U.S. Stock, Fixed Income & Cross-Asset Closing Daily Friday, August 21, 2026 — Full Market Close Report | Data as of: ~4:00 p.m. ET close (Fed-probability cards timestamped 21 Aug 2026 05:05:08 CT / 05:55 p.m. EDT) Prepared for Institutional Investors. Not Personalized Investment Advice; Verify Independently before Acting. Full Data Notes & Conflicts and Source Links are in the companion file US_CrossAsset_Daily_2026-08-21_DataNotes.txt. |
| The tape in one paragraph. Every major U.S. gauge closed higher on Friday except the one that had been the only winner on Thursday — and it happened on a session when the bond market priced the Federal Reserve more hawkishly, not less. The official par curve cheapened almost in parallel: the 2-year rose 5 bp to 4.24%, breaking a three-session freeze at 4.19%; the 10-year 5 bp to 4.74%; the 20-year 5 bp to 5.25%; the 30-year 4 bp to 5.27%, while the 1-month bill did not move at all. CME FedWatch took the September hike from 36.1% to 40.1% on that. Equities rallied through it anyway. The Dow rose 517.80 points, or 0.98%, to 53,277.01, its best session of the week, on healthcare and financials — Merck +2.37%, Johnson & Johnson +1.07%, Amgen +1.88%, Goldman Sachs +3.74% to $1,039.41. The S&P 500 rose 0.43% to 7,674.37, the Nasdaq Composite 0.43% to 26,180.45, the Russell 2000 0.79% to 3,016.21, and VIX fell 5.50% to 15.13. The exception is the tell: the PHLX Semiconductor Index fell 0.51% to 11,740.4, the only major U.S. gauge down on the day, having opened at 11,901.8 and traded as high as 11,944.0 — a 1.71% high-to-close fade — with Marvell −5.57% to $237.04 on the same morning BMO Capital initiated it at Outperform. That is the exact inverse of Thursday, when SOX was the only green index on a red board. On macro, no release rated "Very high" landed in the past twelve hours — Friday's calendar carried only the New York Fed Staff Nowcast at 12:45 p.m. ET, rated Low — and no "Very high" release is due in the next twenty-four hours, because the next twenty-four hours are a Saturday (§7). The second-order tells are dense. Utilities fell 1.97%, the worst of eleven Finviz groups on an up day, which is the one equity sector that read the bond tape correctly. Copper rose 1.72% to $6.5800 after two consecutive sessions of refusing to join the precious bid, while gold added 1.97% to $4,661.60, a three-month high, on a dollar index that moved −0.04%. Bitcoin rose 8.17% to $78,616, capping a 22% week, its best in two years, and Robinhood rose 13.70% to $108.13. And Walmart could not bounce: down another 0.13% to $103.70 the day after its worst session since 2022, even with JPMorgan publicly buying it. For the week, every index finished lower — S&P 500 −1.43%, Nasdaq Composite −2.05%, Nasdaq 100 −2.45%, Dow −0.85%, Russell 2000 −1.75%, SOX −5.45% — and both the S&P and the Nasdaq snapped three-week winning streaks. |
| | Index | Close | Chg | %Chg | Note |
|---|
| S&P 500 | 7,674.37 | +33.21 | +0.43% | Investing.com's board carries 7,674.30, +33.14. Range 7,660.06–7,697.11. 1.60% below the 13 August closing record of 7,798.99 (CNBC); week −1.43% | | Nasdaq Composite | 26,180.45 | +113.28 | +0.43% | Range 26,049.30–26,269.85. Week −2.05%; three-week winning streak snapped | | Dow Jones Industrials | 53,277.01 | +517.80 | +0.98% | Range 52,768.87–53,355.92 — closed 0.15% below the high, the tightest close-to-high of the seven gauges. Second consecutive weekly loss | | Nasdaq 100 | 29,308.86 | +95.69 | +0.33% | Prior close 29,213.17. Range 29,142.44–29,405.12 on 342.4m shares. Snapped a five-day losing run (Bloomberg); week −2.45% | | Russell 2000 | 3,016.21 | +23.78 | +0.79% | Range 2,999.36–3,020.08. Back above 3,000; outperformed the S&P by 36 bp on a day the 2-year cheapened | | VIX | 15.13 | −0.88 | −5.50% | Range 15.08–15.88. Gave back 79% of Thursday's 7.52% rise; still +6.18% on the week | | PHLX Semiconductor (SOX) | 11,740.4 | −59.6 | −0.51% | Prior close 11,800.0. Open 11,901.8, range 11,631.7–11,944.0. The only major U.S. gauge down, and a 1.71% high-to-close fade; week −5.45% | | UST 2Y (official par) | 4.24% | +5 bp | — | The three-session freeze at 4.19% ended. The joint-largest move on the curve | | UST 10Y (official par) | 4.74% | +5 bp | — | Bloomberg's 4:59 p.m. mark 4.73%, +3 bp | | UST 20Y (official par) | 5.25% | +5 bp | — | Unchanged on the week — the only tenor on the curve that is | | UST 30Y (official par) | 5.27% | +4 bp | — | Bloomberg's evening mark 5.273%, +3.9 bp; CNBC quoted 5.276% at 5:05 p.m. | | WTI (Oct, NYMEX front) | $86.64 | −$0.19 | −0.22% | Range $85.81–$87.50 on 209,120 lots; second consecutive weekly gain, +5.15% on the week | | Brent (Oct, ICE front) | $93.93 | +$0.15 | +0.16% | Range $92.76–$94.82 on 286,610 lots; +6.04% on the week | | Gold (Comex Dec) | $4,661.60 | +$90.20 | +1.97% | Range $4,565.51–$4,690.11. Highest since 15 May; fifth straight weekly gain, the longest run since October 2025 | | Silver (Comex front) | $69.010 | +$0.905 | +1.33% | Range $67.980–$70.075. Traded above $70 intraday for the first time in this window | | DXY | 98.83 | −0.04 | −0.04% | Investing.com's dollar-index board, 15:59:31. Range 98.56–98.91. TradingEconomics prints 98.817, −0.08% |
|
| 2 · Market Hot Spots (ranked by tradability) |
| | 1. The front end broke, and nobody in the equity market noticed. For three consecutive sessions the 2-year sat at 4.19% while the long end did all the moving. On Friday it went to 4.24%, and the 3-year with it (4.26% → 4.31%). That is the joint-largest move on the curve, and it is a different animal from the last two sessions: bills did not move (1-month unchanged at 3.80%, 3-month +1 bp to 3.88%), so this is not financing. It is the market repricing policy. The Fed strip agrees to the decimal: CME's September hike probability went 36.1% → 40.1%, and Investing.com's 35.3% → 39.0% (§8). 2s10s finished unchanged at 50 bp because the whole coupon curve moved together, which is the cleanest possible signature of a parallel policy repricing rather than a term-premium event. And equities rose into it — S&P +0.43%, Russell +0.79%, VIX −5.50%. Forward catalyst: Jackson Hole, 27–29 August, with Chair Kevin Warsh's first symposium address on Friday 28 August. | | 2. Read the week, not the day: the buyback worked at 20 years and nowhere else. Over the five sessions the 2-year cheapened 7 bp, the 3-year 7, the 5-year 7, the 7-year 6, the 10-year 6, the 30-year 2 — and the 20-year is exactly unchanged at 5.25%. Treasury named the 10-to-20 and 20-to-30-year buckets, and the 20-year is the single tenor on the curve that gave nothing back across a week in which everything else did. 2s30s flattened 5 bp on the week to 103 and 20s30s steepened 2 bp to +2. The programme has not bought a bond yet — the first operation is 9 September — and it has already re-shaped the week's relative value. Forward catalyst: the operation schedule and the first execution size. | | 3. Semiconductors inverted Thursday exactly, and the fade is the evidence. SOX opened at 11,901.8, printed 11,944.0, and closed at 11,740.4 — down 0.51% on a day the Nasdaq 100 rose 0.33%, an 84 bp underperformance that precisely mirrors Thursday's +125 bp the other way. The composition is the story: Marvell −5.57% to $237.04 on 24.25m shares, Arm −2.91%, Intel −2.25% to $90.06 on 84.17m shares, Teradyne −1.93% after Baird cut it to Neutral, against Broadcom +1.21% to $368.45, GlobalFoundries +1.54% and Lam Research +1.15%. Nvidia fell 0.96% to $214.77 on 82.58m shares — the second consecutive session it underperformed its own index, three sessions before it reports. Forward catalyst: Nvidia, 26 August after the close (§5). | | 4. Bitcoin had its best week in two years and the equity proxies had a better one. The coin closed the U.S. session at $78,616, +8.17%, having ranged $73,025–$79,306, after starting the week at $62,836.88 — a 22% weekly advance. The equity complex ran harder: Strategy +29% on the week and its best since 17 April, Coinbase +25% and its best since May 2025, Circle +22%, Mara +21%, Robinhood +13%. Friday alone: Robinhood +13.70% to $108.13, Coinbase +8.20% to $186.49, Strategy +6.10% to $119.25, Circle +5.16% to $87.98. The driver is legislative — the White House push for the Clarity Act — not monetary. Forward catalyst: the bill's congressional path. The financials group rose 1.06% substantially on this. | | 5. Utilities were the worst group on an up day, and they were right. Utilities −1.97% against a rising index, −3.50% on the week, −7.44% on the month and −0.68% year to date — the only group other than communication services and consumer cyclical in the red for 2026. This is the highest-duration equity sector meeting a 20-year at 5.25% and a 30-year at 5.27%, and it is the one part of the equity market that traded the bond tape as written. When the rate-sensitive sector sells off 2% on a day the index rises 0.4%, the index is not disagreeing with the bond market; it is ignoring it. Forward catalyst: the 26 August PCE deflator and Warsh at Jackson Hole. | | 6. Copper finally joined, which changes what the metals bid means. For two sessions this report flagged copper as the non-participant while gold and silver ran. On Friday copper rose 1.72% to $6.5800, gold 1.97% to $4,661.60 — its highest since 15 May — silver 1.33% to $69.010 after trading above $70, and platinum 2.84% to $1,891.40. Basic materials was the best of eleven groups at +2.99%, and it did this with the dollar index essentially flat at −0.04%. A precious bid that no longer needs a falling dollar and now drags the industrial metal with it is either a reflation trade or a debasement trade that has broadened. Bloomberg's framing on Friday evening — "Bessent's Bond Maneuvers Giving Global Debasement Trade New Life" — argues for the second. Forward catalyst: the PCE deflator; the first buyback operation. | | 7. Healthcare had its best week since June, and it took two sessions of violence to get there. The sector rose 1.25% Friday and 4.29% on the week, the best of the eleven groups over five sessions, after Merck and Moderna's melanoma mRNA readout on Wednesday. Moderna closed +8.86% at $145.13 on 86.54m shares, having ranged $132.42–$159.47, one session after falling 23.55% and two after rising 176.97%. Merck rose 2.37% to $152.52 — back above the $152.20 close that TD Cowen's $137 target sat 10% beneath on Thursday. A stock that has printed +177%, −24% and +9% on consecutive sessions is not being valued; it is being traded. Forward catalyst: the data presentation and filing timetable, still undated. | | 8. Alibaba is the clearest AI-capex verdict of the week, and it is not an American stock. The ADR fell 8.57% to $119.34 on 30.27m shares after fiscal-Q1 2027 results showed capital expenditure of RMB67.7bn, or 25% of revenue, against cloud revenue growth of 45% — new cloud revenue equal to roughly 22% of capex, a 4.5-to-1 spend-to-incremental-revenue ratio — with adjusted earnings per ADS −42% to $1.26 against $1.85 expected on revenue of $39.64bn, +9%. Roughly $21bn of market value went with it. Note what happened here: the stock closed up 1.26% on Thursday on the same numbers, then fell 8.6% on Friday when the market did the capex arithmetic. Not an S&P 500 member. Forward catalyst: Nvidia on 26 August, where the identical question is asked of the seller rather than the buyer. | | 9. Citadel disclosed the size of the Situational Awareness unwind, and it is a market-structure item. Ken Griffin told clients Friday that Citadel has unwound more than 80% of the aggregate risk from the portfolio bought from Leopold Aschenbrenner's fund, via more than 100 block trades totalling over $4bn of market value, with discussions having opened on 29 July. A $4bn-plus block programme executed inside three weeks in AI-adjacent names is a non-fundamental supply that has been sitting underneath the technology tape for the whole period this report has been tracking the semiconductor complex — and it is now, by the firm's own account, four-fifths done. Forward catalyst: the residual 20%, and whether August's AI-complex weakness was partly mechanical. | | 10. Europe rose more than America and its bond market still would not follow. Euro Stoxx 50 +0.68% to 6,465.65, DAX +0.59% to 26,136.56, FTSE 100 +0.64% to 10,816.56, CAC 40 +0.37% to 8,484.43, IBEX 35 +0.76%, SMI +0.62%, with FTSE MIB flat at 52,668.04. On Bloomberg's board the U.S. 10-year rose 3 bp to 4.73% while Germany was unchanged at 3.26%, the U.K. fell 1 bp to 5.06%, and France, Italy and Spain each rose 1 bp. That is the third consecutive session of an American move of 3–5 bp against a flat bund. Three days is no longer noise; it is the market pricing an American duration and policy problem that is not being exported. |
| 3 · Sector Performance — August 21, 2026 |
| | Sector | 1-Day | 1-Week | YTD |
|---|
| Basic Materials | +2.99% | +6.60% | +22.90% | | Healthcare | +1.25% | +4.29% | +12.38% | | Financial | +1.06% | −1.66% | +7.38% | | Communication Services | +0.92% | −1.24% | −1.74% | | Consumer Cyclical | +0.80% | +0.14% | −2.13% | | Industrials | +0.60% | −3.67% | +12.95% | | Consumer Defensive | +0.59% | −1.41% | +7.31% | | Technology | +0.16% | −3.20% | +22.33% | | Real Estate | −0.05% | −0.35% | +10.99% | | Energy | −0.07% | +2.48% | +38.83% | | Utilities | −1.97% | −3.50% | −0.68% |
| | 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. | | YTD reconciliation. Compounding each group's 20 August YTD by Friday's one-day move reproduces the published YTD to within 0.024 percentage points across all eleven groups — the tightest reconciliation in this reporting window. Worked examples: basic materials 1.1934 × 1.0299 = 1.2291 → +22.91% against a published +22.90%, deviation 0.008 pp; utilities 1.0132 × 0.9803 = 0.9932 → −0.68%, published −0.68%, deviation 0.004 pp; healthcare 1.1098 × 1.0125 = 1.1237 → +12.37%, published +12.38%, deviation 0.013 pp. The largest deviation is communication services at 0.024 pp (computed −1.76% against a published −1.74%). Thursday's two persistent drifters, real estate and consumer cyclical, both reconciled to within 0.006 pp this session, which suggests the earlier gaps were membership rebalancing that has now washed through. | | Nine of eleven groups rose and the two that did not are the two that trade off the discount rate. Utilities at −1.97% is a 4.96-point spread to basic materials in a single session; real estate at −0.05% did not participate either. Everything with a duration in its multiple sat this rally out. The leadership was commodity beta and defensives: basic materials on copper +1.72%, gold +1.97%, platinum +2.84%, healthcare on the Moderna–Merck reversal, financials on the crypto complex and Goldman Sachs +3.74%. Note two composition points that the Finviz-versus-GICS mapping hides. Technology at +0.16% is the second-weakest riser on a day the Nasdaq 100 rose 0.33% — because Marvell, Intel, Arm and Teradyne sit in that bucket alongside Broadcom, and the memory-and-equipment complex was the drag. And energy at −0.07% closed red on a week it gained 2.48%, tracking a crude price that fell 0.22% rather than the Brent that rose 0.16% (§11). The weekly table is the sharper document: industrials −3.67% and technology −3.20% were the week's damage, and basic materials +6.60% and healthcare +4.29% were the week's rotation. |
| 4 · Movers & Single-Name Catalysts |
| | Higher | | • | Robinhood (HOOD) +13.70% to $108.13 on 47.89m shares — range $98.77–$109.71, closing 1.4% off the high after Thursday's 6.4% fade from its high. Best week since 2 July, +13% over five sessions, on the Clarity Act push. |
| | • | Moderna (MRNA) +8.86% to $145.13 on 86.54m shares — range $132.42–$159.47, a 16.3% intraday span, and a close 9.0% below the high. On track for its best week on record. |
| | • | Coinbase (COIN) +8.20% to $186.49, Strategy (MSTR) +6.10% to $119.25, Circle (CRCL) +5.16% to $87.98, Mara (MARA) +1.17% to $11.28 — the crypto equity complex on bitcoin's 22% week. |
| | • | IonQ (IONQ) +7.99% to $44.85 — range $42.01–$45.46. Not an S&P 500 member. |
| | • | SoFi (SOFI) +5.52% to $18.91 on 61.48m shares. Not an S&P 500 member. |
| | • | Tesla (TSLA) +5.14% to $362.88 on 57.84m shares — best week since May. The Information reported preparation for an August Cybercab launch, a robotaxi without a steering wheel; the Nevada Transportation Authority approved an autonomous network application in Clark County; rides are already running in Miami, Orlando, Tampa, Austin, Dallas and Houston. |
| | • | Goldman Sachs (GS) +3.74% to $1,039.41 — the largest Dow gainer, and a $37.28 per-share gain in a price-weighted index, which makes it the single biggest contributor to a 517.80-point day. |
| | • | Palantir (PLTR) +3.43% to $179.94 and Oracle (ORCL) +3.10% to $146.47 on 16.69m shares — Oracle recovered the whole of Thursday's 1.21% decline and more. |
| | • | Merck (MRK) +2.37% to $152.52 on 11.43m shares, Amgen (AMGN) +1.88% to $439.33, Johnson & Johnson (JNJ) +1.07% to $270.24, UnitedHealth (UNH) +1.35% to $390.05 — the healthcare block that carried the Dow. |
| | • | SpaceX (SPCX) +2.22% to $136.97 on 74.83m shares — Bloomberg reported SpaceX and AST are seeking grain spectrum valued at $6bn. |
| | • | Salesforce (CRM) +1.82% to $209.16, Caterpillar (CAT) +1.53% to $827.90, American Express (AXP) +1.45%, Visa (V) +1.45% to $371.04, Nike (NKE) +1.37%, Cisco (CSCO) +1.32%, Alphabet A (GOOGL) +1.24% to $344.90, Broadcom (AVGO) +1.21% to $368.45, Procter & Gamble (PG) +1.20%. |
| | • | Meta (META) +0.74% to $549.89, CrowdStrike (CRWD) +0.85% to $191.95, AMD +0.80% to $473.21, Taiwan Semiconductor ADR (TSM) +0.71% to $418.95, Microsoft (MSFT) +0.47% to $483.39. |
| | • | Dillard's and Chewy each rose more than 8% on the week, capping losses for the retail ETF; the SPDR S&P Retail ETF (XRT) closed +1.13% at $87.71 but −1% on the week. |
| | Lower | | • | Alibaba ADR (BABA) −8.57% to $119.34 on 30.27m shares — range $119.23–$125.77, closing at the low. AI capex RMB67.7bn, 25% of revenue; adjusted EPS per ADS −42% to $1.26 vs $1.85; revenue $39.64bn, +9%; cloud +45%. Roughly $21bn of market value. Not an S&P 500 member. |
| | • | Marvell (MRVL) −5.57% to $237.04 on 24.25m shares — range $233.28–$252.52, a 6.1% high-to-close fade, on the morning BMO Capital initiated coverage at Outperform with a $250 target. Gave back a third of its 16.2% two-day gain. |
| | • | Arm (ARM) −2.91% to $243.42 and Intel (INTC) −2.25% to $90.06 on 84.17m shares — the heaviest-traded semiconductor line on the board. |
| | • | Teradyne (TER) −1.93% to $375.74 — Baird downgraded to Neutral from Outperform, $420 target. |
| | • | IREN −1.73% to $41.865 on 41.23m shares — range $40.860–$45.930, a 8.9% high-to-close fade on a day every other crypto-adjacent name rose. Not an S&P 500 member. |
| | • | Honeywell (HON) −1.11% to $215.90 — the worst Dow component. |
| | • | Nvidia (NVDA) −0.96% to $214.77 on 82.58m shares — range $214.50–$218.74, closing 0.13% above the session low. Underperformed SOX by 45 bp and the Nasdaq 100 by 129 bp, three sessions before reporting. |
| | • | Micron (MU) −0.78% to $966.72, Netflix (NFLX) −0.69% to $79.59, Apple (AAPL) −0.60% to $309.42 — Bloomberg reported Apple cut jobs in the Siri, Vision Pro immersive video and gaming teams. |
| | • | Amazon (AMZN) −0.57% to $258.64, Boeing (BA) −0.42% to $214.20, Travelers (TRV) −0.25%, Chevron (CVX) −0.24% to $205.27. |
| | • | Walmart (WMT) −0.13% to $103.70 on 44.1m shares — could not bounce the day after a 9.15% decline, and closed the week down roughly 11% (CNBC), its worst since 2022. Note the vendor conflict flagged in Data Notes: Investing.com's board prints the same $103.70 close but a +0.10% change, which does not reconcile to Thursday's $103.84; StockAnalysis.com's −0.13% does, and is used here. |
| | • | Advance Auto Parts was the other end of the retail week, on track for its worst week since 2023. |
| | Analyst actions | | • | BMO Capital initiated the AI semiconductor complex, three names in one morning. Nvidia (NVDA), Outperform, $340 target — +58.3% above Friday's $214.77 close. Broadcom (AVGO), Overweight, $455 — +23.5% above $368.45. Marvell (MRVL), Outperform, $250 — +5.5% above $237.04. The market's response was a 0.96% decline in Nvidia and a 5.57% decline in Marvell, with only Broadcom up. An initiation with a 58% target on the largest company in the complex moving the stock down is a positioning statement, not a research statement. |
| | • | Baird downgraded Teradyne (TER) to Neutral from Outperform, $420 target — +11.8% above the $375.74 close. Baird also initiated Hubbell (HUBB) at Outperform, $555. |
| | • | JPMorgan is buying Walmart (WMT), Overweight, target cut to $125 from $137 — +20.5% above Friday's close and "nearly 21% upside from Thursday's close" on the firm's own framing. Christopher Horvers: "We are buyers – we view wash out as done as bears got their price and trends should improve while alternate profit pools accelerated." The stock fell again anyway. |
| | • | Baird upgraded Parsons (PSN) to Outperform from Neutral, target $57 from $48, roughly 24% upside; the stock rose about 4%. |
| | • | KeyBanc upgraded InnovAge (INNV) to Overweight from Sector Weight after meeting management; Matthew Gillmor sees "significant opportunities to optimize operations and accelerate growth." The stock rose about 2%. |
| | • | Rothschild & Co Redburn upgraded Equifax (EFX) to Buy from Neutral, target $235 from $214. Wells Fargo upgraded Portland General Electric (POR) to Overweight from Equal Weight, $58 from $51 — a utility upgrade on the worst day of the year for the utility group. Morgan Stanley raised Ternium (TX) to Overweight, $65 from $55. Citigroup upgraded O-I Glass (OI) to Buy, $9, and cut Weibo (WB) to Neutral, $7.80. JPMorgan cut Bowman Consulting (BWMN) to Underweight. Danske cut Frontline (FRO) to Sell, $39.15. |
| | • | Tom Lee (Fundstrat) called for 7,900–8,000 on the S&P by month-end, roughly 3–4% above Friday's close, citing no FOMC meeting in August, a strong second-quarter earnings season, improving 2027 forecasts and a near-complete AI de-leveraging. The index would need to clear its 13 August closing record of 7,798.99 to get there. |
|
| 5 · S&P 500 Earnings Calendar — Current & Next Week (S&P 500 components only) |
| | Times are ET where a clock time was verified. 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 / after-close bucket rather than a clock time, so no clock times are asserted this session; confirm every time against company IR before trading a date. | | | The current week is finished. There is no S&P 500 reporter left on the calendar; the next constituent to report is Intuit (INTU) on Tuesday 25 August, after the close (below). | | | Mon 8/24. No S&P 500 reporter on either bucket. | | Tue 8/25. AMC: Intuit (INTU). | | Wed 8/26. BMO: J.M. Smucker (SJM), Williams-Sonoma (WSM). AMC: Nvidia (NVDA), Salesforce (CRM), CrowdStrike (CRWD), Synopsys (SNPS), Agilent Technologies (A), Veeva Systems (VEEV), HP Inc. (HPQ). | | Thu 8/27. BMO: Dollar General (DG), Dollar Tree (DLTR), Best Buy (BBY), Hormel Foods (HRL). AMC: Marvell Technology (MRVL), Autodesk (ADSK), Workday (WDAY), Ulta Beauty (ULTA). | | Fri 8/28. No S&P 500 reporter on either bucket. | | Changes vs. the prior calendar (8/20 report): | | • | No additions and no removals. Every S&P 500 name on the prior roster for 8/24 through 8/28 reappears on this session's independent Nasdaq capture with the same before-open / after-close bucket, including Williams-Sonoma before the open on 26 August, which was the single change flagged in the prior edition and has now been confirmed by a second consecutive capture. |
| | • | Non-members on the same dates, listed so nobody mistakes their absence for an omission: PDD Holdings (PDD) and XPeng (XPEV) on 8/24; Bank of Montreal (BMO), Bank of Nova Scotia (BNS), Semtech (SMTC), Box (BOX), nCino (NCNO), Boss Zhipin (BZ) and Vipshop (VIPS) on 8/25; Trip.com (TCOM), Li Auto (LI), Dycom (DY) and Donaldson (DCI) on 8/26; Royal Bank of Canada (RY), Toronto-Dominion (TD), CIBC (CM), Rubrik (RBRK), IREN, Harmony Gold (HMY) and Bilibili (BILI) on 8/27; Ubiquiti (UI) and Miniso (MNSO) on 8/28. |
| | • | Membership caveat, unchanged. The constituent screen used this session does not carry Heico (HEI/HEI.A), Zoom (ZM), Dick's Sporting Goods (DKS), PVH, Coty (COTY), Bath & Body Works (BBWI), Kohl's (KSS), Abercrombie & Fitch (ANF), Urban Outfitters (URBN), Nutanix (NTNX), Okta (OKTA), Five Below (FIVE), Burlington (BURL), Gap (GAP), Affirm (AFRM), Elastic (ESTC), SentinelOne (S) or HealthEquity (HQY), each of which reports in the covered window. All are conservatively excluded and listed in Data Notes. |
| | • | What the forward calendar hands the desk. Ten trading days of S&P 500 reporting have collapsed into eighteen names across three days, and nine of them land on Wednesday 26 August, seven of those after the close. The concentration is not merely calendrical: Nvidia, Salesforce, CrowdStrike, Synopsys, Agilent, Veeva and HP are, between them, the AI-compute seller, two software franchises whose multiples depend on AI not being deflationary to seat-based pricing, the design-tool monopolist and the hardware tail. Nvidia alone traded 82.58m shares on Friday while falling 0.96% (§4). Then Marvell reports on Thursday 27 August, the day the market will already be reading Nvidia's read-through — and Marvell has moved +9.85%, +5.82% and −5.57% on the three sessions since the Google equity right was reported. For the reaction function the week just produced, and the dispersion inside it, see §4 and §2. |
|
| 6 · U.S. Treasury Yields — Official Par Curve |
| | Source: U.S. Department of the Treasury daily par yield curve, 21 August 2026, read from the official Text View. All fourteen tenors were extracted; the ten published below are the report's standing tenor set. Changes are versus the 20 August row (1-day) and the 14 August row (1-week). | | Tenor | 21 Aug | 1-Day | 1-Week |
|---|
| 1 Mo | 3.80% | 0 bp | +1 bp | | 3 Mo | 3.88% | +1 bp | +2 bp | | 1 Yr | 4.03% | +4 bp | +5 bp | | 2 Yr | 4.24% | +5 bp | +7 bp | | 3 Yr | 4.31% | +5 bp | +7 bp | | 5 Yr | 4.43% | +4 bp | +7 bp | | 7 Yr | 4.57% | +4 bp | +6 bp | | 10 Yr | 4.74% | +5 bp | +6 bp | | 20 Yr | 5.25% | +5 bp | 0 bp | | 30 Yr | 5.27% | +4 bp | +2 bp |
| | Spread | 21 Aug | 1-Day | 1-Week |
|---|
| 2s10s | 50 bp | 0 bp | −1 bp | | 3M10Y | 86 bp | +4 bp | +4 bp | | 2s30s | 103 bp | −1 bp | −5 bp | | 20s30s | 2 bp | −1 bp | +2 bp |
| | The shape and its diagnostic: an almost perfectly parallel bear shift in coupons, with the bill curve anchored. Every coupon tenor from the 1-year out moved 4 or 5 basis points cheaper, and the two ends of the coupon curve moved by exactly the same amount — 2-year +5, 10-year +5 — leaving 2s10s unchanged at 50 bp. Meanwhile the 1-month did not move at all and the 3-month moved 1 bp, which is why 3M10Y steepened 4 bp to 86. A move of that shape is not a term-premium event and it is not a supply event; both of those pivot the curve. A parallel coupon shift against a fixed bill is the signature of the market repricing the expected policy path, and §8 confirms it independently: the September hike probability rose 4.0 points on CME and 3.7 points on Investing.com in the same session. | | The three-session freeze at the 2-year is the headline. Wednesday, Thursday and the session before all printed 4.19% while 5s through 30s swung by 4–11 bp on the Treasury buyback announcement and its reversal. Friday broke it by 5 bp. That matters because it was the front end's refusal to move that made the last two sessions readable as a duration story. Once the 2-year moves — and moves with the 1-year (+4 bp to 4.03%) and the 3-year (+5 bp to 4.31%) — the story is no longer about who buys the 30-year. It is about what the Fed does in September. | | The weekly table says something different from the daily one, and it is the more useful document. Across five sessions the curve bear-flattened: 2-year +7, 3-year +7, 5-year +7, 7-year +6, 10-year +6, 30-year +2 — and the 20-year exactly unchanged at 5.25%. 2s30s flattened 5 bp to 103 and 20s30s steepened 2 bp to +2. The buyback was announced on Wednesday for the 10-to-20 and 20-to-30-year buckets, and the tenor it named most directly is the only point on the curve that has given nothing back over the week. Treasury has not bought a single bond — the first operation is 9 September — and the announcement effect is already visible as a relative-value fact rather than a level fact. | | Off-table bills, where the financing story lives. The dropped tenors are not neutral this session. The 4-month rose 2 bp to 3.90%, the largest bill move on the curve and double the 3-month's; the 6-month rose 1 bp to 3.95%; the 2-month rose 1 bp to 3.80%; and the 1.5-month was unchanged at 3.77%. That leaves the front of the bill curve still kinked: the 1-month at 3.80% sits 3 bp above the 1.5-month at 3.77% and level with the 2-month, an inversion that has now persisted for three sessions. On the week the shape is sharper still — 1.5-month −3 bp, 2-month −1 bp, 1-month +1 bp, 3-month +2 bp, 4-month +2 bp — a cheapening beyond one month and a richening inside it. That is a September bill-supply signature, not a policy one, and it belongs with the funding data in §9 rather than with the policy repricing above. | | Vendor gap, explained rather than disputed. Bloomberg's 4:59 p.m. board marked the 10-year at 4.73%, +3 bp and the 30-year at 5.273%, +3.9 bp, against official par moves of +5 and +4. CNBC quoted 5.276% on the 30-year at 5:05 p.m. The official par curve is struck at approximately 3:30 p.m. ET from bid-side quotes and carries its own smoothing; real-time vendor marks run to the evening electronic session. The level agreement is within a basis point at both tenors; the change difference is a baseline artefact, because the two sources are measuring from different Thursday reference points. |
| 7 · U.S. Macroeconomic Calendar |
| | Source: Federal Reserve Bank of New York official Economic Indicators Calendar for August 2026. Times ET. Sensitivity is this report's rating, and it drives what §1 must name. | | Current week (Aug 17–21) — remaining | | Nothing remains. The last item on the week's calendar was the New York Fed Staff Nowcast at 12:45 p.m. ET on Friday 21 August, rated Low, and it has been released. The week carried no "Very high" item after Thursday's initial claims. | | Next week (Aug 24–28) | | Date | Time (ET) | Release | Sensitivity |
|---|
| Mon 8/24 | 11:00 | SCE Labor Market Survey (NY Fed) | Low | | Tue 8/25 | 08:30 | Philadelphia Fed Non-Manufacturing Survey | Low | | Tue 8/25 | 10:00 | Consumer Confidence (Conference Board) | Medium | | Tue 8/25 | 10:00 | New Residential Sales | Medium | | Tue 8/25 | 10:00 | Richmond Fed Survey of Manufacturing Activity | Low | | Wed 8/26 | 08:30 | Advance Durable Goods | Medium | | Wed 8/26 | 08:30 | Gross Domestic Product, 2nd release (Q2) | Medium | | Wed 8/26 | 08:30 | Personal Income and the PCE Deflator | Very high | | Wed 8/26 | 10:00 | Corporate Bond Market Distress Index (NY Fed) | Low — but see §9 | | Thu 8/27 | 08:30 | Initial Jobless Claims | High | | Thu 8/27 | 10:00 | Multivariate Core Trend Inflation (NY Fed) | Medium | | Thu 8/27 | 11:30 | Weekly Economic Index | Low | | Thu 8/27 | 14:00 | R-Star (Laubach–Williams estimates) | Low | | Thu 8/27 – Sat 8/29 | — | Jackson Hole Economic Policy Symposium (Kansas City Fed); Chair Kevin Warsh's keynote, Friday morning 28 August | Very high (event) | | Fri 8/28 | 10:00 | Michigan Consumer Survey (final) | Medium | | Fri 8/28 | 12:45 | New York Fed Staff Nowcast | Low |
| The look-ahead — a hollow four days, then everything at once. From Monday's open to Wednesday's 8:30 there is one Medium-rated release with the capacity to move a rate: Consumer Confidence on Tuesday. Then the calendar detonates. The PCE deflator lands Wednesday 26 August at 8:30, and it is the single most important input to a strip that now prices 40.1% for a September hike and 71.2% cumulative for at least one hike by December with zero probability of a cut at any 2026 meeting (§8). Twelve hours later Nvidia reports (§5). The following morning claims print, rated High because Thursday's 206,000 was the number that removed the labour-market route to easing and the four-week average is running near 204,000. And on Friday 28 August, Chair Warsh delivers his first Jackson Hole address, on a symposium theme of financial innovation and payments, into a bond market that has cheapened 6–7 bp across the belly in a week and a 20-year the Treasury has publicly promised to support. That address is not a data release and it is rated Very high anyway, because it is the only scheduled event in the window that can re-anchor the terminal rate — which moved 5.5 bp higher on Friday alone, from 4.020% to 4.075% on the September and October 2027 contracts. The asymmetry: a core PCE at or below +0.2% m/m meets a market that has spent the week taking easing out of 2027, so the dovish surprise has more room to travel than the hawkish one; but a Warsh framing that treats 3.50–3.75% as a floor rather than a destination puts September through 50% for the first time since July, and it arrives on a Friday with no session left to fade it. The concrete hooks, in the order they can move the Fed card: Consumer Confidence 8/25 10:00 → PCE deflator 8/26 08:30 → claims 8/27 08:30 → Warsh 8/28 morning → Michigan final 8/28 10:00. |
|
| 8 · Fed Funds Futures & Rate Path |
| | CME FedWatch — the four-column headline (September 16 meeting) | | Meeting information. Meeting date 16 September 2026; contract ZQU6; expires 30 September 2026; mid price 96.3225; prior volume 32,465; prior open interest 249,354. Current target range 3.50–3.75%. | | Target rate (bp) | NOW | 1 day (20 Aug 2026) | 1 week (14 Aug 2026) | 1 month (21 Jul 2026) |
|---|
| 350–375 (current) | 59.9% | 63.9% | 66.9% | 30.3% | | 375–400 | 40.1% | 36.1% | 33.1% | 54.5% | | 400–425 | 0.0% | 0.0% | 0.0% | 15.2% | | EASE / NO CHANGE / HIKE | 0.0% / 59.9% / 40.1% | 0.0% / 63.9% / 36.1% | 0.0% / 66.9% / 33.1% | 0.0% / 30.3% / 69.7% |
| | Provenance of every column, stated. All four columns were read live and numerically off CME FedWatch's "Current" view, whose footer stamps the data as of 21 Aug 2026 05:05:08 CT (≈6:05 p.m. ET). This is the first edition in this reporting window in which CME published a complete numeric four-column table rather than only plotting the comparison series, so no column here is chart-read, reproduced from a prior edition, or borrowed from another vendor. The 1 MONTH column carries a reference date of 21 July 2026, exactly one calendar month back, and is therefore usable in calculations. One standing caveat applies: a live CME read taken after 5:00 p.m. ET is indicative rather than a settlement snapshot, and if the settled figures differ this edition will be corrected in the next one. | | The CME-versus-Investing.com gap, reconciled and quantified. CME puts the September hike at 40.1%; Investing.com's Fed Rate Monitor, timestamped 21 Aug 2026 05:55 p.m. EDT, puts it at 39.0% — a gap of 1.1 percentage points. The mechanical cause is the quoted contract price: CME's mid is 96.3225 and Investing.com's is 96.325, a difference of 0.25 basis points. Because the 16 September decision is effective from 17 September, only 14 of September's 30 days carry the post-meeting rate in the ZQ average, so the contract weights the new rate at 14/30 = 46.7%. A 0.25 bp difference in price therefore scales to 0.25 × 30/14 = 0.54 bp of implied post-meeting rate, which is 0.54 / 25 = 2.1 percentage points of hike probability. The price difference alone over-explains the observed 1.1 pp gap; the residual is each vendor's assumed effective-rate base. That base can be checked: back-solving CME's own numbers from a price of 96.3225 gives an implied current effective rate of 3.631%, and the New York Fed's published EFFR for 20 August is 3.63% (§9). The two vendors are pricing the same distribution and rounding it differently. | | Multi-day momentum, and the month that has to be read backwards. The hike has been bid for a week: 33.1% → 36.1% → 40.1% on CME across 14, 20 and 21 August, and 32.3% → 35.3% → 39.0% on Investing.com's independent previous-week and previous-day columns. That is +7.0 points in a week and +4.0 points in a session, and Friday's move is the largest single-day repricing of the September meeting in this window. But the one-month column reverses the sign of the story: on 21 July the market priced 69.7% cumulative hike including 15.2% on a full 50 bp, against 40.1% and 0.0% today. So the correct arc is a large dovish repricing through late July and early August that has now retraced roughly a quarter of itself in five sessions — and it retraced on a week containing a 206,000 claims print, a Philadelphia Fed manufacturing index at 47.4 with prices paid at a six-month low, and a failed sovereign intervention in the long end. A hike being bid on disinflationary data is the anomaly to hold on to; the market is trading the supply of duration and the identity of the Chair, not the inflation path. | | (a) Current-year meeting distributions — current / [prev-day] / [prev-week] | | Source: Investing.com Fed Rate Monitor, updated 21 Aug 2026 05:55 p.m. EDT. Current target range 3.50–3.75%. | | Meeting | Contract | 3.50–3.75 (hold) | 3.75–4.00 (+25) | 4.00–4.25 (+50) | 4.25–4.50 (+75) | Cumulative hike |
|---|
| 16 Sep 2026 | 96.325 | 61.0% [64.7] [67.7] | 39.0% [35.3] [32.3] | 0.0% | 0.0% | 39.0% [35.3] [32.3] | | 28 Oct 2026 | 96.265 | 47.0% [51.3] [53.7] | 44.1% [41.4] [39.6] | 9.0% [7.3] [6.7] | 0.0% | 53.1% [48.7] [46.3] | | 9 Dec 2026 | 96.140 | 28.9% [33.4] [32.8] | 45.2% [44.9] [45.1] | 22.5% [19.2] [19.5] | 3.5% [2.6] [2.6] | 71.2% [66.7] [67.2] |
| | Cut probability is 0.0% at every 2026 meeting, unchanged for the whole of this reporting window. The October meeting crossed a threshold on Friday: cumulative hike 53.1% against a hold at 47.0% means the strip now expects, on balance, that the Fed has moved by the end of October — the first time in this window that a 2026 meeting other than December has had a hike as its majority outcome. Read the momentum across the three meetings: the September hike gained 3.7 points in a session, October 4.4 and December 4.5. The whole 2026 strip moved together and the back end moved most, which is the opposite of what a single-meeting reassessment looks like and consistent with a level shift in the expected path. | | (b) Next-year path — modal range, cumulative and the contracts that draw the terminal | | Meeting | Contract price | Modal range | Modal prob. | Cumulative above 3.50–3.75 | Cumulative below |
|---|
| 27 Jan 2027 | 96.110 | 3.75–4.00 | 41.7% | 77.3% | 0.0% | | 17 Mar 2027 | 96.030 | 3.75–4.00 | 36.6% | 83.4% | 0.0% | | 28 Apr 2027 | 95.995 | 3.75–4.00 | 33.9% | 85.5% | 0.0% | | 9 Jun 2027 | 95.940 | 4.00–4.25 | 32.2% | 87.7% | 0.0% | | 28 Jul 2027 | 95.930 | 4.00–4.25 | 32.2% | 87.9% | 0.0% | | 15 Sep 2027 | 95.925 | 4.00–4.25 | 32.2% | 87.9% | 0.0% | | 27 Oct 2027 | 95.925 | 4.00–4.25 | 31.8% | 87.2% | 0.3% | | 8 Dec 2027 | 95.945 | 3.75–4.00 | 30.9% | 84.7% | 1.3% |
| | The terminal moved 5.5 bp higher in one session. The cheapest contracts on the strip are September and October 2027 at 95.925, implying 4.075% — against 95.980 and 4.020% in the prior edition. The modal range flips from 3.75–4.00% to 4.00–4.25% at the June 2027 meeting and stays there until December, when it flips back. Cumulative pricing above the current range peaks at 87.9% in July and September 2027, and the strip prices a cut at some point in 2027 at only 1.3% by December — down from 1.7% a day earlier and 2.8% a week earlier. The market has spent the week removing more than half of the little easing it had priced for 2027 while simultaneously pushing the first hike right; the December-2027 contract richening to 95.945 from the 95.925 trough is the only shape on the strip that resembles an eventual cut, and it is worth 2 bp. | | (c) Year-end probability ladders | | Year-end 2026 — the 9 December meeting. Current target range 3.50–3.75%. | | Outcome | Range | Current | [prev-day] | [prev-week] |
|---|
| −25 bp | 3.25–3.50 | 0.0% | 0.0% | 0.0% | | Hold | 3.50–3.75 | 28.9% | [33.4] | [32.8] | | +25 bp | 3.75–4.00 | 45.2% | [44.9] | [45.1] | | +50 bp | 4.00–4.25 | 22.5% | [19.2] | [19.5] | | +75 bp | 4.25–4.50 | 3.5% | [2.6] | [2.6] | | +100 bp | 4.50–4.75 | 0.0% | 0.0% | 0.0% |
| | Year-end 2027 — the 8 December meeting. | | Outcome | Range | Current | [prev-day] | [prev-week] |
|---|
| −50 bp | 3.00–3.25 | 0.0% | 0.0% | [0.1] | | −25 bp | 3.25–3.50 | 1.3% | [1.7] | [2.7] | | Hold | 3.50–3.75 | 14.0% | [18.1] | [19.9] | | +25 bp | 3.75–4.00 | 30.9% | [34.3] | [34.9] | | +50 bp | 4.00–4.25 | 30.7% | [28.7] | [27.5] | | +75 bp | 4.25–4.50 | 16.7% | [13.1] | [11.6] | | +100 bp | 4.50–4.75 | 5.3% | [3.5] | [2.8] | | +125 bp | 4.75–5.00 | 1.0% | [0.5] | [0.4] | | +150 bp | 5.00–5.25 | 0.1% | 0.0% | 0.0% | | +175 bp and beyond | 5.25–5.50+ | 0.0% | 0.0% | 0.0% |
| | Cumulative for year-end 2027: above the current range 84.7% [80.1] [77.2]; below 1.3% [1.7] [2.8]. The +25 and +50 outcomes are now within 0.2 percentage points of each other (30.9% versus 30.7%), so the modal cell is not meaningfully modal, and the distribution's centre of gravity has shifted a full step up in five sessions. | | Rounding, stated transparently. The Investing.com rows sum to 100.0% for September 2026, January, July, September and December 2027; 100.1% for October and December 2026 and June 2027; and 99.9% for April 2027. Cumulative figures in this section are computed from the published cells without renormalising, so a cumulative may inherit up to 0.1 point of that rounding. |
| | (a) IG and HY credit spreads | | As-of date, stated rather than implied. FRED's ICE BofA series publish with a one-business-day lag; all three carried 20 August as their last observation when read this session, so Friday's session is not in these numbers and publishes on 24 August. Same-day direction was cross-checked against the cash-ETF tape: LQD closed −0.13% at $105.92 and HYG +0.06% at $79.61 on a day the 10-year cheapened 5 bp. HY cash outperformed IG cash by 19 bp and closed higher outright while duration sold off — a clean statement that Friday's move was a rates move, not a credit move. | | Series | Level | As-of | 1-day | 1-week | YTD (31 Dec 2025) |
|---|
| IG — ICE BofA US Corporate OAS (BAMLC0A0CM) | 82 bp | 20 Aug | +1 bp (81) | +2 bp (80 on 14 Aug) | +3 bp (79) | | HY — ICE BofA US High Yield OAS (BAMLH0A0HYM2) | 275 bp | 20 Aug | +2 bp (273) | +8 bp (267) | −6 bp (281) | | CCC & lower OAS (BAMLH0A3HYC) | 1,035 bp | 20 Aug | +5 bp (1,030) | +23 bp (1,012) | +150 bp (885) | | CCC minus HY differential | 760 bp | 20 Aug | +3 bp (757) | +15 bp (745) | +156 bp (604) | | CDX IG 5y | see retrieval note | — | — | — | — | | CDX HY 5y | see retrieval note | — | — | — | — |
| | Mark the prior edition's test honestly, because it landed between the goalposts. Thursday's report wrote: "a 21 August IG print back at 83 bp or wider would say Thursday's selloff hit credit as well as duration and would restore the trend this edition just declared broken; an IG print at or inside 79 bp would confirm the widening episode is over." The print came at 82 bp — one basis point short of the widening test and three wide of the tightening test. Neither condition triggered. What did happen is that the five-session widening run (78 → 79 → 80 → 81 → 82) that broke at 81 on 19 August has resumed at 82, matching the month's widest reading set on 18 August. The correct statement is therefore: the trend was not confirmed by the written test, and it was not falsified either; credit re-widened on Thursday's selloff by exactly the amount duration would predict and no more. The tail is where the information is. CCC widened another 5 bp to 1,035, a fresh monthly wide, +23 bp on the week against HY's +8 and IG's +2 — the tail widened nearly three times as fast as the index across five sessions, and the CCC-minus-HY differential is now 760 bp, +156 bp on the year while HY itself is still 6 bp tighter than it started 2026. | | CDX retrieval note — the six-step ladder, worked and named. (1) Bloomberg in Chrome: /markets and /markets/rates-bonds both rendered fully; the Fixed Income Indices panel published Global Aggregate 501.07, U.S. Aggregate 2,340.79, Asian-Pacific 194.01, Pan-Euro 226.14 and EM USD Aggregate 1,404.41, and a programmatic search of the rendered page text returned no CDX string. (2) WSJ Market Data: wsj.com/market-data/bonds rendered, but its instrument tables are script-gated against programmatic extraction and the article-level read returned only the day's story, "Treasury Yields Rise as Buybacks Could Top $4 Billion an Issue"; no CDX level appeared. (3) Cbonds / ICE / S&P Global: this session reached the Cbonds CDX.NA.IG 5Y page, which is a step further than the prior edition's Cloudflare block — it confirms the series is current, with a 20/08/2026 value and a previous value on 19/08/2026 — but every numeric cell renders masked behind the subscription wall, and the same is true of CDX.NA.HY 5Y and the whole iTraxx family. ICE's Markit CDX.NA.IG product page and the S&P Dow Jones CDX family page publish methodology and index-news documents, not levels. (4) FT Markets Data and Reuters credit wraps: no instrument page and no wire wrap quoting a level in the reviewed material. (5) TradingView / Barchart / CME CDS index products: documentation only. A general search again surfaced an undated third-party digest asserting "81 basis points" for IG; under this report's standing rule an undated, unattributed digest number is not a CDX level and it is withheld. (6) Cash-market proxy: LQD −0.13% and HYG +0.06%, published as a proxy for the direction of cash credit and not as a CDX level. Quoting conventions restated: CDX IG 5y is quoted in basis points of spread; CDX HY 5y in price points, where a rising price means tightening credit spreads. | | (b) Money-market and funding plumbing | | NY Fed reference rates publish at roughly 8:00 a.m. ET for the prior business day; the overnight series below carry a 20 August effective date. The SOFR averages, the index and the reverse-repo take-up carry 21 August. | | Measure | Level | Detail |
|---|
| SOFR | 3.63% | 20 Aug, up 1 bp from 3.62%. Volume $2,922bn; 1st pct 3.58%, 25th 3.60%, 75th 3.68%, 99th 3.71% | | EFFR | 3.63% | 20 Aug. Volume $102bn, up from $95bn; 1st pct 3.60%, 25th 3.62%, 75th 3.63%, 99th 3.69% | | OBFR | 3.63% | 20 Aug. Volume $229bn; 1st pct 3.53%, 99th 3.68% | | TGCR | 3.60% | 20 Aug, unchanged. Volume $1,192bn | | BGCR | 3.60% | 20 Aug, unchanged. Volume $1,220bn | | SOFR − IORB | −2 bp | IORB 3.65%. A second consecutive print through the administered rate, but 1 bp closer to it than Wednesday's −3 bp | | 30-day average SOFR | 3.64319% | 21 Aug effective date; 90-day 3.63920%, 180-day 3.66017%; SOFR index 1.25565902 | | ON RRP take-up | $0.200bn | 21 August, from $0.225bn on 20 August and $0.317bn on 19 August. Effectively empty against a $2.4tn peak | | Reserve balances (week to 19 Aug) | $2,935.3bn | No new weekly print — the next release is 27 August. −$8.8bn on the week and −$207.4bn from the 15 July peak of $3,142.7bn | | 4-month bill (par, 21 Aug) | 3.90% | +2 bp on the day — the largest bill move on the curve, double the 3-month's, and off-table under this report's tenor set (§6) |
| | The funding question closed, and it closed one basis point tighter than it opened. Wednesday's report set the test in writing: "a second consecutive SOFR print below IORB on 21 August closes the funding question outright." The 21 August publication carried the 20 August effective date at 3.63%, which is the second consecutive print below the 3.65% IORB after 19 August's 3.62%. The question is closed. But read the shape rather than the headline: the mean backed up 1 bp toward the administered rate and the 75th percentile went the wrong way, from 3.67% to 3.68%, while the 1st-to-99th band held at 13 bp — so the distribution's upper half re-tightened even as the average stayed through IORB. Tri-party and broad general collateral were both unchanged at 3.60% on roughly $1.2tn a day each, which says the secured core is genuinely comfortable; the pressure is in the tail, not the median. The unsecured tail eased fractionally — EFFR's 99th percentile at 3.69%, OBFR's at 3.68% against 3.69% — on $102bn of EFFR volume, up from $95bn. The bill curve remains the live item. The 4-month cheapened 2 bp to 3.90% and the 6-month 1 bp to 3.95% while the 1-month did not move and the 1.5-month did not move, leaving the front kinked (1-month 3.80% above 1.5-month 3.77%) for a third session. That is a September supply and month-end financing shape, and it sits against reserves $207.4bn below their July peak and a reverse repo facility holding $200m — no cushion if buyback settlement pushes bill issuance higher. | | (c) Rates volatility and swap spreads | | The MOVE series did not update for Friday. Investing.com's board still carries 73.18, +2.69%, dated 20/08, with a session range of 71.26–73.18. The last matched-date ratio is therefore MOVE 73.18 against VIX 16.01 on 20 August = 4.57×. Pairing Thursday's MOVE with Friday's VIX of 15.13 produces 4.84×, and that number should not be traded on because it compares two different dates; it is quoted only to show the direction of the distortion. What can be said cleanly is this: VIX fell 5.50% on Friday to 15.13 and rose 6.18% on the week, while MOVE's last three prints were 74.98 → 71.26 → 73.18. Equity volatility spiked and unwound inside 48 hours; rates volatility has done neither. A rates market that cheapened 6–7 bp across the belly in a week, broke a three-session freeze in the 2-year and added 7 points of September hike probability without its own volatility index moving more than 3% in a day is a market where the level is being repriced and the distribution is not — which is exactly the configuration that under-prices a Jackson Hole address. | | No verified 2y, 10y or 30y swap-spread level was obtainable this session. The vendors publishing them sit behind entitlement walls and no wire quoted a level in the reviewed material. That is a gap and is recorded as one for a third consecutive edition. | | (d) Issuance, leveraged loans and private credit | | The supply story is the credit story, and it now has a full-year number attached. August IG supply reached $145.2bn by 17 August, topping August 2020's $136bn full-month record and marking the third consecutive monthly record, against an August average since 2019 of roughly $95bn. Through July, 2026 IG issuance stands at $1,681bn, +26.9% year on year, and the Street's projection is that gross IG issuance tops $2tn in 2026 against $1.7tn in 2025. Bloomberg's framing on 17 August — "US High-Grade Bond Sales Set August Record in Year Full of Them" — attributes the pace directly to AI build-out borrowing. Set that against the Broadcom package reported Thursday: more than $60bn of senior debt plus roughly $30bn of junior financing through a special-purpose vehicle, potentially approaching $100bn, with Blackstone and Apollo in discussions; and against BofA's flag of $370bn of AI-related debt across the complex. Friday added the equity-side mirror: Alibaba disclosed RMB67.7bn of quarterly capex, 25% of revenue, producing new cloud revenue equal to roughly 22% of that spend — and lost 8.57% for it (§4). The market is now discriminating between AI capex that is funded and AI capex that is returning, and it is doing so in the equity market faster than in the credit market. No verified HY primary volume, Morningstar LSTA leveraged loan index level or bank CDS level was published in the reviewed material this session; those three fields are gaps and are recorded as gaps. Ken Griffin's disclosure that Citadel has unwound more than 80% of the Situational Awareness risk through over 100 block trades worth more than $4bn is the private-market liquidity item of the week and belongs here as much as in §2. | The credit take — credit is pricing duration, not risk, and the tail is the only honest signal. On the last published day IG widened 1 bp to 82, HY 2 bp to 275 and CCC 5 bp to 1,035. Line those up against the rates market: the 10-year cheapened 4 bp on the same date and 5 bp the next. IG's 1 bp of widening on a 4 bp duration move is not a credit event; it is a beta. HY's +2 is the same statement with more spread to work with. The tail is different in kind: CCC has widened 23 bp in a week against HY's 8, the differential is at 760 bp and +156 bp on the year, and it has widened in every one of the last four published sessions including the one where IG tightened. So the divergence to watch is no longer credit-versus-equity-vol — VIX at 15.13 and IG at 82 bp are, for now, telling the same benign story. It is core-versus-tail: an index whose spread is 82 bp absorbing a record $145.2bn August calendar without flinching, while the cohort that actually needs the financing widens 23 bp in five sessions. What breaks it, in order of likelihood: a 24 August IG print at 84 bp or wider would say Friday's 5 bp parallel coupon shift reached the core and would make the widening a trend rather than a beta; a CCC print inside 1,025 bp would say the tail move was a summer-liquidity artefact and end this thread; and the NY Fed's Corporate Bond Market Distress Index on 26 August is the one scheduled reading that can arbitrate between them. On the plumbing side, the funding question is closed — but a SOFR print back at 3.65% or above with the 75th percentile through 3.70% would reopen it as a bill-supply problem, and the 4-month at 3.90% is where that would show up first. |
|
| | Quote basis. Levels are the TradingEconomics currency board on its Aug/21 dated rows, pulled at approximately 6:35 p.m. ET, with the header order verified as Price | Chg | %Chg | Weekly | Monthly | YTD | YoY | Date before any column was quoted. Unlike the prior edition, the board had not rolled its daily boundary when captured, so its %Chg column measures the full Friday session against Thursday's close and is reproduced directly. TradingEconomics prints the change column unsigned; signs below are taken from the %Chg column. The dollar-index cross-check is Investing.com's board at its 15:59:31 stamp. | | Pair | Level | 1-Day | Weekly | Monthly | YTD | Read |
|---|
| DXY | 98.817 | −0.08% | −0.85% | −2.28% | +0.50% | Investing.com's board closed 98.83, −0.04%, range 98.56–98.91. The dollar did not bid a hawkish repricing | | EUR/USD | 1.16784 | +0.01% | +0.94% | +2.34% | −0.53% | Unchanged to four decimal places on a day the bund did not move either | | USD/JPY | 158.985 | −0.04% | −0.21% | −2.55% | +1.42% | Flat for a second session. The yen would not move on risk-off Thursday and would not move on risk-on Friday | | GBP/USD | 1.36420 | +0.08% | +0.80% | +2.00% | +1.35% | Gilts outperformed Treasuries by 6 bp and sterling gained 8 bp | | USD/CHF | 0.80134 | +0.11% | −1.47% | −1.61% | +1.07% | The franc weakened on a risk-on day — the textbook response, and the first one this week | | USD/CNY | 6.72019 | −0.08% | −0.34% | −0.81% | −3.68% | Managed; a fifth of the won's move | | USD/KRW | 1386.01 | −0.61% | −2.15% | −6.16% | −3.79% | The won strengthened on a day the Kospi rose 0.88% — see the take | | USD/TWD | 31.8370 | −0.11% | −0.58% | −1.74% | +1.56% | Barely moved with SOX down 0.51% | | AUD/USD | 0.71691 | +0.79% | +1.22% | +2.45% | +7.44% | The best G10 performer, on the day copper finally joined the metals bid | | NZD/USD | 0.59674 | +0.59% | +1.28% | +2.60% | +3.67% | Confirms the antipodean move was commodity-led, not idiosyncratic | | USD/CAD | 1.37592 | −0.22% | −0.84% | −2.33% | +0.28% | The loonie firmed with crude down 0.22% — a dollar move, not an oil move | | USD/MXN | 16.8930 | −0.25% | −0.77% | −2.90% | −6.25% | Still the strongest major EM currency of 2026 | | USD/SGD | 1.26891 | −0.21% | −0.79% | −1.72% | −1.34% | Added as the Asian funding cross-check |
| | The take — the dollar refused to buy a hawkish Fed, and that is the most important thing in this section. On Friday the 2-year cheapened 5 bp, the whole coupon curve cheapened 4–5 bp, and the market added 4.0 points to the September hike probability and 5.5 bp to the 2027 terminal rate (§8). The textbook response is a stronger dollar. DXY fell 0.08%, and on Investing.com's independent board it fell 0.04%. Meanwhile gold rose 1.97% to a three-month high (§11). A currency that will not rally on a hawkish repricing, alongside a metal that rallies without needing the currency to fall, is the same trade seen from two sides — and Bloomberg named it on Friday evening: "Bessent's Bond Maneuvers Giving Global Debasement Trade New Life." The dollar is not being priced off the policy rate this month; it is being priced off the fiscal and institutional questions that sit behind the buyback programme and behind Jackson Hole. | | The Korean mechanism failed its first out-of-sample test, and the failure is informative. The prior edition proposed that Korean equity flows are domestically levered, so a Kospi drawdown forces repatriation and lifts the won while a Kospi rally should weaken it, and it drew the falsifiable prediction explicitly. Friday: Kospi +0.88% and the won strengthened 0.61% — the opposite sign. The two-day version of the rule survives (a 5.80% drawdown moved the won 1.82% stronger; a 5.89% rally moved it 0.21% weaker), but the third observation breaks the directional claim outright. The honest revision is that the won is not trading Korean equity direction at all on ordinary days; it is trading the dollar, and it happens to have the highest beta to it in the Asian complex — −2.15% on the week and −6.16% on the month against a DXY that fell 0.85% and 2.28%, a beta of roughly 2.5× on both horizons. The tradeable statement is now about beta, not about the equity market: the won is the highest-octane expression of a falling dollar in Asia, and it should be sized as such rather than as a Kospi proxy. | | Read what the commodity currencies did and what the havens did not. AUD +0.79% and NZD +0.59% were the two largest G10 moves, on the session copper broke its two-day refusal with +1.72%. For two consecutive sessions this report flagged the Australian dollar as ignoring a precious-metals rally; the moment the industrial metal joined, the currency moved. That is a clean confirmation that the antipodean currencies trade the industrial complex and not the monetary one, and it means AUD is a poor expression of the gold trade and a good expression of the copper trade. On the other side, USD/JPY moved 0.04% and EUR/USD moved 0.01% — on a day with a 500-point Dow rally, a 22% weekly move in bitcoin and a 5 bp parallel curve shift. Two of the three largest currency pairs in the world did nothing at all. USD/CHF's +0.11% is the only orthodox risk-on print in the haven complex, and it is small. |
| | Commodity | Settle / level | Chg | %Chg | Weekly | Monthly | YTD | Driver |
|---|
| WTI (Oct, NYMEX front) | $86.64 | −$0.19 | −0.22% | +5.15% | −0.22% | +50.89% | Range $85.81–$87.50 on 209,120 lots; first decline in seven sessions | | Brent (Oct, ICE front) | $93.93 | +$0.15 | +0.16% | +6.04% | −0.21% | +54.26% | Range $92.76–$94.82 on 286,610 lots; the two benchmarks diverged in sign | | Gold (Comex Dec) | $4,661.60 | +$90.20 | +1.97% | +5.30% | +11.54% | +6.66% | Range $4,565.51–$4,690.11 on 190,440 lots. Highest since 15 May; fifth straight weekly gain | | Silver (Comex front) | $69.010 | +$0.905 | +1.33% | +6.64% | +15.47% | −3.23% | Range $67.980–$70.075 — traded above $70 and closed 1.5% under it | | Platinum | $1,891.40 | +$52.30 | +2.84% | +7.66% | +14.40% | −8.63% | The largest percentage gain in the complex for a third consecutive session | | Copper (Comex front) | $6.5800 | +$0.1110 | +1.72% | −0.36% | +1.93% | +15.73% | Range $6.4865–$6.6183 on 47,760 lots. The non-participant participated | | Natural gas (Henry Hub) | $2.7550 | +$0.0220 | +0.81% | +0.81% | −5.81% | −25.26% | Still the only major commodity down more than 20% on the year | | Gasoline (RBOB) | $3.3220 | +$0.0591 | +1.81% | +4.33% | −2.71% | +94.17% | Rose while crude fell — the crack widened for the first time in three sessions | | Heating oil | $4.4625 | −$0.0178 | −0.40% | +4.19% | +7.56% | +110.35% | Fell nearly twice as fast as crude; the distillate crack compressed again |
| | Basis caveats, stated before the analysis. Settles for WTI, Brent, gold, silver and copper are Investing.com per-contract historical boards for the front contract, verified internally consistent within this session's board; platinum, natural gas, gasoline and heating oil levels and all weekly/monthly/YTD columns are TradingEconomics on its Aug/21 dated rows. Three disclosures. First, gold's basis: TradingEconomics prints spot gold at $4,607.35 against Bloomberg's board at $4,680.60 and the Comex December future at $4,661.60. The TradingEconomics level is 1.59% below Bloomberg, outside this report's ~1% tolerance, so the TradingEconomics gold level is withheld and only its ratio columns are used; the level quoted is the Comex December settle, which agrees with Bloomberg to within 0.41%. The residual spot-to-December gap of roughly 1.2% is carry, not disagreement. Second, crude's basis: TradingEconomics' crude level of $86.640 matches the Investing.com October settle exactly, and Bloomberg's evening board at $87.06 sits 0.48% above it on the electronic session — all three within tolerance, so nothing is withheld. Third, board re-basing: Investing.com's boards have re-based since the prior edition — Friday's board reads $86.83 for 20 August WTI against the $86.20 published Thursday, and $4,571.40 for 20 August gold against $4,582.44 — so every day-on-day change above is computed within this session's board and the cross-edition discrepancy is documented in Data Notes rather than smoothed over. | | The take — the crack spread finally answered, and it answered against the barrel. For two consecutive sessions this report ran the same test: a genuine physical-supply threat widens product cracks, and Wednesday and Thursday both compressed them while crude rallied on Iran headlines, which said the bid was in the paper barrel. Friday ran the test in reverse and got the confirming answer. Crude fell 0.22% and gasoline rose 1.81% — the gasoline crack widened by roughly 2.0 points in a session on a day the sanctions story went quiet. Products are now leading crude in the direction that physical scarcity would imply, on a day with no war headline, which is what a real product-market tightness looks like as distinct from a macro bid. Heating oil's −0.40% says the tightness is in the light end only. Positioning frames the whole thing: crude is +50.89% year to date and −0.22% on the month, so the entire six-session bounce has done nothing but round-trip August, and the Brent–WTI spread at $7.29 is where the geopolitical premium actually sits — Brent rose on the day WTI fell, which is the seaborne barrel pricing the Hormuz question and the landlocked one not. | | The metals bid broadened, and that is a different trade from the one it was on Wednesday. Wednesday and Thursday were gold and silver alone, with copper down 0.09% and then down 0.42% while the precious complex ran 4.6% cumulatively. Friday: copper +1.72%, gold +1.97%, silver +1.33%, platinum +2.84% — all four together, with the dollar index down 0.08% and real yields higher across the coupon curve. Gold's year to date has gone from +0.49% on Tuesday to +6.66% on Friday; silver's from −11.04% to −3.23%, recovering nearly three-quarters of 2026's loss in four sessions; platinum's from −13.5% to −8.63%. When only the monetary metals move, the correct read is debasement. When the industrial metal joins on the same day, the read splits: either the debasement bid has broadened into anything scarce and physical, or a reflation impulse has arrived — and the second one is testable next week, because a reflation copper holds its gain through a hawkish PCE print and a debasement copper does not. Basic materials at +2.99% was the best equity group on the board (§3), so the equity market has already taken the second view. |
| | 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 — the calendar spread worked on day one; take a quarter and raise the stop | | Mark on the prior edition's rates idea, first. The book is long ZQU6 (September 2026) against short ZQZ6 (December 2026), DV01-matched one-for-one at $41.67 per basis point per contract, entered Thursday at 96.325 / 96.160 for a spread of 16.5 bp. Friday's mark: ZQU6 96.325, ZQZ6 96.140, a spread of 18.5 bp — +2.0 bp in a single session, worth $83.34 per contract pair. No invalidation triggered: the spread did not approach 13.0 bp; the September cumulative hike printed 39.0% on Investing.com and 40.1% on CME, both under the 50% stop; and no 2026 meeting shows a non-zero cut probability. The thesis was that the hike keeps sliding right and every session of sliding lifts the front contract against the back. Friday did something better than that — the whole strip repriced hawkishly and the spread still widened, because December absorbed more of the move (cumulative hike 66.7% → 71.2%) than September did (35.3% → 39.0%). | | The modal path, the base case and the tails. Modal path: hold on 16 September (CME 59.9%, Investing 61.0%, ease 0.0%); hike by 28 October, where cumulative hike 53.1% now exceeds hold at 47.0% for the first time in this window; one 25 bp hike delivered by 9 December, modal 3.75–4.00% at 45.2% with cumulative hike 71.2%; the modal range steps to 4.00–4.25% from the June 2027 meeting; terminal 4.075%, drawn by the September and October 2027 contracts at 95.925, 5.5 bp higher than Thursday's 4.020%; year-end 2027 modal 3.75–4.00% at 30.9% with cumulative easing of just 1.3%, down from 1.7% a day and 2.8% a week earlier. Base case: the front end has stopped being a spectator. A parallel 4–5 bp coupon shift with the bill anchored (§6) plus +4.0 points of September hike probability in one session is the market pricing the policy path, not the term premium — and it did it on a week whose data were disinflationary. The driver is the identity and framing of the Chair, not the inflation print. Tail one, dovish: a core PCE deflator at or below +0.2% m/m on 26 August lands on a strip that has just taken a quarter of its 2027 easing out and would compress the spread hard from the front. Tail two, hawkish: a Warsh keynote on 28 August that treats 3.50–3.75% as a floor takes September through 50% and the spread through 22 bp. Practical implication: the September-versus-December leg has now captured two of the roughly 5.5 bp that separate 16.5 bp from the top of its plausible band, and it did so before either event. Take a quarter of the position off at 18.5 bp, carry three-quarters, and raise the stop from 13.0 bp to 15.5 bp so the trade cannot give back more than it has made. | | Expression: long ZQU6 / short ZQZ6, DV01-matched one-for-one, three-quarters of the Thursday size. Catalyst: PCE deflator 8/26 08:30; Jackson Hole 8/27–8/29 with Warsh Friday morning; claims 8/27 08:30; the first buyback operation 9 September; the 16 September FOMC. Invalidation: the spread back through 15.5 bp; or the September cumulative hike printing above 50% on either vendor; or any 2026 meeting showing a non-zero cut probability. Sizing: one-for-one DV01 at $41.67 per basis point per pair; risks 3.0 bp to make a further 3.5. | | 2. Long the 20-year against the 30-year, on the buyback's own map — hold; the weekly table is the argument | | Expression: long the 20-year Treasury point against short the 30-year, DV01-matched, small. Mark: entered Wednesday with 20s30s at +2 bp, +3 bp Thursday, back to +2 bp Friday — flat from entry on the day and flat from entry on the week. The daily mark is uninteresting; the weekly one is not. Across five sessions the 20-year is exactly unchanged at 5.25% while the 2-year cheapened 7 bp, the belly 6–7 and the 30-year 2. 20s30s has steepened 2 bp on the week and 2s30s has flattened 5. Treasury named the 10–20 and 20–30-year buckets and the 20-year is the only tenor on the curve that has given nothing back. Catalyst: the operation schedule; the first execution on 9 September; the quarterly refunding statement that sizes the buckets. Invalidation, unchanged: 20s30s back through −2 bp; or Treasury clarifying that operations will be spread evenly across the 10–30y complex; or a 30-year auction tailing more than 2 bp. Sizing: small, unchanged. | | 3. New — short the utility complex against the S&P 500 | | Expression: short an equal-weighted large-cap utility basket against long the S&P 500, beta-adjusted, small. Thesis: the utility sector is the longest-duration cash-flow stream in the equity index, and on Friday it was the only group to fall more than 0.1% on a day the index rose 0.43% — −1.97%, a 4.96-point spread to the best group — with real estate at −0.05% as the corroborating non-participant. The setup is the sector's own arithmetic against the curve: −3.50% on the week, −7.44% on the month, −0.68% year to date while the 20-year sits at 5.25% and the 30-year at 5.27%, both within 2 bp of their highs for this cycle. Friday's move was not a sector event — there was no utility headline; Wells Fargo actually upgraded Portland General Electric to Overweight that morning — it was a rate event that only one part of the equity market chose to price. If §6's parallel coupon shift is right that the market is repricing the policy path rather than the term premium, the discount rate applied to a regulated 30-year cash flow goes up in either the hawkish or the fiscal branch. Catalyst: PCE deflator 8/26 08:30; Warsh at Jackson Hole 8/28; the 9 September buyback operation, whose failure would cheapen the long end further. Invalidation: the 30-year back inside 5.10%, which would remove the mechanism; or the utility basket outperforming the S&P by more than 2% over five sessions; or a September hike probability back below 30% on either vendor. Sizing: small. The sector has already fallen 7.4% in a month, so this is a continuation trade entered late, and it should be sized as one. | | 4. Long the memory and storage complex against short megacap platform technology — hold at small; day one lost | | Expression, as struck Thursday: long an equal-weighted memory and storage basket against short an equal-weighted megacap platform basket (Apple, Amazon, Alphabet, Microsoft), dollar-neutral, small. Mark: the pair lost on its first session. Micron fell 0.78% to $966.72 while the platform basket averaged +0.14% (AAPL −0.60%, AMZN −0.57%, GOOGL +1.24%, MSFT +0.47%), for roughly −0.9 points. The wider read is worse: SOX underperformed the Nasdaq 100 by 84 bp, precisely inverting Thursday's +125 bp, and Marvell fell 5.57%. The honest reading: Thursday's split ran on a Korean buyback headline, and a headline-driven factor split reverses when the headline stops being new. Nothing in the thesis broke — no memory name cut a buyback, and the platform names did not re-rate on demand — but the entry was one session too late. Catalyst: Nvidia, 26 August after the close; Marvell, 27 August; the pricing terms of the Broadcom package. Invalidation, unchanged: the megacap basket outperforming the memory basket by more than 4% over five sessions — it is at 0.9 after one; or a Nvidia print that re-rates the platform names on demand rather than on capex; or any memory name cutting or deferring its buyback. Sizing: small, unchanged, and no adds before Nvidia. | | 5. Protection on the CCC cohort funded in IG — hold at a half; the best week the structure has had | | Expression: long CCC-exposed credit protection (or short a levered-loan / CCC-heavy vehicle) against long IG cash. Mark: the protection leg gained again — CCC +5 bp to 1,035 bp on 20 August, a fresh monthly wide — against a funding leg that cost only 1 bp, IG to 82. Over five sessions the ratio is what the structure exists to harvest: CCC +23 bp against IG +2 bp, and the CCC-minus-HY differential at 760 bp, +15 bp on the week and +156 bp on the year. Catalyst: the 24 August FRED update, carrying the first post-Friday credit spread; the record $145.2bn August IG calendar clearing into a cheaper curve; the NY Fed Corporate Bond Market Distress Index on 26 August. Invalidation, unchanged: the differential back through 735 bp, or IG widening beyond 85 bp — which would say the funding leg has become the risk. Sizing: a half, unchanged. | | 6. On-balance-sheet AI funding against off-balance-sheet AI funding — cut to a half; the worst session the pair has had | | Expression: long customer-funded suppliers against short self-funded ones, small. Mark: Marvell −5.57% against Broadcom +1.21% is a 6.78-point loss in one session, and it wiped out roughly two-fifths of the pair's two-day gain. The proximate cause is instructive: BMO Capital initiated Marvell at Outperform with a $250 target on Friday morning — just 5.5% above the price — and initiated Broadcom at Overweight with a $455 target, 23.5% above its. When the incremental sell-side view puts four times more upside on the self-funded name than on the customer-funded one, the discrimination this trade harvests is being priced out by the analyst community rather than by the tape. Invalidation test status: the written stop was "a customer-funded structure trading below its announcement price within ten sessions"; Marvell at $237.04 remains roughly 10% above its pre-announcement level, so the stop is intact. Action: cut to a half and let Nvidia and Marvell's own prints decide it. Catalyst: Nvidia 26 August, Marvell 27 August, the pricing terms of the Broadcom package. | | 7. Short the bank complex against the S&P, into the flattener — closed at a token loss; the mechanism is dead | | Expression: short a large-cap bank basket against long the S&P 500, beta-adjusted, cut to a token on Thursday. Mark: financials +1.06% against an S&P +0.43% cost another 63 bp. The written invalidation, "2s10s back through 55 bp," did not trigger — 2s10s finished at exactly 50 bp, unchanged — but the thesis was that a buyback programme is structurally a flattening instrument and every basis point of flattening is a debit to net interest margin. The curve has now spent three sessions refusing to flatten, and Friday's driver of the financials group was crypto and capital markets, not net interest margin: Goldman Sachs +3.74%, Robinhood +13.70%, Coinbase +8.20%, against JPMorgan +0.01%. Action: close the token. A short whose thesis is rate-driven and whose sector is being driven by something else is a beta position, not a view. The idea can be re-struck only on an actual operation on 9 September that flattens 2s10s through 46 bp. | | Two prior closes, marked forward. The healthcare-versus-semiconductors pair was closed Thursday at its written invalidation, and Friday would have paid it: healthcare +1.25% against SOX −0.51%, +1.76 points. That is the second position this week stopped out one session before its thesis worked — the consumer-discretionary put spread was the first. Both stops were correct as written, and the cost of having written stops is precisely this. Neither is being re-entered on the strength of one session. | | The vol note. VIX closed at 15.13, −5.50%, having given back 79% of Thursday's spike inside a single session, and finished +6.18% on the week. MOVE has not updated for Friday and remains 73.18 on a 20 August stamp (§9), so the only clean matched-date ratio is Thursday's 4.57×. Look at what VIX is being asked to price into a four-day window: the PCE deflator and Nvidia on the same day (26 August), claims on the 27th, and Chair Warsh's first Jackson Hole address on the 28th — into a market 1.60% below its record close, with the 2-year having just broken a three-session freeze, the terminal rate 5.5 bp higher in a session, and a strip that prices zero chance of a cut at any 2026 meeting. Between Monday's open and Wednesday's 8:30 there is exactly one Medium-rated release. That shape argues, again and more strongly than last week, for owning gamma dated 26–29 August and financing it out of Monday and Tuesday rather than out of the wings — the calendar is emptier in front and fuller behind than a 15-handle VIX implies. |
|
The crowded consensuses to stress-test, with the numbers. 1. "The front end is anchored." It was, for three sessions at 4.19%, and then it moved 5 bp to 4.24% in one afternoon with the 3-year alongside it at +5 bp to 4.31% and the 1-year +4 bp to 4.03% — while the 1-month bill did not move at all. The whole coupon curve shifted in parallel and 2s10s finished unchanged at 50 bp. That is a policy repricing, and the strip agrees: September hike 33.1% → 36.1% → 40.1% across a week on CME, and a terminal rate 5.5 bp higher in a single session. The anchor was an assumption, not a fact, and equities spent Friday rallying on the assumption. 2. "The Fed is done for 2026." The strip shows 0.0% probability of a cut at every 2026 meeting for the whole of this reporting window, and now shows cumulative hike of 53.1% by October — a majority for the first time — and 71.2% by December. It got there on a week containing a 206,000 claims print and a Philadelphia Fed price series at a six-month low. A consensus with zero probability on one side is a position; this one has now survived a dovish data week and got more hawkish. The cheapest way to be wrong is a core PCE at or below +0.2% m/m on 26 August, into a market that has removed a quarter of its 2027 easing in five sessions. 3. "Equity vol is correctly priced." VIX at 15.13 is pricing a roughly 0.95% daily move into a four-day window containing the PCE deflator and Nvidia on the same morning, claims, and a first Jackson Hole address by a Chair appointed in May. Bloomberg carried the buy-side version on Friday: "Allspring's Miletti Sees Jackson Hole as Bigger Risk Than Nvidia." The week's VIX path — 14.89 → 16.01 → 15.13, net +6.18% — describes a market that gets nervous for one session and forgets by the next. The ratio to rates vol remains near 4.6× on the last matched date against a five-year average nearer 4×. 4. "The equity market and the bond market are telling the same story." They are not, and Friday is the cleanest example this month. The S&P rose 0.43% and the Russell 0.79% on a day the entire coupon curve cheapened 4–5 bp and the market added 4 points of September hike probability. Only two of eleven equity groups priced it — utilities −1.97% and real estate −0.05%. Either the index is right that a 25 bp hike is irrelevant to 2027 earnings, or the utility sector is right that the discount rate just moved. Both cannot be. Utilities are −7.44% on the month; the S&P is 1.60% from a record. 5. The two-sided geopolitical tape. Washington has threatened "crushing" sanctions on Iran, ruled out negotiations, and the UAE has suspended trade with Tehran. Yet Friday's crude tape was WTI −0.22% and Brent +0.16%, with the gasoline crack widening 2 points — the market pricing product tightness rather than a transit interruption. The asymmetry is unchanged and it is large: an actual Hormuz interruption is not in the price at all, and neither is a de-escalation, which would take a complex up 50.89% year to date back through $80 quickly. The $7.29 Brent–WTI spread is where the risk premium actually sits. 6. Structural watch items. AI capex is now being marked in the equity market before the credit market: Alibaba lost 8.57% and roughly $21bn for spending 25% of revenue on capex that generated new cloud revenue equal to 22% of it, and Nvidia answers the same question on 26 August. Record IG supply — $145.2bn in August, $1,681bn through July, +26.9% year on year — is clearing into an index at 82 bp while the CCC cohort widens 23 bp a week; the core is absorbing what the tail cannot. Citadel has unwound more than 80% of the Situational Awareness risk through 100-plus blocks worth over $4bn, which means a large mechanical seller has been underneath the AI-adjacent tape all month and is now four-fifths gone. The bill curve stays kinked — 1-month 3.80% above 1.5-month 3.77%, 4-month +2 bp to 3.90% — with reserves $207.4bn below their July peak and the reverse repo facility holding $200m. And the 20-year is unchanged on the week while everything around it cheapened, which is either the buyback working or the market front-running an operation that has not happened. 7. What VIX is and is not pricing. At 15.13 it is pricing a market that rose 0.43% and might do so again. It is not pricing: a front end that broke a three-session freeze; a terminal rate that moved 5.5 bp in a session; the PCE deflator and Nvidia arriving within twelve hours of each other; a new Fed Chair's first symposium address; a 22% weekly move in bitcoin driven by a bill that has not passed; or a $100bn debt structure still being negotiated by an S&P 500 constituent. There is no macro release of any rating between now and Monday 11:00, and exactly one Medium-rated item before Wednesday's 8:30 — which is why the optionality is cheap and why it is the calendar, not the level, that makes it cheap. |
|
| Sources — used this session: Bloomberg.com (US edition — /markets and /markets/rates-bonds, rendered in the local Chrome browser), WSJ.com (/finance/stocks and /market-data/bonds, rendered), CNBC daily market live blog and Market Insider, Investing.com (major-indices, Nasdaq 100, PHLX Semiconductor, US-30 components, trending-stocks, dollar-index, MOVE, bitcoin and per-contract commodity historical boards, plus the Fed Rate Monitor), Finviz group screener (Performance table view), U.S. Department of the Treasury daily par yield curve Text View, CME FedWatch, Federal Reserve Bank of New York markets API and Economic Indicators Calendar, FRED (ICE BofA OAS series, IORB, WRESBAL, RRPONTSYD), TradingEconomics currency and commodity boards, Nasdaq earnings calendar API, StockAnalysis.com, Cbonds, and Kansas City Fed symposium material. |
| Full Data Notes & Conflicts and the complete categorized Source Links appendix are in the companion file US_CrossAsset_Daily_2026-08-21_DataNotes.txt. |
| U.S. Stock, Fixed Income & Cross-Asset Closing Daily — Friday, August 21, 2026. Prepared for institutional investors. Not personalized investment advice; verify independently before acting. |
|