Fiat Elpis · High-conviction macro note 14
Four high-conviction macro trades for the new dollar regime
Short the crowded hedges where Fed credibility and dollar carry now bite. Buy earnings and policy compression where the levered book still has room to chase.
Fiat Elpis high-conviction cross-asset note · 30 August 2026
The regime shift is actionable. Renewed Fed credibility rewards dollar carry, while AI earnings and China policy support create upside with cleaner positioning. The result is a four-leg book—short gold, short EUR/USD, long Nasdaq-100 and long Hang Seng—with independent anchors rather than one generic risk-on bet.
01 · The information delta
The policy reset creates four clean cross-asset expressions
The market is repricing policy credibility and technological scarcity at the same time. The message from Jackson Hole was unambiguous: the 2% inflation target is fixed, broad financial conditions are not restrictive and short-term rates remain the predominant policy tool.1 The immediate transmission—two-year yields up 12.8bp, DXY up 0.61% and gold down 3.19%—shows where the pressure lands first.2
That regime produces four expressions: short gold, short EUR/USD, long Nasdaq-100 and long Hang Seng. Gold and the euro monetize renewed dollar carry. Nasdaq monetizes an AI earnings engine against exceptionally light levered exposure. Hang Seng monetizes policy-tail compression and a technology-led profit recovery.
The book is deliberately not one macro bet repeated four times. Each leg has a separate fundamental anchor, a separate positioning profile and a single observable condition that ends the thesis. Positioning support ranks Nasdaq first, gold second as unwind fuel, EUR/USD mixed and Hang Seng unproven. The fundamental case is strong across all four.
“Short the crowded hedges. Buy the earnings and policy compression.”
Fiat Elpis cross-asset view, 30 August 2026
02 · Signals to track
The four trades
The crowded long is ready to unwind
Speculative length and four straight $1bn-plus ETF-inflow weeks have loaded the same side of the boat just as higher real yields and a stronger dollar raise the cost of holding the metal.3–4
US carry meets Europe’s bad-options tree
The euro now carries a weaker energy-import bill, French fiscal fragmentation and a hawkish ECB path that is already largely priced. Another European hike risks becoming a growth-negative bad hike.6–8
Buy the AI earnings machine
AI demand is converting into revenue across compute, memory, packaging, networking, power and commissioning while the levered book remains unusually light. Earnings lead; positioning supplies the chase.9–11
Buy policy-tail compression
China’s property package lowers systemic downside while paid AI inference and electronics profits become measurable. This is a technology-profit and policy-risk trade—not a broad reflation call.12–16
The construction
This is not a generic risk-on book. Each leg monetizes a different part of the regime and owns a separate failure condition. Dollar carry pressures gold and the euro; AI earnings drive Nasdaq; Chinese policy and technology profits drive Hang Seng.03 · What may be mispriced
The market is pricing every leg from the wrong starting point
Gold is still being treated as an untouchable secular hedge even though the marginal holder is now crowded. The euro is still being rewarded for prospective ECB tightening even though most of that tightening is priced and the next hike can damage growth. Nasdaq is still being framed as a duration asset despite a live earnings and physical-capacity boom. Hang Seng is still carrying a property-disaster discount after policy has shifted from developer-level rescue toward project finance, completed-home delivery and capital-market support.
The cleanest convexity sits in Nasdaq because earnings are accelerating against very low hedge-fund exposure. Gold ranks next because crowded ownership can convert a macro catalyst into forced liquidation. EUR/USD needs no positioning squeeze: carry, energy and France are enough. Hang Seng has the weakest positioning evidence and the strongest policy-repricing potential.
- Short gold — the crowded long is ready to unwind. The secular store-of-value case is intact; the tactical trade is lower. Non-commercial accounts hold 277,159 longs against 33,825 shorts—a net long of 243,334 contracts after adding 21,145 in one week.3 Spot-gold ETFs received $4.4bn in the latest desk print and were tracking a fourth consecutive week above $1bn.4 That marginal exposure is vulnerable to the same two variables: a firmer dollar and higher US real yields. A negotiated Hormuz reopening would remove another layer of geopolitical premium.5
- Short EUR/USD — US carry meets Europe’s bad-options tree. US policy rates remain materially above the ECB deposit rate while Europe absorbs the larger energy-import shock. France compounds the problem: real consolidation damages growth; a diluted budget damages debt credibility; failure revives no-confidence and rollover risk. OAT-Bund has already traded above 87bp.7 Meanwhile, a September ECB hike is almost fully priced and another move is embedded by February 2027.6 Asset managers remain net long roughly 287,000 euro contracts while leveraged funds are already net short roughly 55,000, leaving liquidation capacity without making positioning the core case.8
- Long Nasdaq-100 — buy the AI earnings machine. Compute and Networking revenue more than doubled, demand remains supply-constrained and supply commitments jumped to $279bn.9 AI accounts for more than half of this year’s 9% growth in equipment and intangible investment, while S&P 500 profits are up more than 20%.1 Prime-book net leverage and the fundamental long/short ratio sit at first-percentile one-year readings; software exposure is at the fourth percentile.10 Leveraged funds remain net short about 44,000 Nasdaq-100 contracts even after covering roughly 23,000 in the latest week, while asset managers remain net long.8 Equity-fund inflows and buybacks at 1.7× last year’s pace provide the cash buyer behind any hedge-fund chase.11
- Long Hang Seng — policy tail risk is falling as technology profits rise. The property package moves financing toward projects, supports developer equity and bond issuance, expands ABS/CMBS/REIT channels, lengthens mortgage terms and eases land-payment pressure.12 A one-year US-China trade truce is already framing the September summit, compressing external risk premium.13 Paid inference is scaling: enterprise/API revenue rose 703%, July token usage reached 20× January and August ARR was reported above $800m; generative AI now supplies almost 80% of peer revenue.14–15 The detailed official table shows computer, communications and electronics profits up 105% in January–July versus 17.6% for industry overall.16 Disclosed HSI participant data do not show a washout, so the call rests on policy and profits rather than a squeeze.17
04 · What would change my mind
Only four developments break the book
The calls remain active unless one of these regime signals appears:
- Gold rises persistently with the dollar while ETF inflows continue, proving that geopolitical, diesel or fiscal-credit hedging has overwhelmed the real-yield channel.
- Weak US employment or inflation forces a Fed hold while the ECB delivers September and keeps February live, shifting the front-end differential decisively toward the euro.
- Long real yields and credit spreads rise fast enough to compress Nasdaq multiples despite continued AI revenue growth, turning AI financing into the discount-rate shock.
- China’s property measures prevent defaults but fail to revive home sales, household demand and broad corporate cash flow, leaving Hang Seng dependent on a narrow technology cohort.
Bottom line
Short the crowded hedges; buy the earnings and policy compression
The four-leg book is short gold and EUR/USD, long Nasdaq-100 and Hang Seng. It is short the assets most exposed to renewed dollar carry and long the assets where earnings or policy are improving faster than positioning and valuation imply.
Nasdaq owns the earnings engine. Gold owns the cleanest crowded unwind. EUR/USD owns the carry, energy and fiscal divergence. Hang Seng owns the policy-tail and technology-profit inflection. Four trades, four independent anchors, four explicit break conditions.
Sources & method
Primary sources, thesis separated from fact
- 1. Federal Reserve — In Our Time, Chairman Kevin Warsh, 28 August 2026
- 2. Barchart/Yahoo Finance — Hawkish Fed Chair Warsh Boosts the Dollar and Sinks Gold, 28 August 2026
- 3. CFTC — COMEX gold Commitments of Traders, futures only, positions as of 25 August 2026
- 4. Goldman Sachs Global Banking & Markets — ETF Color, 28 August 2026 (client report; publisher context)
- 5. Associated Press — Iran and Oman hold talks on managing the Strait of Hormuz, 25 August 2026
- 6. European Central Bank — account of the 22–23 July 2026 monetary-policy meeting, published 27 August 2026
- 7. TS Lombard — France fiscal-risk client note; Reuters/Boursorama — OAT-Bund market update, 21 August 2026
- 8. CFTC — Traders in Financial Futures, futures and options combined, positions as of 25 August 2026
- 9. NVIDIA — Form 10-Q for the quarter ended 26 July 2026
- 10. Goldman Sachs Prime Services — hedge-fund leverage and software-exposure client reports through 28 August 2026 (publisher context)
- 11. J.P. Morgan Prime Services and Goldman Sachs flow-of-funds excerpts — equity inflows and buyback pace through 28 August 2026 (publisher context)
- 12. CSRC and Chinese financial regulators — capital-market and credit support for the new real-estate development model, 28 August 2026
- 13. Associated Press — reporting on the one-year US-China trade truce and planned September summit, 24 August 2026
- 14. MiniMax — first-half 2026 results and results-call operating metrics, 26 August 2026
- 15. SenseTime — first-half 2026 results, 26 August 2026
- 16. National Bureau of Statistics of China — industrial profits, January–July 2026
- 17. HKEX — market-share and participant data for Hang Seng Index futures
Footnotes 1–17 correspond to the superscript references above. Sell-side and news attribution appears only in these footnotes. CFTC figures are dated snapshots; prime-broker and desk figures come from the named client reports, while the linked public pages identify the publisher and methodology context. Market levels are as cited at publication.