — have not been triggered; the terminal contract rose from 95.900 to 95.925. The thesis was that Monday’s terminal re-rating was a crude artefact rather than a trend, and one session of crude fading took back 2.5 bp of the 4 bp it added, which is consistent with the thesis without yet proving it.

Modal path, base case and tails — this is what the trade expresses. Modal path: no hike on 16 September on both vendors for the first time since Friday (CME 52.0% hold, Investing.com 52.6%), with the vendors now 0.6 points apart rather than 2.1 (§8). One 25 bp hike delivered in 2026 with December carrying 78.5% cumulative against October’s 62.0%; year-end 2026 modal 3.75–4.00% at 44.2%; terminal 4.075% at the July 2027 meeting; year-end 2027 modal 3.75–4.00% at 31.4% with 4.00–4.25% 1.9 points behind. Base case: an economy that has stopped hiring, is absorbing a crude shock that is fading, and has a small-business survey saying hiring plans just hit a four-year high — a mix that keeps a central bank on hold for longer than either tail implies. Hawkish tail: core CPI at or above +0.4% m/m, worth roughly 1.5 bp on ZQU6 and therefore about 12 points of September probability, dragging the whole 2027 strip. Dovish tail: claims above 240,000 on 8/13 with a soft PPI, which would produce the first non-zero 2026 cut probability of the year — currently 0.0% for a seventeenth consecutive session.

Expression, unchanged: long ZQZ7 against short ZQU6, DV01-matched one-for-one (both legs 30-day fed funds futures, DV01 $41.67 per basis point per contract), carried at two-thirds of the original size into the print. Rationale for holding through CPI rather than flattening: ZQU6 at 96.310 has roughly 6.0 bp of upside to a certain hold and 6.5 bp of downside to a certain hike, so the short leg is close to symmetric; the long leg carries the 2027 easing optionality that a hot print destroys and a soft print re-prices, and it is 4.5 bp cheaper than at inception. Catalyst: CPI 8/12 08:30; PPI and claims 8/13 08:30; retail sales 8/14 08:30. Invalidation, unchanged: July core CPI at or above +0.4% m/m; or the 28 July 2027 contract trading below 95.850. Sizing: small; size on roughly 6 bp of maximum adverse move in the front leg and note the spread’s carry is slightly negative.

2. Buy the event, not the direction — own short-dated S&P optionality into CPI

Expression: a short-dated (one- to two-week) S&P 500 strangle or a long straddle rolled off the 8/12 open, or long VIX September calls funded by short August. Thesis: the VIX fell 1.16% to 15.28 on a down day, twelve hours before the single “Very high” release of the fortnight, with the September future at 16.78 — a 1.50-point contango that says the demand for protection is for next month, not for tomorrow morning (§9). At 15.28, one-month implied prices a daily move of roughly 0.96%; §8’s arithmetic says a 1.5 bp move in ZQU6 is a 12-point move in the September probability, and both vendors are sitting within 3 points of a coin flip. An event with a 50/50 distribution and a 12-point sensitivity to 1.5 bp of contract price is not a 0.96% day. Catalyst: CPI, 8/12 at 8:30. Invalidation: the print landing exactly on consensus with a core at +0.2%, in which case the vol is a straight decay loss — size it as a premium-at-risk trade, not a delta trade. Sizing: small, defined-risk. Note the skew: §7 argues a soft print produces a smaller rally than a hot print produces a selloff, so a put-side tilt is cheaper than a symmetric structure.

3. Own the distillate crack, not the barrel — long refiners against short crude

Expression: long a basket of Marathon Petroleum, Phillips 66 and Valero against a short in September WTI, sized so the futures leg matches the basket’s crude beta. Thesis: the trade worked on Tuesday and the internal composition improved. Heating oil +2.04% against crude +1.30% and gasoline +0.41%, and the equity followed exactly — MPC +5.01%, PSX +4.10%, VLO +2.85% against APA −0.98% and OXY flat (§11). Distillate is the transit-exposed product; gasoline is the demand-exposed product; the divergence between them on Tuesday is the cleanest evidence yet that the market is pricing a logistics disruption rather than a reserve disruption. Catalyst: the EIA weekly on 8/12 at 10:30 — distillate inventories are the direct test. Invalidation: a confirmed Hormuz reopening announcement, which compresses the crack faster than the barrel and loses on both legs; or a distillate build above 3m barrels. Sizing: medium, unchanged. Two sessions, roughly +5 points of refiner-versus-producer outperformance.

4. Fade the crude-consumer short — cover the airlines

Expression: close or substantially reduce the short-airline/short-cruise basket against long S&P futures. Honest mark: the 10 August idea was short Southwest, United, Carnival, Royal Caribbean and Norwegian against long index, beta-adjusted; it made roughly 3.5–4 points on Monday and gave back roughly 1.3 points on Tuesday, when UAL rose 2.06%, LUV 1.38% and DAL 1.35% on a day crude settled higher. Thesis for cutting: the group rallying into a rising barrel means it is no longer trading fuel — it is trading the reopening headline, and the reopening headline is a binary the short leg cannot survive. The stated invalidation was WTI below $78.18, which has not triggered, so this is a discretionary size reduction on a change in the driver rather than a stop. The remaining position should be the cruise lines only, which did not participate in Tuesday’s rebound (CCL −0.09%, RCL −0.28%) and whose bunker exposure is to the distillate complex that is still outperforming. Catalyst: any confirmed Hormuz statement; Michigan preliminary 8/14. Invalidation for the residual: WTI closing below $78.18. Sizing: small, reduced from medium.

5. Long gold against the Swiss franc — keep it, halve it

Expression: long spot gold funded in CHF, notional-matched. Thesis, and what changed: the core observation still holds and got stronger — the franc weakened for a second consecutive session while gold made a fifth consecutive gain, and Tuesday removed the alternative explanations, because there was no oil shock, no VIX spike and no equity decline worth the name (§10, §11). The market is buying the non-yielding real asset and selling the low-yielding currency, which is an inflation-hedge configuration. What changed is the entry: Bespoke reports gold closed a full standard deviation above its 50-day for the first time in 103 sessions and entered overbought territory for the first time since 10 March, with the metal up roughly 8% in August. A right thesis at a stretched entry is a sizing problem, not a thesis problem. Catalyst: CPI 8/12, the direct test of the inflation-hedge leg. Invalidation: the franc strengthening more than 1% against the dollar in a session where gold is flat or lower; or gold breaking below $4,300 spot. Sizing: small, halved from medium.

6. Protection on the CCC cohort funded in IG — cut again

Expression: long CCC-exposed credit protection (or short a levered-loan/CCC-heavy vehicle) against long IG cash. Thesis and the crack, updated: the CCC credit spread at 1,014 bp is still 129 bp wider year to date while the IG credit spread is 1 bp tighter and the HY credit spread is at its 2026 tight; the CCC-minus-HY differential at 744 bp against 604 bp in January is still the only series that has tracked the funding cycle all year. But it has narrowed in three of the last five sessions and is 5 bp inside its 31 July level of 749 bp, and the stated cut trigger of 730 bp is 14 bp away. Correction to the prior edition: it published 763 bp for 31 July; the correct figure from the FRED series is 749 bp (CCC 1,034 bp less HY 285 bp), so the month-to-date narrowing is 5 bp rather than 19 bp. The idea is cut again, to a quarter of its original size, and it will be closed on a 730 bp print rather than held to it. Catalyst: CPI 8/12; the September quarter-end funding test, still binary with ON RRP at $1.250bn. Invalidation: the CCC-minus-HY differential through 730 bp. Sizing: small, quartered from the original.

7. Own the electrical-equipment second derivative against the semiconductor first derivative

Expression: long a basket of Eaton, Emerson, Hubbell, Rockwell Automation and Johnson Controls against short an equal-dollar SOX exposure. Thesis: the five names rose 2.69% to 3.57% as a bloc on the day Nvidia announced a $500bn third-party financing coalition, while the SOX rose 0.87% and Nvidia itself closed −0.03% (§4). The financing announcement changes who pays for the build-out, not who sells the chips — and the constrained input in a data-centre build is currently switchgear, transformers and cooling, not silicon. The credit evidence supports it: Nvidia’s own five-year CDS is 73 bp against roughly 41 bp for most of the year, so the chip vendor is carrying counterparty risk the electrical suppliers are not (§9). Catalyst: Applied Materials 8/13 AMC as the semiconductor leg’s test; any incremental hyperscaler capex disclosure. Invalidation: the SOX outperforming the basket by more than 4% over five sessions, which would say the market has decided the financing accrues to the chipmaker after all; or Nvidia’s CDS re-widening through 82 bp, which would take both legs down together and break the pair. Sizing: small-to-medium; note the basket’s beta is below the SOX’s, so this needs beta-adjusting rather than dollar-matching.

Volatility note

VIX 15.28 (−1.16%), range 15.23–15.61 — a decline on a down day, twelve hours before the print; September VIX futures 16.78, a 1.50-point contango; MOVE 75.46 on a stale 10 August vintage that did not publish an update, so the 4.94× ratio is not tradable (§9). At 15.28, one-month S&P implied prices a daily move of roughly 0.96%. Realised over the last three sessions has been −0.06%, −0.32% and, before that, +0.62% — implied remains comfortably above realised and the short-vol carry has kept working, which is precisely why nobody is positioned for the event. The specific asymmetry to price is unchanged and has got cleaner: §8 shows one basis point of the September contract is worth about eight points of headline probability and both vendors sit within 3 points of a coin flip; §7 explains why a soft CPI should produce a smaller rally than a hot CPI produces a selloff, because July’s data predates the crude move entirely. A short-dated put spread struck below Tuesday’s low of 7,717.25 is a cleaner expression of that skew than outright VIX calls at 15.28, and it does not require the vol level to rise to pay.

The crowded consensuses, with the numbers that would break them

1. Zero probability of a 2026 rate cut, for a seventeenth consecutive session. Investing.com’s cards print 0.0% in every sub-3.50% bucket at every 2026 meeting, and CME’s four-column table has no cut bucket at all. A payroll contraction, a 5% crude spike and a crude fade have all left it untouched. This is the single most crowded position in this report, and the thing that would break it is not CPI — it is a claims print above 240,000 on 8/13 alongside a soft PPI (§7). The asymmetry is severe because the price of being wrong is asymmetric: the strip has 6 bp of room to a certain hold in September and no priced room at all below the current range for the rest of the year.

2. Tight credit spreads against a rising long end and a low VIX. IG 78 bp for a sixth consecutive print, HY 270 bp at the 2026 tight, VIX 15.28 falling on a down day, the 10-year +7 bp on the week and the 20-year +7 bp and unmoved on the session. The index will not warn you. The three dispersion readings that will are named in §9: Nvidia’s five-year CDS at 73 bp against roughly 41 bp for most of the year, the CCC-minus-HY differential at 744 bp against 749 bp on 31 July, and the SOFR 99th-minus-1st span widening to 13 bp with a falling floor.

3. Megacap concentration masquerading as index risk. On Tuesday 270 of 495 screened S&P 500 constituents rose, the median constituent gained 0.19%, and the index fell 0.32%. Alphabet alone, at −3.84%/−3.60%, took Communication Services down 2.06% while the other ten Finviz groups had a median move of −0.07%. A book that is long the index and short single names has been carrying an unintended short-breadth position all week, and the two sessions have run in opposite directions: Monday’s decline came with 221 up and 267 down, Tuesday’s with 270 up and 221 down.

4. The AI-infrastructure financing chain, now with a public price. Nvidia’s five-year CDS nearly doubled to 82 bp in under three weeks and retraced to 73 bp on a financing announcement; Intel raised $20bn of equity in two tranches in two days; Riot’s twenty-year, 191-megawatt lease was marked down 5.46% and then up roughly 20% in 24 hours; CoreWeave and Super Micro both beat after Tuesday’s close. This is a chain in which the vendor, the customer, the landlord and the financier are increasingly the same set of balance sheets, and the only publicly observable price on the counterparty risk is a single-name CDS that moved 41 bp in three weeks. The $500bn coalition is a genuine mitigant and it is also an admission that the risk needed mitigating.

5. Everything priced for a benign July CPI that structurally cannot contain the crude move. Consensus is +0.1% headline and +0.2% core; Bloomberg Economics projects energy subtracting 11 bp from the headline and core y/y at its lowest since March 2021. WTI settled $78.18 on 7 August and $83.20 on 11 August. July’s print predates all of it. A soft number will be read as vindication and it will be measuring a month that no longer exists.