Hawkish scores at a five-year high into the blackout; the long end's best week in a month comes from AI-credit fear, not disinflation

0. Weekly Arc

The week ran from a data shock into an institutional silence, and the argument migrated from timing to terminal level. A hot September CPI on Oct 14–15 pushed October hike odds toward a coin flip; officials then declined to validate it, and Chair Warsh’s refusal to guide collapsed October odds below 15% by Oct 17. Oct 18–19 produced the sequence’s oddity — hot prints, lower long yields, crowded duration shorts. Oct 20 opened the blackout with the December consensus intact and the Treasury buyer base, not the Fed, setting the long end; Oct 21 hardened December into a near-unanimous survey fact and split the sell side on where the cycle stops. Today the silence holds, the hawkish tone scores at a five-year high, and the long end rallies on AI-credit anxiety rather than on cooling prices.

1. Policy Narrative & Expectations

The net change over the past ~24h is that the pricing barely moved while the documentation of the hawkish consensus became systematic. CME FedWatch puts October at an 83.9% hold against a 16.1% cumulative 25bp, and December at 9.9% hold / 76.0% cumulative 25bp / 14.2% cumulative 50bp ^[1]; a parallel tracker has October at 15.1% (−1.8pp) and December at 85.4% (−4.4pp) ^[2], and BMO reads December at roughly 90% ^[3]. Nomura’s decomposition shows the market pricing 4bp at the October meeting, 25bp in December and a cumulative 55bp by March, with the two-year forward OIS terminal proxy back up to 4.57% ^[4]. Against that, the marginal new information is a communication audit rather than a decision: Deutsche Bank’s AI-based scoring of Fed communication ahead of the blackout puts the average hawkish score for all officials at 7.5, up 0.2 from September and above the pandemic-era peak of 7.4, with the 2026 voting members also at 7.5 and 18 of 19 officials above the neutral 5.0 threshold ^[5]. Goldman Sachs’ synthesis reaches the same place from the transcripts — nearly all participants have said since the September FOMC that further tightening may be necessary, with most judging more than one hike needed, and the rhythm read as every-other-meeting, i.e. skip October and hike in December, a plan Goldman says last week’s Warsh and Daly remarks kept valid after the stronger September CPI ^[6]. The unresolved part is urgency, not direction: DB notes even the most hawkish officials, Hammack and Kashkari, showed reduced urgency for immediate action ^[5], while the range of magnitude calls inside the committee runs from Williams’ “one more may be appropriate” to Logan’s 50bp-or-more ^[6].

1.1 FOMC Officials’ Remarks

• [ONGOING] No public FOMC remarks in the past 24h: the Committee is in its pre-meeting communication blackout, so official policy signals are temporarily absent ^[7]. The most recent remarks reach this batch only through secondary syntheses and are captured in §1.2.

1.2 Policy Signals & Institutional Communication

• [NEW] Deutsche Bank’s communication scorecard puts the hawkish consensus at a five-year high: ahead of the blackout, the all-official average rose 0.2 to 7.5 and the 2026-voter average from 7.3 to 7.5, both above the pandemic peak of 7.4, with 18 of 19 officials above the neutral 5.0 threshold ^[5]. Dallas Fed President Logan scored highest at 9.5, up 1.6, on viewing policy as “not there yet” and needing “a few more hikes” to retract last year’s insurance cuts; Jefferson rose about a point to 7.7 and Waller to 7.4, with Waller’s latest speech directly pointing to further hikes and detailing “signal options” totalling 75bp of tightening; Warsh stood at 8.1, his October IMF remarks scoring about 7.5 against 8.6 at Jackson Hole; Hammack fell 1.3 to 8.3 while emphasising labour-market tightening risk, and Kashkari slipped 0.2 to 8.8 ^[5]. Bowman, at 3.6, was the only official below neutral, and Powell held at 5.2 because he has made no public speech since May ^[5].
• [NEW] The same scorecard flags a nuance inside the consensus: 13 officials moved by less than half a point — including Warsh, Bowman, Williams at 6.1, Cook at 8.4, Barr at 8.6 and Paulson at 7.9 — which DB reads as broad agreement on continued tightening at a measured pace, and it maintains its baseline of a pause in October followed by hikes in December and next March ^[5].
• [NEW] Goldman’s remarks synthesis fixes the reaction function at every-other-meeting: nearly all participants have indicated since September that further tightening may be necessary, most think more than one hike is needed, but the pace is not urgent, pointing to a skip in October and a hike in December ^[6]. The magnitude range is wide: Logan estimates the target range needs to rise 50bp or more; Waller expects additional hikes if data meet expectations but says timing is flexible and need not be consecutive; Williams says he has not changed his view that one more hike may be appropriate; Hammack says inflation pressure is relatively broad and not only supply-side, with AI buildout demand pushing up computers, electronics and high-tech component prices; Jefferson sees no spillover yet from tariffs, energy shocks and commodity spikes into broader persistent inflation; Barr notes only 2 of the past 20 months were consistent with 2% core PCE, while Paulson puts underlying inflation at 2.5%–3% and Williams notes underlying indicators have already declined ^[6]. Goldman itself cautions that the Williams and Jefferson remarks predate last week’s CPI ^[6].
• [NEW] BMO prices the tape, not the path: Treasury yields may be coiling for a breakout after weeks of narrow-range trading, with the Fed’s Oct 28 decision the next catalyst — the Bollinger Band spread has compressed to 24bp from as much as 80bp on Sept 28, the 2-year has closed between 4.70% and 4.82% in every session since Oct 1, and the average daily range of 10bp is the highest monthly average since March ^[3]. Its base case is a hold in October and 25bp in December, but it warns markets may be placing too much confidence in a year-end hike with December priced near 90% ^[3].
• [NEW] DWS argues for a hawkish hold rather than a hike: chief US economist Christian Scherrmann expects the Fed to stay unchanged in October while keeping a hawkish bias, since high inflation still leaves the market uneasy but weakening labour-market momentum and stable core PCE expectations both support waiting, making a hawkish hold more likely than a direct hike ^[8].
• [NEW] A deliberately hawkish outlier: Prometheus Research published more than 30 pages of fundamental and systematic work concluding that risks remain skewed toward the Fed needing to raise rates materially more than is currently priced (single source / unverified) ^[9].
• [NEW] Nomura supplies the framework for how a credit shock interacts with the hiking cycle: the current credit worries around US tech and France resemble a historical “second wave” shock that central banks can contain by easing, but escalation into a “third wave” crisis and recession depends on why oil is rising — fundamentals-driven oil forces continued hikes and blocks easing, while geopolitics-driven oil preserves policy flexibility; with US and European central banks still hiking on high oil prices, the escalation risk is live ^[4].
• [NEW] A former market participant’s framing of the September move: per Reuters commentary from Federated Hermes’ Deborah Cunningham, the FOMC “had little choice last month but to lift the band to 3.75-4.00%”, Chair Warsh delivered an increase many did not expect from him and “did what the data demanded”, and the Committee may have felt pressure from the White House, the bond market and price-conscious consumers to hold ^[10].

2. Key Data & Market Read

• [NEW] Activity tracking stayed firm: BofA held its Q3 US GDP tracking estimate unchanged at a firm annualized pace above its own official forecast, with personal consumption and equipment investment the supports and net exports the drag; the only entry on the day’s calendar was the weekly MBA mortgage applications print ^[7]. The US economic surprise index also firmed, with both the 20-day and 5-day averages edging higher ^[2].
• [ONGOING] September CPI: the strong print did not derail the officials’ every-other-meeting hiking plan ^[6].
• [ESCALATED] Housing remains the cleanest transmission casualty: pending home sales fell sharply, homebuilder confidence dropped to a 14-year low on insufficient demand, and September single-family permits declined, with Citi expecting further permit declines as builders struggle to absorb existing supply against 30-year mortgage rates near 7.5% ^[11].
• [ONGOING] Narrative impact: the growth leg is still being marked up by tracking models while housing prints cycle lows — the divergence the briefings have flagged for a week — and with the Committee silent, no official can arbitrate whether that divergence is resilience or the first visible lag effect of a higher policy path ^[7][11].

3. Financial-Conditions Signals

• [ESCALATED] Dollar & rates — the haven bid has taken the long end to a one-month low: Treasuries climbed across the curve, pushing long-end yields to the lowest in a month with yields down at least three basis points, as renewed doubts about the durability of the artificial-intelligence trade drove investors toward havens and Asian equities retreated ^[12]. Nomura records a twist-flattener with long bonds attracting safe-haven inflows ^[4], and BMO notes haven demand keeping the 10-year near the lower end of its October range ^[3]. Volatility eased alongside, with MOVE at 95.9 and VIX at 16.7 ^[4].
• [NEW] The driver decomposition has flipped from real yields to breakevens: the 10-year real yield was flat at 2.75% while the 10-year breakeven rose to 2.37%, i.e. inflation compensation rather than real rates is now doing the marginal work on nominal yields ^[4] — a rotation away from the real-yield-led adjustment that dominated the prior week.
• [ESCALATED] Credit — the AI complex is repricing from the weakest link outward: Citi argues the consequences of a hawkish policy path are already visible, with rising funding costs lowering corporate valuations and widening spreads for corporate and some sovereign borrowers, and AI infrastructure especially sensitive because it is now funded through credit markets rather than free cash flow ^[11]. JPMorgan records the high-yield AI basket widening 147bp over the past month to 478bp, 97bp wider than July’s widest level, CoreWeave’s senior unsecured bonds repriced to roughly a 15% yield with about 25% of market value lost in a week, and early stress appearing in the lowest-quality slice of US consumer credit, especially subprime ^[13]. Nomura judges the tech and French credit concerns less severe than historical crises, while noting the widening France-Germany spread is pushing flows into long duration ^[4].
• [NEW] Liquidity — the money-fund bid is being rerouted, not withdrawn: US money market funds cut Treasury bill holdings by $276bn month-on-month in September, the largest monthly decline since June 2025, with government funds down $182bn and prime funds $95bn, and rotated defensively into floating-rate notes and repo while shortening weighted average maturity ^[14]. JPMorgan expects incremental bill demand to come mainly from reduced repo allocations, modestly tightening repo conditions with SOFR averaging near IORB, notes SOFR volumes below $3trn since July as a possible sign of lower hedge-fund leverage, and reports quarter-end FICC-cleared repo up $107bn against dealer repo down $97bn ^[14]. Total money-fund assets fell $45.45bn to $7.89trn in the week to Sept 30, and a Fed hike takes roughly three to four weeks to reach a fund’s yield because portfolios are laddered ^[10].
• [ESCALATED] Fiscal arithmetic is now a quantified feedback loop: BofA puts the 10-year about 100bp above CBO’s February baseline — between CBO’s +50bp and +150bp scenarios — and argues that at current levels the effective interest rate on federal debt breaches 4% within a decade, above CBO’s long-run nominal growth, with rising interest costs and larger deficits themselves supporting long-end yields ^[7]. In JPMorgan’s investor survey, 56% expect the 10-year above 5.0% by end-2027 and 60% name fiscal sustainability the biggest upside risk to long yields, against US, German and Japanese 10-year yields at their highest since 2002, 2009 and 1996 respectively ^[13].
• [NEW] Swap spreads and curve structure: swap spreads opened slightly tighter in a lag to the Treasury pullback, with light investment-grade supply having a bigger impact than the Treasury move and unwinding of deal-tied long positions dominating thin overnight action as expectations for a large AI-related issuance wave turned more cautious; the swap spread curve is tilted flatter on higher money-market rates and a dip in SOFR futures (single source / unverified) ^[15][16].
• [NEW] Financial conditions and the dollar are flat at the margin: the US financial conditions index was essentially unchanged on both the 20-day and 5-day averages and the dollar index held at 101.7 ^[2].
• [NEW] The diversification caveat inside the income trade: Charles Schwab notes the 10-year above 5% — near a two-decade-plus high — improves the starting yield on new bond purchases, but a scatter of aggressive income sectors shows most clustered in the upper-right quadrant, highly correlated with each other and more correlated with US equities than core bonds, so diversification in stress may be weaker than expected ^[17].

4. Global Central-Bank Linkages

• [NEW] ECB — no October hike advocated, with financial conditions doing the work: Goldman’s read of Governing Council remarks after the October meeting finds no policymaker advocating a hike, most keeping a balanced tone that guides to possible further tightening while stressing a “measured” approach, with Nagel saying “the work is not done” and “a bit more may still be needed”, Lane calling the long-rate rise “a material tightening of euro-area financial conditions”, and Moulin and Rehn arguing that higher long rates may themselves substitute for rate action ^[18]. Goldman’s comparison of three euro-area FCIs supports that substitution: its MF-FCI has tightened 0.5 standard deviations since summer on rate-driven weights, the ECB staff’s ECB-BIG is essentially flat with credit volumes holding up, and Goldman’s own GSFCI sits in between and leads ECB-BIG, implying credit quantities may tighten further, with a VAR pointing to a 0.1%–0.5% hit to euro-area real GDP by Q4 2027 and a peak quarterly growth drag of up to 15bp in Q1 2027, most pronounced in France ^[19].
• [ESCALATED] BOJ — December is priced, and the argument is that it is underpriced: market pricing assigns roughly 80% probability to a December hike ^[4], JPMorgan argues a weak yen is eroding purchasing power and requires faster normalisation ^[13], and BofA’s Taylor-rule work says BOJ hikes are underpriced by about 66bp, with the Bank’s own forecasts implying a 2.4% policy rate against 1.8% priced ^[20].
• [NEW] BOE — the mirror image, overpriced: BofA’s forward-looking Taylor rule shows markets overpricing BOE hikes by roughly 120bp, with the policy rate priced near 4.6% for mid-2027 against a rule-implied 3.4%, making the UK the only G10 economy whose policy rate sits above the rule level; Chief Economist Huw Pill has said the four hikes priced are too many and BofA expects two ^[20].
• [ONGOING] PBoC — the anchor switch is plumbing, not a cut: the Financial Street Forum set three arrangements — moving the main policy rate from 7-day OMO to overnight reverse repo, diversifying loan pricing anchors to DR001 or LPR, and retiring the old deposit and lending benchmarks — with the new anchor about 15bp below the old one and 7-day OMO and LPR unchanged; DR-benchmarked loans have been piloted since July 2026, basis risk between the DR001 asset anchor and average-deposit liability anchor is rising, and one house expects 1-year AAA NCD yields to fall to 1.35%–1.4% ^[21]. The central bank drained a net RMB 1.25trn via pledged reverse repos in October to date while injecting a net RMB 400bn via outright reverse repos, and Governor Pan Gongsheng confirmed DR001 as the operational target ^[22].
• [NEW] BOK: the Bank of Korea said it will determine the timing and speed of additional base-rate hikes while assessing inflation trends, the domestic economy and financial stability, and that it is essential to evaluate changes in domestic and external risks and the impact of its two previous increases (single source / unverified) ^[23][24].
• [NEW] BofA’s G10 Taylor-rule map puts the Fed at the dovish extreme: on an output-gap rule the Fed is the most accommodative central bank in the G10, with the dollar’s policy-stance z-score at −1.67 against the New Zealand dollar at −1.22, the Australian dollar at −0.80, sterling at +0.29, the Swedish krona at +0.42 and the Canadian dollar at +0.48 ^[20]. The clearest FX signals are anti-carry — short AUD/NZD and short USD/JPY, with short NOK/SEK secondary — set against a volatility-adjusted carry backdrop BofA calls among the strongest since 2000, with the Bloomberg FX cumulative carry index up 15.3% over 52 weeks (92nd percentile) on FX volatility of 6.89 (10th percentile) ^[20]. JPMorgan, by contrast, sees ECB hikes as overpriced at two and keeps a medium-term bullish dollar view on Fed policy and supportive rate differentials ^[13].

5. Asset Implications

Quadrant	Current probability tilt	Key asset implication	Anchoring narrative
Growth↑ + Inflation↑	Rising at the margin	Commodities↑ TIPS↑ nominal long bonds↓ — the communication audit puts the hawkish consensus at a five-year high with 18 of 19 officials above neutral, Waller’s “signal options” detail 75bp, Logan wants 50bp or more, the every-other-meeting plan survives a strong CPI, and activity tracking plus the surprise index are firming	§1.2 / §2
Growth↑ + Inflation↓	Falling	Stocks↑ long bonds↑ gold↓ — the only support is DWS’s hawkish-hold call, which rests on weakening labour momentum and stable core PCE expectations; there is no fresh disinflation data and the breakeven is rising, not falling	§1.2 / §3
Growth↓ + Inflation↑	Rising	Commodities↑ gold↑ stocks↓ — high-yield AI spreads at 478bp, CoreWeave near a 15% yield, early subprime consumer stress, housing at cycle lows with mortgages near 7.5%, and a fiscal r>g feedback loop quantified at roughly 100bp above the CBO baseline	§3 / §2
Growth↓ + Inflation↓	Rising modestly	Long bonds↑↑ stocks↓ credit spreads↑ — the long end reached a one-month low and MOVE and VIX both eased, but the rally’s stated cause is doubt about the AI trade rather than cooling activity or inflation	§3 / §1.2

Stock-bond correlation call: today’s negative print is the same species as the previous three — haven demand, not disinflation — but the internal decomposition has changed in a way that matters for regime classification. The prior week’s rally ran through real yields, which meant TIPS and nominals lost or gained together and the inflation hedge was absent. Today the 10-year real yield is flat at 2.75% while the breakeven has risen to 2.37%, so inflation compensation is now the leg doing the marginal work ^[4]. That is a mechanically friendlier configuration for a risk-parity book in one narrow sense — TIPS are the relative winner inside duration rather than a co-victim — but it is also the classic inflation-driven positive-correlation signature, arriving on a day when the headline print was negative because of AI-credit fear. The honest read: the base case remains an inflation-, supply- and term-premium-driven positive regime, and the negative episode is a growth-scare window whose trigger is the funding cost of the AI complex. Two structural supports for the positive reading strengthened today rather than weakened — the fiscal feedback loop is now quantified at the level where r exceeds g within a decade if yields hold ^[7], and 60% of surveyed investors already name fiscal sustainability the biggest upside risk to long yields ^[13]. Against that, the credit channel is now fully one-directional: spreads widen on rising funding costs, which lowers equity valuations, which is the same variable hitting both sleeves ^[11].

Risk-budget implication: the bond sleeve should not capitalise this rally into a larger long-end risk budget, because its own stated fuel is AI-credit anxiety and the covering of a crowded duration short rather than any cooling in prices or activity. Express duration in the belly and the curve rather than as a long-end level bet, and shift the composition of the sleeve toward TIPS: with the real yield flat and the breakeven rising, inflation compensation is the part of the curve being repriced, so owning the breakeven is cleaner than owning the nominal level. Credit should be run on quality and dispersion, not index beta: the AI-adjacent cohort is where the marginal widening is concentrated — a 478bp high-yield AI basket and a compute provider’s unsecured debt near 15% — while the senior part of the capital structure and front-end financials remain the defensive expression, and the low-quality consumer loan segment is the next place stress will surface. Equities deserve the smallest tactical risk budget relative to their volatility, and the reason is now published rather than inferred: aggressive income sectors are clustered in the upper-right of the risk/return scatter, correlated with each other and more with equities than with core bonds, so the portfolio’s apparent diversification is thinner than its sleeve weights imply. Gold earns an above-neutral weight as the reserve-diversification and dollar-hedge sleeve rather than a rates hedge: the gold-dollar correlation has strengthened to −81.2% in 2026, gold rose in 84 of the 91 dollar-decline episodes of at least 14 days since 2000 for a 92.3% hit rate and a 5.46% median interval return, and the same study shows 2-year and 10-year yields fell in only 68.1% and 64.8% of those episodes — i.e. the hedge works through the currency, not through duration. Energy stays at or above benchmark but booked explicitly as a directional inflation position, and the one scenario that breaks the whole construction is the one the same research names: a genuine dollar-liquidity scramble, in which the dollar strengthens and moves inversely to risk assets, rendering the dollar hedge useless.

6. Contrarian & Tail Risks

• Consensus fragility: the baseline is “October is a hold near 84%, December is somewhere between 76% and 90%, and the committee is uniformly hawkish.” Four assumptions are falsifiable. First, the December number is three different statistics wearing one label — 76.0% as a cumulative 25bp CME probability, 85.4% on a second tracker, and “roughly 90%” in a third house’s reading — and BMO explicitly warns that confidence at 90% leaves room for expectations to shift on fresh data. Second, the hawkish consensus is real in tone and contested in urgency: the same scoring exercise that puts the average at a five-year high notes that the two most hawkish officials, Hammack and Kashkari, both showed reduced urgency for immediate action, and that 13 of 19 officials moved by less than half a point. Third, the magnitude distribution is unbounded on the upside — Waller’s detailed 75bp of “signal options” and Logan’s 50bp-or-more sit far above the market’s 55bp cumulative to March, while DWS argues the next move is a hold at all and Prometheus argues materially more than priced is required; those cannot all be hedged with the same duration position. Fourth, the whole hawkish read rests on remarks delivered before the last CPI — Goldman says so explicitly about Williams and Jefferson — so the committee’s stated tolerance for 2.5%–3% underlying inflation is being tested by data it has not yet commented on, and cannot comment on for another week.
• Second-order transmission: four chains are not in the price. The AI-financing chain is the most advanced and now has a term sheet: roughly $350bn of public AI-related issuance expected in 2026 against about $5.5trn of AI capex over five years, of which some $4.1trn is debt-funded, with that borrowing crowding out consumer credit and emerging-market issuance and eroding the old assumption that Asian external surpluses automatically recycle into Treasuries — a central-bank backstop that is weakening exactly as supply grows. The fiscal chain is the second and is now arithmetic rather than narrative: rates roughly 100bp above the CBO baseline push the effective rate on the debt stock above long-run nominal growth within a decade, and rising interest costs plus wider deficits feed back into the long end, a loop that no policy-rate decision closes. The non-bank leverage chain is the third: SOFR volumes below $3trn since July hint at lower hedge-fund leverage, non-bank activity and leverage are rising with volatility amplification concentrated outside the banking system, and an IMF stress test of an AI-productivity repricing plus higher rates still leaves bank CET1 near 11.5% — i.e. the banks are fine and the risk sits elsewhere. The funding-plumbing chain is the fourth: money funds have already cut bills by $276bn in a month, and JPMorgan’s expectation that future bill demand comes out of repo allocations means the marginal buyer of duration is being funded by squeezing the repo market rather than by new cash.
• Source quality control: volume is not confirmation. Several items are explicitly single-source and are carried as colour rather than reporting: the Prometheus conclusion that the Fed must hike materially more than priced, and the Bank of Korea’s guidance on the timing of further hikes, both arrive as one-line social relays. The swap-spread commentary arrives twice from the same macro-chart account and should be read as one observation, not two. Two blocks are same-house duplicates rather than independent corroboration — the BofA Securities morning note and the BofA Taylor-rule piece come from the same franchise, and the JPMorgan money-market update and the JPMorgan investor survey are one franchise’s two products. One large block is out of scope: the CEPR discussion paper on currency competition contains no FOMC decision, official remark, US data release, financial-conditions or cross-asset content and is excluded from the Fed narrative. Two further blocks are dated framework pieces from mid-September and are used for structure only, not as fresh signals. Direct conflicts deserve side-by-side reading rather than averaging: the October hike probability appears at 16.1% and 15.1% across two trackers; December appears at 76.0%, 85.4% and roughly 90%, which are a cumulative-move probability, a hike probability and a rounded reading of the same event; the growth read conflicts internally, with tracking estimates marked up while housing permits, pending sales and builder confidence all deteriorate; and the credit read conflicts across houses, with one desk calling the AI widening an overreaction unsupported by fundamentals earlier in the week while today’s evidence shows the basket 97bp beyond its July wides. Finally, several items are model-, survey- or score-derived — the AI communication scores, the three euro-area FCIs and their VAR output, the Taylor-rule gaps, the GDP tracking estimate and the equity-scatter quadrant work — and frame a distribution rather than a price.

---

Appendix: Additional Sources

• ^[25] CEPR — discussion paper on currency competition and self-fulfilling inflation beliefs; no Fed, US data or cross-asset content, excluded
• ^[26] GF Securities Strategy — dated 09/16 methodology deck on A-share liquidity; no Fed content, used for structure only
• ^[27] Zheshang Securities — dated 09/15 note on China credit weakness and narrowing domestic easing room; framework only
• ^[16] Deer Point Macro — duplicate relay of the swap-spread commentary in ^[15] (single source / unverified)
• ^[24] Financial Juice — BOK: timing and speed of additional hikes to depend on inflation, growth and financial stability (single source / unverified)
• ^[18] Goldman Sachs — ECB Chatterbox October edition: full roster of Governing Council remarks; only the summary lines are used above
• ^[19] Goldman Sachs — European Daily: FCI comparison and VAR output in full
• ^[20] BofA — “Taylor-made for FX”: G10 policy-stance z-scores and carry metrics

---

This report is a macro-mechanism analysis, not investment advice.