U.S. Stock, Fixed Income & Cross-Asset Closing Daily Monday, September 21, 2026 · U.S. cash close Institutional readership. Data captured 18:05-18:55 ET. Section 12 contains trade ideas that are not personalized investment advice. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Source: Finviz group screener, Performance table view (g=sector&v=140&o=name), read in the local Chrome browser after the close. 1-Day is the Change % column, 1-Week Perf Week, YTD Perf YTD. Finviz classification, not GICS. Seven green, four red, and a best-to-worst spread of 5.52 percentage points — more than two and a half times Friday's 2.11 and the widest of the reporting window. The shape is a barbell with nothing in the middle: two groups above two and a half per cent, one below minus two, and the remaining eight inside a 1.56-point band from consumer cyclical's +1.42% to basic materials' -0.39%. That is a market expressing two views, one on silicon and one on the barrel, and no view at all on anything else. The YTD reconciliation holds. Compounding each group's 18 September YTD by Monday's one-day move reproduces the published YTD to 0.05 percentage points or better at all eleven groups, and to 0.01 or better at nine. Worked examples: technology 1.2639 x 1.0252 = 1.29575, or +29.58% against a published +29.58%, deviation zero; communication services 1.0049 x 1.0343 = 1.03937 → +3.94% against +3.94%, deviation zero; energy 1.3917 x 0.9791 = 1.36262 → +36.26% against +36.25%, deviation 0.01. The largest deviation is industrials at 0.05 pp — 1.0880 x 1.0028 = 1.09105, or +9.11% against a published +9.06% — on a group that moved 0.28%, which is too small a price move to generate five basis points of drift and is therefore flagged as a probable constituent change and carried forward rather than withheld. The composition traps run one way and it is the reverse of Friday's. Communication services at +3.43% is Meta +11.43% and Warner Bros. Discovery +10.79% against Charter -3.76% and Paramount Skydance -2.94%, a 15.19-point range inside the best group on the board, and the two winners are large enough to carry a bucket their carriers dragged. Technology at +2.52% is broad for once: Intel +12.14%, AMD +9.95%, Qualcomm +9.29%, Akamai +12.33% and Super Micro +5.40% at the top with HP -4.19% the only material negative, so the 12.65-point intra-group range of Friday has compressed to something closer to a sector move. Energy at -2.09% is the refining split inverted: Marathon Petroleum -5.30% and Valero -4.84% are the two worst names in the group on a day Halliburton fell 0.74% and Kinder Morgan 0.38%, so the downstream took roughly six times the upstream's loss. And consumer cyclical at +1.42% is worth naming because Amazon is inside it: Amazon +1.87% is the single largest weight, and stripping it leaves a bucket that still rises on General Motors +2.13% and Lululemon +3.30% but by materially less than the headline.
Levels and percentages from the 494-line Investing.com S&P 500 component capture taken after the close unless another vendor is named. The capture loads 494 of roughly 500 index lines, so counts below are ratios of what loaded, not a census. Upside, with catalysts Akamai (AKAM) $117.41, +12.33% — the largest single-name gain in the capture. No dated catalyst for this session was found across Investing.com, Bloomberg, WSJ or the wire aggregators; the company's two prior double-digit sessions this year both had named causes, and this one is reported without one rather than attached to a stale item. It sits inside the same AI-inference bid that moved Intel and Qualcomm. Moderna (MRNA) $172.93, +12.26% — FDA clearance of the updated 2026-2027 Spikevax and mNEXSPIKE formulations for the JN.1-lineage XFG variant, alongside melanoma data for intismeran. Wolfe Research upgraded to Peer Perform from Underperform citing roughly $9.2bn of potential peak sales across four cancer indications; Argus moved to Buy with a $180 target; Barclays raised its target to $125 from $48 while holding Equal Weight. Intel (INTC) $121.78, +12.14% — a Micro LED development partnership with AUO Optronics, with reported SK Hynix discussions alongside it. Meta Platforms (META) $741.25, +11.43% — Wells Fargo's Ken Gawrelski raised his target $156 to $796, Buy maintained, on the strength of recent AI product launches; the Muse assistant reached number one on the U.S. App Store. Warner Bros. Discovery (WBD) $30.80, +10.79% — WSJ reported advanced settlement talks between Paramount and California's attorney general over the $110bn merger. Concessions reported include $1.5bn of California production investment, no sale of the studio lots, penalties against a thirty-films-a-year pledge, a possible cable-channel divestment and an editorial-independence board. AMD (AMD) $615.52, +9.95% — reported price increases of about ten per cent on selected AI accelerators and GPUs. Qualcomm (QCOM) $194.23, +9.29% — the AWS custom-silicon and optical-networking agreement, with a warrant over 25 million shares at $161.26. The rest of the upside tier, left plain because the run reads better that way: GoDaddy +7.50%, Shopify +7.33%, AppLovin +7.18%, Corning +5.89%, Super Micro Computer +5.40%, United Airlines +5.22%, Lam Research +4.92%, CrowdStrike +4.92%, Monolithic Power +4.90%, Applied Materials +4.49%, Synopsys +4.38%, Cadence Design +4.24%, F5 Networks +4.18%, Amphenol +4.09%, KLA +3.94%, Delta Air Lines +3.62%, Equinix +3.52%, Agilent +3.50%, Lululemon +3.30%, Fortinet +3.17%, Ulta Beauty +3.12%, Estee Lauder +3.11%, Palantir +3.07%. Downside, with catalysts Mohawk Industries (MHK) $117.27, -7.00% — the worst line in the capture, and the housing-cyclical read into Thursday's August new home sales print. Marathon Petroleum (MPC) $402.38, -5.30% and Valero (VLO) $393.27, -4.84% — refining margins on a session crude fell over four per cent and the distillate product fell with it. United Parcel Service (UPS) $94.75, -4.35% and FedEx (FDX) $295.79, -2.59% — freight fell on a broadly positive tape, with Union Pacific -3.49% and CSX -2.36% alongside them. Four transport names lower on a 1.49% index day is a coherent negative signal about volumes, not a single-company event. HP Inc (HPQ) $32.96, -4.19% — the only material technology decline in a group that rose 2.52%. Phillips 66 (PSX) $261.75, -4.17%, CF Industries (CF) $123.27, -3.47%, ConocoPhillips (COP) $127.54, -3.25%, Exxon Mobil (XOM) $158.30, -3.20%, Occidental (OXY) $57.25, -2.71% and Chevron (CVX) $203.67, -2.79% — the energy complex, in order of damage. Charter Communications (CHTR) $123.35, -3.76% and Paramount Skydance (PSKY) $9.91, -2.94% — the two communication-services names that fell while their group gained 3.43%. Campbell's (CPB) $19.93, -3.67%, General Mills (GIS) $35.41, -2.51%, J.M. Smucker (SJM) $119.88, -2.50% and Bunge (BG) $112.61, -2.57% — packaged food lower into General Mills' Wednesday morning report. Steel Dynamics (STLD) $229.99, -2.24% and Nucor (NUE) $242.40, -2.41% — a second session of guidance-driven weakness after Friday's 6.32% Nucor decline. Analyst actions Ciena (CIEN) — Evercore ISI to Outperform from In-Line, target $375 to $550, a 46.7% increase and the largest on the day's list. Microsoft (MSFT) — Cantor Fitzgerald Overweight, target $522 to $608, roughly 21% above the $501.61 close. Waters (WAT) — Goldman Sachs Buy, target $465 to $515. Invitation Homes (INVH) — Mizuho to Outperform from Neutral, target $31 to $32; American Homes 4 Rent (AMH) — Mizuho to Outperform from Neutral, target $35 to $36, two residential-REIT upgrades on the same morning from the same house. Macerich (MAC) — Evercore ISI to Outperform from In-Line, target held at $26. Cutting the other way: Credo Technology (CRDO) Mizuho target $290 to $245; TTM Technologies (TTMI) Needham $220 to $175; T-Mobile (TMUS) JP Morgan $275 to $260; Colgate-Palmolive (CL) Piper Sandler $98 to $95. Inspire Medical (INSP) — Freedom Capital Markets Buy, target $71 to $97. Cleveland-Cliffs (CLF) — GLJ Research Buy, target $15.60 to $17.48. Ciena, Credo, TTM, Inspire, Securitize, Candel Therapeutics and Cleveland-Cliffs are flagged as names whose S&P 500 membership was not confirmed against this session's 494-line component capture; they are reported as analyst actions, not as index movers. The intraday fade worth recording The Nasdaq Composite printed 27,183.93 and closed 27,122.09, giving back 61.84 points, or 22.7% of the 272.34-point advance measured from its own low to its own high, and finishing 68.12 points below the 27,190.21 52-week high it came within 6.28 points of touching. The S&P closed 14.42 points off its high and SOX 58.44 points off 12,491.61. Three indices that ran into the close and did not hold is a smaller fade than Friday's but it is the same shape, and it is the reason the record-high framing carried by several wire summaries is reported here as a level and a gap rather than as a label.
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 come from the Nasdaq earnings calendar API for each date, screened name by name against the 494-line Investing.com S&P 500 component capture taken the same session. Nasdaq's buckets are before-open or after-close rather than clock times, so no clock times are asserted; confirm every time against company investor relations before trading a date.
Tue 9/22. BMO: AutoZone (AZO). Wed 9/23. BMO: Cintas (CTAS), Paychex (PAYX), General Mills (GIS). Thu 9/24. BMO: Darden Restaurants (DRI). AMC: Costco Wholesale (COST). Fri 9/25. No S&P 500 reporter on either bucket.
Mon 9/28. No S&P 500 reporter on either bucket. Tue 9/29. BMO: Carnival (CCL), CarMax (KMX). Wed 9/30. BMO: Jabil (JBL), FactSet (FDS), Conagra Brands (CAG). AMC: Micron Technology (MU). Thu 10/1. BMO: Accenture (ACN), McCormick (MKC). AMC: Nike (NKE). Fri 10/2. No S&P 500 reporter on either bucket. Changes vs. the prior calendar (9/18 report):
U.S. Department of the Treasury daily par yield curve, month-scoped Text View for September 2026, read with a server-side fetch. The 21 September row published on a first attempt at about 18:15 ET. Rate up = red. Below one year only the 1-month and 3-month appear in the table; the other bills are extracted and cited in prose and in Section 9 block b where they carry a financing story.
Shape and diagnostic. This is a bull flattener pivoting on the 2-year, and the pivot is the whole message. The front is nailed down — the 2-year did not move at all and the 1-year cheapened a basis point — while everything from five years out richened three to five. A curve that rallies only where the policy rate is no longer the dominant input, on a session crude fell over four per cent and a voting regional president argued for more tightening, is repricing the inflation path rather than the policy path. Term premium is doing the work that a week ago the meeting calendar was doing. The spreads. 2s10s flattened 5 bp to 20 bp and is 12 bp flatter on the week, the single largest weekly move on this table; 3M10Y flattened 8 bp to 79 because the bill cheapened while the note richened, which is a different mechanism from the same arithmetic sign. 2s30s at 53 bp is 16 bp flatter on the week, so the whole week's flattening has come from the front cheapening rather than the back rallying. 20s30s held -4 bp for a third consecutive session and remains the one spread this curve refuses to resolve. Vendor cross-check. WSJ's 17:04 ET quotes read 2-year 4.753%, 10-year 4.956% and 30-year 5.286% against the official par 4.76, 4.96 and 5.29, inside two basis points at every coupon tenor, and its change fields (+0.1 bp, -4.1 bp, -3.9 bp) reconcile to the par moves for a fourth consecutive session. Bloomberg's board carried the 10-year at 4.95. The off-table bills. The financing story sits in tenors this table no longer prints: the 1.5-month cheapened 4 bp to 4.02%, the 4-month 2 bp to 4.26% and the 6-month 3 bp to 4.27%, against a 1-month that richened a basis point. Section 9 block b reads that against Monday's auction stops.
Source: TradingEconomics United States calendar, read in the local Chrome browser after the close. The board serves times in UTC+8 and its date headers are shifted accordingly; every time below has been converted to ET and is stated in ET. Sensitivity is this report's own rating and drives which releases Section 1 must name. Consensus figures are the board's consensus column where populated and its own forecast where not, flagged as such. Current week — remaining releases only Tuesday 22 September
Wednesday 23 September
Thursday 24 September
Friday 25 September
Next week Monday 28 September. Dallas Fed manufacturing index for September at 10:30, prior 11.6, Low. Three-month and six-month bill auctions at 11:30 against Monday's 4.015% and 4.155% stops, Medium. Tuesday 29 September. S&P CoreLogic Case-Shiller home price index for July at 09:00, prior +2.1% year on year, Low. FHFA house price index for July at 09:00, prior 0.0% month on month, Low. Conference Board consumer confidence for September at 10:00, prior 89.4, High. JOLTS quits for August at 10:00, prior 3.056M, Medium. Dallas Fed services at 10:30, Low. Fifty-two-week and six-week bill auctions at 11:30, Low. Wednesday 30 September. MBA mortgage applications at 07:00, Low. ADP employment change for September at 08:15, prior 38K, board forecast 75K, High. Core PCE price index month on month for August at 08:30, prior +0.2%, Very high. Personal income at 08:30, prior +0.4%, and personal spending, prior +0.2%, both High. Final Q2 GDP at 08:30, prior +2.1%, consensus +1.5%, Medium, and the GDP price index, prior +3.6%, Medium. Quarter-end. Thursday 1 October and Friday 2 October. The captured board returned no rows for either date at capture time and nothing is asserted for them. On the Bureau of Labor Statistics' standing first-Friday convention the September employment report would fall on 2 October; it did not appear on the board this session and is named as a convention rather than a confirmed listing. This gap is recorded in Data Notes. The look-ahead
Order of when they can move the Fed card. Wednesday's flash PMIs at 09:45 are the first broad activity read since the hike and the first that touches the median company rather than the megacap; a services print through 56.5 against a 56.0 consensus makes the October hike harder to price out. Thursday's claims at 08:30 matter more than usual because the prior is 196K, below two hundred thousand, and a consensus of 203K asks for deterioration that has not been arriving. Friday's Michigan one-year inflation expectation, with a 4.6% consensus against a 4.0% prior, is the one number on the board that speaks directly to the thing Musalem said on Monday needs more tightening to fix. And the core PCE print on 30 September is the release that actually decides the October meeting; everything before it is positioning. Completed prints may be cited here where they change the forward distribution, and one does. The Chicago Fed National Activity Index came in at -0.04 for August against a -0.06 consensus, with July revised up to +0.08, so the first hard-activity read since the hike landed close to expectation and slightly better than feared. It is not a Very-high release and it did not move the market, but it removes one of the two arguments for the October hike being priced out, the other being the labour market that reports on Thursday.
Current target range: 3.75%-4.00%, raised a quarter point on 16 September by a 12-0 vote, with interest on reserve balances at 3.90% and the overnight reverse repo offering rate at 3.75%. CME FedWatch headline — 28 October 2026 meeting.
Data as of 21 Sep 2026, 05:02:53 p.m. CT (6:02 p.m. ET), read from the FedWatch probability table; the page's own last-updated stamp is 05:18:31 p.m. CT. The 1 WEEK legend reads 14 September and the 1 MONTH legend 21 August, both advanced from the prior edition, so all four columns are used rather than marked chart-read. Column provenance, the live-read correction and the vendor gap are in Data Notes. (a) Current-year meeting distributions Investing.com Fed Rate Monitor, updated 21 Sep 2026 05:55 p.m. EDT. Format: current [prior day] [prior week]. Modal range in bold.
October sums to 100.0% and December to 99.9% on the vendor's rounding. The card's current and prior-day columns are identical at both meetings and ZQV6 and ZQZ6 were unchanged at 96.105 and 95.835, which is consistent with a 2-year that did not move; CME has the October hike 1.1 points lower on the day. (b) Next-year meeting path
No modal bucket changed. The implied terminal rate at the cheapest contract is 100 - 95.315 = 4.685%, against 4.680% on Friday. (c) Year-end probability ladders Year-end 2026 — the 9 December meeting.
Year-end 2027 — the 8 December meeting.
Transparent rounding. The 2026 ladder sums to 99.9% and the 2027 ladder to 100.0%, the residual sitting in the vendor's own rounding rather than in any omitted outcome. All outcomes are stated relative to the 3.75%-4.00% target range.
(a) IG and HY credit spreads ICE BofA option-adjusted spreads via FRED, read from the plain /data/<SERIES> tables with a server-side fetch. FRED publishes with a one-business-day lag: the levels below carry the 18 September 2026 as-of date, not the 21 September close. Same-day direction is cross-checked against the cash proxies underneath. 1-Week is versus the 11 September row.
CDX — the six-step ladder was worked and all six steps were executable, the fourth clean six of the reporting window. (1) Bloomberg in Chrome: /markets rendered and a full-text scan returns zero occurrences of the index name and of the calculating agent's name. (2) WSJ Market Data bonds page: rendered fully, Treasury quote table populated and stamped 5:04 p.m. EDT, and a full-text scan returns zero occurrences. (3) Cbonds rendered normally in the Chrome extension, and its CDX.NA.IG 5Y and CDX.NA.HY 5Y records both carry a previous-value stamp of 17/09/2026, IHS Markit named as the calculating organisation and the basis-point figure masked behind the request-access wall — one day older than the record this report found on 18 September, so Cbonds is now running behind rather than advancing. (4) ICE: ice.com/data-services/indices returns a 404. (5) FT: markets.ft.com loads and /data/indices returns an error page, a sixth consecutive session. Barchart returned a CloudFront 403 "Request blocked" on its own symbol search, which is a data failure at the source rather than a tooling refusal. (6) Cash-market proxies, labelled as proxies: HYG closed $78.68, +0.19%, and LQD $105.09, +0.37%. No CDX level is published here. The barbell finally cracked at the tail. The IG credit spread tightened a basis point to 77 bp and the HY spread two to 268, both new tights for the reporting window, while the CCC tail widened 7 bp to 1,083. That is the first session in the window in which the three series have moved in genuinely different directions rather than together or not at all, and it takes the CCC-minus-HY differential 9 bp wider to 815 bp after a flat print and three narrowings. On the week the shape is the same: IG 3 bp tighter, HY 3 bp wider and the CCC tail 7 bp wider, against -2 bp, -15 bp and +195 bp on the year. Investment-grade credit is at its best level of 2026 and the weakest cohort in high yield is near its worst, in the same tape, on the same day the equity index rose 1.49%. The cash proxies confirm the direction and the arithmetic says most of it was duration. LQD closed $105.09, up 0.37%, on a session the 10-year richened five basis points; a fund with roughly seven years of effective duration earns about 0.35% on that move, so essentially the whole gain is the curve and the one basis point of spread tightening is rounding. HYG rose 0.19% to $78.68, about half as far, which is the ordinary shorter-duration asymmetry and again says nothing about credit quality on its own. LQD's 52-week low of $104.05 was set on 14 September, so Monday's close sits $1.04 above it. (b) Money-market & funding plumbing New York Fed reference rates, published at approximately 8:00 a.m. ET for the prior business day. The 18 September 2026 row is the latest published at capture, so the endpoint is running one business day behind after three consecutive sessions two behind. These rates are on the new 3.75%-4.00% regime. Rate up = red.
The bills are paying up and the coupons are not, which is the session's cleanest divergence. Monday's auctions stopped 4.5 basis points cheaper at three months and 9.5 at six than the same tenors a week earlier, and the par curve agreed — the 3-month +3 bp, the 4-month +2, the 6-month +3, the 1.5-month +4 — on a day every coupon tenor from three years out richened three to five. Two mechanisms are available and they point the same way. The first is calendar: quarter-end is seven business days out, and dealer balance sheets rent themselves dearly into it. The second is supply, and the 9.5 basis point move at six months against 4.5 at three says the concession is being demanded where the paper is, not where the policy risk is. Nothing here is stress. SOFR printed 3.85%, five basis points below a 3.90% interest on reserve balances, on $2,955bn of volume with a 13 basis point tail band between the first and ninety-ninth percentiles, and it printed exactly the same rate on the day before at $2,992bn. A funding market under pressure does not repeat a rate to the basis point two days running. The reverse repo facility doubled off a tiny base and remains empty in any meaningful sense. Take-up rose to $576m on the 18 September operation from $276m, which is a 109% increase on a number that is four ten-thousandths of the $1.5tn the facility once absorbed. No 21 September operation had published at capture, and the standing repo facility endpoint returned a 400 error on the same query shape that worked for the reverse facility, so no 21 September standing-facility figure is asserted. Reserve balances remain $3.0138tn on the week ended 16 September with no new observation published this session. (c) Rates volatility & swap spreads
The rate-volatility card is one day behind and it inverts what the prior edition published. The Investing.com card carries an 18/09 stamp at 80.64 with a day range of 76.22 to 80.64 and an open of 76.22, so the vendor's bar for 18 September ran from exactly the 76.22 this report published on a 17/09 stamp up to 80.64 — a rise of 4.42 points, or 5.80%. Two of the three internal checks pass: the level sits at the top of its own day range, and the open reproduces the prior published vintage to the hundredth. The card's own change field reads 0.00, which cannot be right against its own open, and its "previous close" field remains 95.74, outside the day range, for an eighth consecutive session. Neither field is used; the level is published with its vintage and the session change is withheld. So rate volatility rose on Friday while equity volatility fell, and on Monday equity volatility rose too. MOVE gained 5.80% on the session VIX lost 4.08%, and then VIX added 0.41% to 14.87 on a 1.49% index rally. The MOVE-to-VIX ratio at 5.42 is up from 5.15 and the numerator did all of it. A rates market whose volatility index climbs nearly six per cent while its 2-year does not move is pricing uncertainty about the path rather than a direction, which is the opposite of what the prior three sessions showed. Swap spreads at the 2-year, 10-year and 30-year were not obtainable from a primary source this session and are not asserted; the substitute evidence is the cash curve, where 20s30s held -4 bp while the coupon sector richened three to five. (d) Issuance, leveraged loans & private credit The primary market said the same thing the equity market did, one rung lower in the capital structure. Bloomberg reported that a CoreWeave-tied data-centre project has begun a junk-bond offering, which follows the same company's $3bn convertible plus at-the-market programme on 17 September — so the data-centre complex has now raised equity, equity-linked and high-yield paper inside five sessions. Bloomberg also carried Truist forecasting asset-backed issuance up 15% this year on narrowing spreads and KKR committing to the Korean artificial-intelligence supply chain after a record $3bn of deals, which is the private-credit and private-equity expression of the same trade the Kospi put on overnight. Post-Labor-Day dollar IG supply remains at its weakest pace since 2020 after an August near a record $130-145bn, with year-to-date supply above $1.68tn, up 27% on 2025. No new large dollar IG deal cleared against the 4.96% ten-year in the session under review and none is asserted.
Source: TradingEconomics currency board, read in the local Chrome browser after the U.S. close. Quote basis: EUR, GBP, AUD and NZD are quoted as dollars per unit of foreign currency, so a fall is a weaker foreign currency; every other pair is quoted as units of foreign currency per dollar, so a rise is a weaker foreign currency. Thirteen of fifteen rows carried a Sep/21 date stamp; USD/JPY and USD/SEK carried 06:15 and 06:16 live stamps and are flagged below. The %Chg column is computed over twenty-four hours against the prior edition's levels for the same vendor; the vendor's own field is reproduced only where it agrees, and every disagreement is named.
The take: the dollar index has now risen four sessions running and Monday is the first of the four with a reason attached to it. DXY closed 100.413, up 0.20%, on a day the American 10-year richened five basis points and the German, French, Italian and British ten-years richened six to ten. A currency that gains while its own long end outperforms nobody is not being paid for yield; it is being paid because eleven of fourteen crosses on this board went the dollar's way on a session global risk appetite was strongly positive. That is the configuration that usually marks a funding bid rather than a carry bid — and the two crosses that did not follow are the two with their own equity stories. The won's five-session losing streak ended and the arithmetic is worth writing out. USD/KRW fell 0.80% to 1,375.28, the largest single-session won gain on this board in the reporting window, on the day the Kospi closed above 7,000 for the first time at 7,007.72, up 1.65%, and the American 2-year did not move. The five sessions that preceded it took the won from roughly 1,341 to 1,386.39, a 3.4% depreciation against an index that rose in four of them, and this report has been recording the failure of every available explanation in turn — carry, equity direction, and then both. Monday supplies the first clean one: a flat American front end plus a Korean equity market making a new high equals an 11.11-won rally, and the won is still 2.04% weaker on the week. One session against five is the beginning of a reversal or the end of a squeeze, and the way to tell them apart is whether it holds through Tokyo's reopening on Tuesday. The commodity currencies did the work the commodity did not. USD/CAD rose 0.36% to 1.40315, the worst major on the board, on a session WTI's November contract fell 4.43% — a loonie that finally trades its own export after ignoring a 2.34% crude decline on Friday with a 0.08% gain. USD/NOK rose 0.39% to 9.44180 on the same mechanism, the krone having also strengthened on Friday's falling barrel. Two petrocurrencies that refused to follow crude down on Friday followed it down on Monday, which reads as a lagged adjustment rather than a new view, and both remain far better on the year than the board's average: the krone is 6.41% stronger year to date against a dollar index up 2.13%. Three vendor fields disagree with their own arithmetic and one pair is stale. AUD/USD printed 0.71183 with a vendor change of -0.04%, which implies a prior of 0.71214 against the 0.71130 this report published for the same vendor twenty-four hours earlier; the computed move is +0.07% and that is what appears above. USD/SEK printed 9.83580 with +0.10%, implying a prior of 9.82598 against a published 9.80609; the computed move is +0.30%, and the row also carries a 06:16 live stamp rather than a Sep/21 close stamp. USD/CHF printed -0.16% against a computed -0.07%. And USD/JPY's row is stamped 06:15, so its 157.337 is a live Tuesday-Asia quote taken while Tokyo cash equities were shut for Respect for the Aged Day on Monday — the computed 24-hour move of +0.30% is published, the level is flagged, and the yen's behaviour is not read into a session Japan did not trade.
Settlement basis, stated, and reconciled to the prior edition. The Investing.com per-contract historical board remains the settle series of record, unchanged for ten editions. Rows were captured at approximately 18:40 ET. Two contracts rolled and are not comparable to the prior edition. Investing.com's crude page states it "rolled over on Sep 20, 2026 to the Nov 26 contract", so WTI below is the November contract against a prior edition that published October; the whole historical series re-based with it. Gasoline RBOB remains on October by the vendor's own contract panel, yet its 17 and 18 September rows were restated upward by 5.76 and 1.54 cents, which is a vendor revision rather than a roll and is recorded in Data Notes. Because this is a Monday, every contract carries an interleaved 20 September electronic row and the vendor's own %Chg column measures against Sunday rather than against Friday's settle — it is not reproduced. All changes below are computed against the 18 September finalised row. Weekly and year-to-date columns are TradingEconomics spot returns, whose header order was verified as Price, Chg, %Chg, Weekly, Monthly, YTD, YoY, Date.
*YTD figures marked with a single asterisk are TradingEconomics spot year-to-date returns, not futures returns on the contracts quoted above. Weekly columns are TradingEconomics one-week changes on the same caveat. The restatement, and the third-party rule went three-for-three to the cent again. The 18 September edition named the Investrade review's and Bloomberg's figures in-line rather than adopting them, on rows that had passed the volume test. Published against finalised for 18 September: Brent $103.19 against $103.87, gold $4,415.90 against $4,424.90, silver $66.785 against $67.149, heating oil $5.0471 against $5.0578, natural gas $2.899 against $2.912 and copper $6.7155 against $6.6915. The named third-party figures were Brent $103.87, gold $4,424.90 and silver $67.15 — all three exact, gold and Brent to the cent and silver to a tenth of a cent, on a capture the prior edition itself described as the cleanest of the window. That is the eighth consecutive session in which a dated third party has beaten the board, and the second in which every named figure was exact. Two published directions invert: natural gas from -0.07% to +0.38% and copper from +0.81% to +0.45%. WTI cannot be restated because the contract rolled; the October settle the third parties named was $100.30 against a published $99.53, and that figure is used below for the October arithmetic. Four rows are forming and the volume test is screaming on a Monday, exactly as it did a week ago. WTI at 0.04% of the prior session's volume, silver at 0.07%, natural gas at 0.03% and gold at 0.16% are intraday electronic prints, not settles. Three of the four corroborate closely against TradingEconomics spot taken on the same capture — natural gas 2.8258 against the board's 2.828, a fifth of a cent; gasoline 3.4477 against 3.4455; heating oil 4.8711 against 4.8680 — and Investing.com's own crude quote page carries a previous close of 91.96 against the board's 91.82, fourteen cents. Gold is the exception and three independent vendors are lower than the board: WSJ $4,381.00 at 17:15 ET, Investrade $4,386.00 and Bloomberg $4,385.20, against the board's $4,387.37. The gaps are $6.37, $1.37 and $2.17, far tighter than the $9.00 and $15.94 of the last two sessions, and on the record of this window the $4,381-$4,386 band is the better estimate of where the row finalises. The board basis is published for continuity across ten editions and no basis switch is made. The crude roll is the thing to get right, and it changes two other numbers. The board's WTI is now the November contract at $91.82; the October contract, which the prior edition published and which is the leg in this desk's Brent-WTI position, closed at $95.49 on WSJ's board and 95.345 on TradingEconomics spot, fifteen cents apart. That puts October-November backwardation at $3.67, against $4.22 on Friday's finalised numbers — so the prompt-month premium is compressing as the geopolitical scare unwinds, which is what should happen and is a cleaner read of the story than the flat price. Brent-WTI on the original Brent-November-against-WTI-October basis is $4.53, widened 96 cents from a restated $3.57; on a like-for-like November-against-November basis it is $8.20. Both are published because the position in Section 12 is on the first and the board is on the second. The crack spreads, computed on a consistent October-product-against-October-crude basis using the $95.49 October close:
The distillate crack has now given back $4.52 of the $5.42 it gained on Friday, and gasoline has taken the other side twice running. Heating oil fell 3.75%, the largest product decline on the board and steeper than the October crude leg's own move, while RBOB fell only 2.33% and remains +3.94% on the week against crude at -5.95%. A product that outperforms its feedstock by nearly ten points over five sessions while the other product underperforms it is a refinery-configuration event, and Marathon Petroleum -5.30% and Valero -4.84% — the two worst energy names on the session — are what that looks like in the equity market when the barrel falls faster than the margin can widen. Copper was the only thing on this board that went up, and it went up on volume. Copper closed $6.7920, up 1.50%, on 113% of the prior session's volume — a completed settle, not a forming row — and it is +6.11% on the week and +18.21% on the year on the spot basis. Set that beside gold at +0.75% year to date and crude at +66.07%, and the industrial metal is doing something neither of the others is: rising on the day the growth-sensitive commodity collapsed four per cent and the monetary metal fell nearly one. The gold-silver ratio narrowed to 65.84 from a restated 65.90, a fifth consecutive narrowing and the lowest of the window, but both metals fell — so the ratio is compressing through the numerator rather than through a silver bid, which is a weaker signal than the four sessions before it.
Desk-style ideas for institutional investors. Each carries an explicit expression, catalyst and invalidation. These are not personalized investment advice; verify independently and size to your own mandate before acting. 1. The rates trade — long ZQZ6 against short ZQZ7 widened again on a session the front end did not move at all; hold Mark first. Long ZQZ6 (December 2026) against short ZQZ7 (December 2027), DV01-matched one-for-one at $41.67 per basis point per contract, entered on 11 September at 95.910 / 95.450 for a spread of 46.0 bp, quarter size. Monday's mark: ZQZ6 95.835 unchanged, ZQZ7 95.345 from 95.350 — a spread of 49.0 bp. That is +0.5 bp on the session, worth +$20.84 per contract pair, and it takes the position to +3.0 bp, or +$125.01, from entry. The reading. This is the quietest session the trade has had and it still gained, which is the point of the structure. The front five contracts did not move at all — October, December, January, March and April all unchanged to three decimals — while every contract from June 2027 out cheapened half a basis point, and the par 2-year was likewise unchanged while the 10-year and 30-year richened five. The market added a fraction of tightening to 2027 on a day it added none to 2026, on no Fed news except Musalem arguing for more increases, which is precisely the political-and-terminal risk a 2027 contract prices and a 2026 contract cannot. The implied terminal rate rose to 4.685% from 4.680%. Catalyst: the 2-year auction Tuesday 13:00 ET against a 4.204% prior stop; the flash PMIs Wednesday 09:45; claims Thursday 08:30 against a 196K prior and a 203K consensus; core PCE 30 September. Invalidation, unchanged: the spread through 40.0 bp; or December 2026's probability of no further hike above 20%, against 10.0% today; or the 2027 modal range back at 4.25%-4.50% or lower at five or more of the eight meetings, against one today. Sizing: a quarter, at $41.67 per basis point per pair. Mark to date: +3.0 bp. 2. The belly butterfly — long the five-year against the two-year and the thirty-year; a basis point back Mark. A DV01-weighted butterfly: receive the 5-year against paying the 2-year and the 30-year in equal DV01 halves, entered on 17 September at the official par close of 2 x 4.78% - (4.67% + 5.29%) = -40 bp, quarter size. Monday: 2 x 4.83% - (4.76% + 5.29%) = -39 bp. The position gains as the fly goes more negative, so this is +1.0 bp on the session and -1.0 bp from entry. The reading. Friday's damage came from the near wing cheapening nine basis points; Monday reversed a basis point of it through the belly, the 5-year richening 3 bp against an unchanged 2-year and a 30-year 5 bp richer. So the far wing is now the problem and the near wing is inert, which is the mirror of day one. The thesis survives intact — the belly is the only sector that benefits from both halves of "hike sooner, stop lower" — but the supply schedule is about to test it directly: the 2-year prices Tuesday, the 5-year Wednesday and the 7-year Thursday, so the belly's own concession arrives one session after the near wing's. Catalyst: the three auctions; the flash PMIs Wednesday; ten committee appearances between Tuesday and Friday. Invalidation, unchanged: the fly through -30 bp; or CME's October probability back below 45%, against 56.5% today; or two consecutive coupon auctions tailing more than five basis points. Sizing: a quarter, DV01-matched two-for-one. Mark to date: -1.0 bp. 3. Long the power and electrical tier against short the artificial-intelligence security complex — the worst session the pair has had Mark. Long an equal-weight basket of GE Vernova, Eaton, Constellation Energy, Vistra and Quanta Services against short an equal-weight basket of CrowdStrike, Palo Alto Networks and Fortinet, dollar-neutral, quarter size, entered at the 14 September closes. Monday: the long basket averaged +1.40% — Constellation +2.91%, Eaton +2.51%, Quanta +0.93%, GE Vernova +0.57%, Vistra +0.08% — against a short basket averaging +3.45%: CrowdStrike +4.92%, Fortinet +3.17%, Palo Alto +2.25%. The pair lost 2.05 points, taking it from +3.39 to +1.34 points. The reading, and it is the risk the last note named. Friday's gain came entirely from three security names falling two to three per cent on valuation with no news; Monday they went back up on no news either, and faster. A short leg that moves three and a half per cent in a day on nothing is a momentum exposure, not a valuation one, and this pair is short momentum in the hottest part of the tape — SOX rose 4.29% and the whole software-security cohort travels with it. The long leg did what it was supposed to: it rose 1.40% with the independent power producers finally participating rather than fighting the curve, which is the first session the two tiers inside it have moved together. Action: hold the quarter and narrow the short leg to CrowdStrike and Palo Alto, dropping Fortinet, whose beta to the semiconductor tape is the highest of the three. Catalyst: the flash PMIs Wednesday; Micron 30 September after the close; any hyperscaler capital-expenditure confirmation. Invalidation, unchanged: the spread widening a further 8 points from entry against the position; or any credible report of a deferred or cancelled data-centre programme at a named operator. Sizing: a quarter, dollar-neutral. Mark to date: +1.34 points. 4. Long the 20-year against the 30-year — flat for a third session Mark. Entered on 15 September at the official par close of 20-year 5.40% against 30-year 5.36%, a spread of -4 bp, DV01-matched, quarter size. Monday: 20-year 5.33%, 30-year 5.29% — a spread of -4 bp. Mark to date: 0.0 bp. The reading. Both tenors richened exactly five basis points, so the spread has now held -4 bp through a selloff and a rally in consecutive sessions and is 1 bp wider on the week. The evidence for the liquidity-discount thesis is getting weaker in a specific way: a discount being absorbed should close on rallies, and this one did not move on a five-basis-point rally with no supply in either tenor. This is the fourth session with no progress and the position is on notice; the auctions are the last scheduled test before the review. Catalyst: the 5-year Wednesday and the 7-year Thursday, the first supply into a long end that has richened five basis points without help; the Fed balance sheet print Thursday 16:30 ET. Invalidation, unchanged: 20s30s through -8 bp; or a second consecutive coupon auction tailing through the afternoon fix; or no movement by Friday's close, at which point the position is closed for lack of thesis. Sizing: a quarter, DV01-matched. Mark to date: 0.0 bp. 5. Long Brent against WTI — $1.61 in profit, and the October contract expires tomorrow, which forces the mark-out forward Mark. Entered on 15 September at settles of $108.52 Brent November against $105.49 WTI October, a differential of $3.03 on the published basis and $2.92 on the finalised one. Monday, on the same contract pairing: Brent November $100.02 against WTI October $95.49, a differential of $4.53. The position is +$1.61 from the restated entry, per barrel-for-barrel pair, after +$0.74 on Friday. The reading, and the instrument now decides the timing. The differential widened 96 cents in a session on a four per cent decline in both legs, which is the physical story asserting itself in the right direction: Brent fell 3.71% against an October WTI that fell roughly 4.8% on WSJ's basis. But the written instruction to mark out on 25 September is no longer executable, because the October WTI contract expires on 22 September and Investing.com's own board has already rolled to November. That is not an ambiguous clause and it is not discretion — the leg ceases to exist. Action: hold through Tuesday and mark the position out at the October settlement on 22 September, three sessions earlier than written, and do not roll into the November-against-November spread, which printed $8.20 on Monday and is a different trade with a different carry. The reason for the difference is visible: October-November backwardation compressed to $3.67 from $4.22, so the prompt-month premium this position was partly harvesting is itself unwinding. Catalyst: the expiry itself on 9/22; the EIA petroleum status report Wednesday 10:30. Invalidation: superseded by the forced mark-out above; the $2.75 stop was never triggered. Sizing: a quarter, barrel for barrel. Mark to date: +$1.61. 6. Long the equal-weighted index against the capitalisation-weighted index — a point against on day one Mark. Long RSP against short SPY, dollar-neutral, entered at Friday's closes, quarter size. Monday: RSP $212.68, +0.56% against SPY $773.50, +1.55%. The pair lost 0.99 points. Mark to date: -0.99 points. The reading, and it is an honest first-day loss rather than a broken thesis. The trade was entered on a session where 347 of 494 lines fell against 145 and the index still rose; Monday the decliner ratio normalised completely — 276 up against 215 down, a 1.28-to-one advancer ratio — and the pair lost anyway, because breadth normalising is not the same as the median member keeping pace. Seven of eleven Finviz groups were green, which is the participation the thesis asked for, and the capitalisation premium survived it at 0.99 points because four names did most of the index's work. The written invalidation asks for the decliner ratio to normalise below 1.5-to-one within three sessions, and it did so in one, so that clause has been satisfied in the position's favour and the loss came from the other leg of the thesis. Action: hold the quarter; the second clause is the live one. Catalyst: the flash PMIs Wednesday 09:45, the first broad read that touches the median company; six defensive S&P 500 reporters between Tuesday and Thursday; quarter-end rebalancing on 30 September. Invalidation, revised: the pair 3 points against entry, against -0.99 today; or any single megacap earnings event entering the window, which would re-concentrate the index and is the specific risk this trade is short; the breadth clause is retired as satisfied. Sizing: a quarter, dollar-neutral. Mark to date: -0.99 points. 7. New — long Paramount Skydance against short Warner Bros. Discovery The expression. Long PSKY against short WBD, dollar-neutral, entered at Monday's closes of $9.91 and $30.80, quarter size. The position gains when the acquirer closes the gap to the target. The thesis. The Wall Street Journal reported advanced settlement talks between Paramount and California's attorney general that would remove a named regulatory obstacle to the $110bn combination, with concessions running to $1.5bn of in-state production investment, no sale of the studio lots, penalties against a thirty-films-a-year pledge and a possible cable divestment. Warner Bros. Discovery rose 10.79% and Paramount Skydance fell 2.94% — on a session the index rose 1.49% and their shared Finviz group rose 3.43%, so the acquirer underperformed its own sector by more than six points on news that makes its own deal more likely to complete. Deal risk falling should compress the spread from both ends. What the market did instead was price the concessions as a pure cost to the buyer and the $7m-a-day delay fee as a liability Paramount now certainly capitalises rather than possibly escapes. Both readings can be right and still leave the acquirer too cheap, because a settlement that closes the deal is also the event that ends the fee. Catalyst: confirmation or denial of the settlement; the 30 September fee threshold; any regulatory filing naming the concession package. Invalidation: the pair 6 points against entry; or a credible report that the California talks have collapsed; or any second-state or federal action that reopens regulatory risk, which would justify the market's split and invalidate the premise. Sizing: a quarter, dollar-neutral, and smaller than the book's other pairs because a single headline can gap both legs. Mark to date: new. Closed positions, marked forward The long distillate crack, closed 14 September at a restated +$0.44, is vindicated again. The differential printed $59.75 on Monday against the $66.79 at which it was entered on 9 September, so holding would now be -$7.04 from entry and $7.48 worse than the booked close. The avoided loss has grown for a fourth consecutive session. Protection on the CCC cohort funded in IG, closed on 16 September at +43 bp, would have gained again. The CCC-minus-HY differential widened 9 bp to 815 bp on the 18 September stamp, so the cumulative had it been held rises to +49 bp against the +43 bp booked. That is the first session in which the closed position would have been ahead of the booked exit, and it is recorded as such. The credit-bureau pair, closed at -1.72 points, went the wrong way again. Fair Isaac fell 2.65% and Equifax 0.16% against Finviz financials at +0.52%, so the pair gained 1.93 points in the session after it was closed, taking the cumulative had it been held to +5.03 points. That is now clearly the worst close on this page. TransUnion did not appear in the 494-line component capture for an eighth consecutive session and no mark is asserted on it. Long October volatility on the semiconductor complex, closed on 17 September at roughly -9.7% on the index, would have recovered part of the loss: SOX rose 4.29% to 12,433.17, its largest single-session gain of the window, which is the first move in the position's favour since it was closed. The index leg would now be -11.2% from entry against the -9.7% booked, so the close is still ahead but by less. The short-debasement basket against long dollar, closed on 3 September, would have gained for the first time in three sessions: gold -0.85% and silver -0.76% against a dollar index at +0.20%. The vol note VIX closed 14.87, up 0.06 points or 0.41%, on a range of 14.60 to 15.13, on the best index session in more than a month. The five-observation path is 17.20, 17.71, 15.44, 14.81, 14.87, so the index has stopped falling one session after making the window's low. A 14.87 handle asks for roughly a 0.94% daily move against realised index moves of 1.14%, 0.16% and 1.49% over the last three sessions, an average absolute 0.93% — so index volatility is now essentially fair to realised at 1.01-to-one, against 1.61-to-one on Friday and 1.43-to-one on Thursday. That is a two-session collapse in the risk premium and it happened through the denominator: realised volatility rose 60% in three sessions while the implied level did not move. The case for owning convexity is therefore better than it was, not worse. Dispersion widened rather than collapsed: the 494-name distribution ran from Akamai +12.33% to Mohawk -7.00%, a 19.3-point spread on a 1.49% index day, against 15.4 points on Friday's 0.16% day. Rate volatility is rising too — the MOVE card's 18 September bar shows +5.80% — and the two have now moved together in the same direction for the first time in the window. And the calendar is asymmetric: no Very-high release lands before Friday's Michigan inflation expectation, but three coupon auctions land in three consecutive sessions and a Trump-Xi summit sits on Thursday with no scheduled time. Buying a 14.87 handle against 0.93% realised, into an unscheduled-risk week, is a better trade than it was at 14.81 against 0.58%.
Crowded consensuses worth stress-testing with numbers.
The two-sided geopolitical tape. The de-escalation side carried the session: Hormuz transit at a six-month high, credible United States and Iran diplomatic signals, and an accelerating Saudi pipeline restoration timetable took the November crude contract down 4.43% and October's down roughly 4.8%, with October-November backwardation compressing to $3.67 from $4.22. The escalation side is entirely unscheduled and sits on Thursday: a Trump-Xi summit with no published time, covering trade, tariffs, Taiwan and artificial intelligence, inside a week in which the American semiconductor complex rose four per cent on Korean and Taiwanese leadership and KKR committed to the Korean artificial-intelligence supply chain after a record $3bn of deals. A tape that is long North Asian silicon into an unscheduled bilateral on Taiwan is carrying a risk it has not been asked to price. Structural watch items. Turkey is three weeks into a fund-redemption episode and USD/TRY has moved 0.11%, 0.04%, 0.04% and 0.09% across four sessions while the BIST 100 has swung four per cent peak to trough — a currency that is administered rather than discovered, and the exit is still closed. The French long end recovered 4.2 bp of Friday's 8.5 bp widening with no domestic event in either direction, so the OAT-Bund spread at 101.3 bp is a political premium trading on sentiment rather than news. Reserve balances have not printed a new observation in two sessions and sit at $3.0138tn into quarter-end. And TransUnion has been absent from the 494-line component capture for eight consecutive sessions, which is now a vendor problem worth naming rather than a data point.
Companion files for this edition: US_CrossAsset_Daily_2026-09-21.md (canonical report of record, including the Overnight / Asia & Europe read-through, the Source Links appendix and the Data Notes & Conflicts appendix) and US_CrossAsset_Daily_2026-09-21_DataNotes.txt. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| U.S. Stock, Fixed Income & Cross-Asset Closing Daily · Monday, September 21, 2026. Prepared from public vendor data; figures are as captured and may be revised by the source. Not personalized investment advice. |