Tycho Arete Macro Fund

Overview

Tycho Arete Macro Fund aims to deliver competitive risk-adjusted returns while maintaining a low correlation with all major asset classes. It strives to construct and update macro-analytical frameworks that incorporate the rapidly changing macroeconomic conditions around the world, as well as the significant idiosyncrasies of large global actors such as China and Japan.

Strategy & Manager

Performance

Commentary

Investment Manager’s Commentary – September 2026

September Outlook

September was the month the bond market presented the bill. The selloff in Treasuries accelerated, the US 10-year yield rising 53bp to 5.28% — its largest monthly climb in four years — while the Fed, which entered the year priced for cuts, hiked for the first time in three. On the surface there were culprits enough: oil back above $100, a hawkish turn in Washington. But the deeper force remains structural: the investment boom continues to gobble up capital, the sums now so great that even historically cash-rich tech groups must fund the build-out with debt — presenting lenders with a set of complex calculations that demand an increasingly higher term premium. Our short in duration — the book’s longest-held expression of that view — collected +2%, the month’s largest contributor.

Equities, remarkably, did not break on a 53bp yield move; they split. The Nasdaq gained 3.2% even as the Russell 2000 fell 5.4% and most stocks corrected quietly beneath a resilient index surface. Against the mounting pressure of rising rates, markets were repricing AI demand and adoption: new frontier model and agent roll-outs from OpenAI (Astra) and Meta (Muse) drew millions of users within weeks, agentic workloads promised a step-change in inference demand, and the revenue reads have so far confirmed it. The numerator, for now, is outrunning the discount rate. Everything outside the complex — rate-sensitive and without an AI growth story — absorbed the full 53 basis points. Leadership did not rotate; it narrowed. We were positioned for precisely this shape of tape — directional longs in Nasdaq, FTSE Taiwan and our hardware basket against shorts in the broad S&P and the AI laggards — and both legs paid.

China was the month’s casualty, and for three compounding reasons. The Xi-Trump summit optimism that had supported the tape through the summer faded without concrete deliverables. The AI gap widened — September’s agentic breakthroughs were all American, a reminder that the ecosystem bifurcation cuts both ways. But most crucially, the fundamentals kept weakening at home: profits softened into Golden Week, and the domestic economy keeps adding gravity faster than the innovation complex can generate escape velocity. The CSI 300 lost nearly 6%. Our onshore and HK longs — financials, internet, AI hardware — bled roughly 0.3% in sympathy, but more than offset by our hedges in China A50 and HSI. With gravity winning for now, we have further reduced our long exposure in HK/China.

Current Outlook

To understand the most consequential trade in the world today, it helps not to start in Silicon Valley. Start instead in China, circa 2010 — a build-out visible from everywhere, suppliers booking profits years before end demand arrived, credit flowing freely to a sector deemed too important to fail. We spent our careers studying and trading that boom from the inside, through its long ascent and potentially longer unwind. Watching the AI build-out today, we keep encountering the same economic mechanisms.

For the AI boom is, before it is anything else, an investment boom — capital-intensive, debt-hungry, and strategically important. The technology is genuinely new; the economics are not. Because the breakthrough is real, the market reads the build-out as a secular trend and prices it accordingly. But secular trends are claims about demand, and what is being built is supply — at staggering capital intensity and high marginal cost. Chips depreciate, power is metered, every incremental token carries a bill. The parallel most often reached for is the dot-com era, and on the infrastructure side it is not wrong — the fiber build-out was also capital-hungry, also ran ahead of demand, and ended in a glut.

But the business model this time is fundamentally different. The internet era sold software and websites: expensive to create, nearly free to serve, so that once demand arrived the marginal economics were extraordinary. The AI era sells intelligence by tokens, and every token is manufactured — on depreciating hardware, with metered power.High capital intensity and high marginal cost is not the anatomy of the internet business model; it is the anatomy of an industrial one. Which is why the closer relative, to our eyes, is China’s property boom.

Same Mechanism, Different Sector

Strip the two booms to their core mechanics and they share more similarities than differences, running the same four gears:

1. Investment-led growth: Capex is booked as supplier earnings years before end demand arrives, if it arrives. Upstream profits led China’s market for a decade while the consumption share fell, just as hyperscaler and supply-chain earnings now dominate index EPS growth while the consumer stalls.

2. A rising cost of capital: The build-out’s demand for savings outruns supply, and real yields grind to cycle highs. Financial repression masked this in China but did not prevent it — the true price surfaced off balance sheet, in trust loans and private credit. The West’s version prints more honestly, every day: a 10-year real yield at fifteen-year highs.

3. Crowding-out: Capital abandons whatever is not the priority sector, and investment outside the complex goes flat. In China, property and infrastructure absorbed the banking system’s credit for a decade while private, non-property investment stagnated — “the state advances, the private sector retreats” (国进民退). The world now runs the same pattern with new labels: AI, power, grid and defence take the capital and the policy priority, while S&P capex ex-AI is flat and rate-sensitive activity — housing, non-data-centre construction — wilts under the very yields the build-out created. The boom does not merely outgrow the rest of the economy; it actively de-funds it.

4. Policy complicity: Once a sector is declared strategic, leverage is tolerated and risk premia compress. In China, LGFV debt priced as quasi-sovereign for a decade on the belief that Beijing would not let property fail; today, AI framed as a national-security race compresses risk premia across the complex.

Where We Are in the Cycle

Every investment boom is, at its core, a race between two lines — the cost of capital, which rises as the build-out absorbs the world’s savings, and the return on invested capital, which deteriorates as each marginal project gets worse. The boom does not end when the story breaks; it ends when the lines cross — when the marginal dollar invested no longer clears its funding cost.

What complicates the picture is that circular financing masks the crossing. In China, local governments — the owners of the land — created their own buyers: financing vehicles, funded with borrowed money, that purchased land from the very governments that sponsored them, with the proceeds booked as the fiscal revenue that justified further borrowing. The supplier manufactured its own demand and called it income. In the AI complex, the chip-maker’s equity funds the model labs whose multi-year compute commitments return as the chip-maker’s revenue — the seller financing its own customer, demand counted at face value. Both loops inflate the very revenue used to judge returns. Measured ROIC stays flattering while marginal ROIC quietly sinks, and so the lines can cross long before the market notices they have.

Where do the signposts place us now? Late in the acceleration stage, with the first indicators of transition flickering: capex approaching parity with free cash flow, debt becoming the marginal source of funding, circular commitments a growing share of booked growth. We are not calling a top — China taught us how long this stage can reward the patient bull, and how expensive it is to be early. But September’s bond market, charging 53 basis points more for the month, suggests the cost-of-capital line is no longer waiting patiently.

How to Trade This

Our book is positioned for a late-stage acceleration of investment and hedged for the peak, built on four pillars. First, the rising cost of capital itself: short US duration, a curve steepener and short gold — three legs of one trade, tied respectively to nominal yields, the term premium and real yields. Second, the acceleration’s beneficiaries: long Nasdaq and Taiex futures, long global AI hardware and the CSPs — contracted demand and visible cash flow still command the market’s only premium. Third, the peak-investment hedge: short Nikkei, among the most levered equity markets to AI spend and uniquely exposed to a rising cost of capital through its own bond market — higher JGB yields and a firmer yen strike from both sides. Fourth, a moderate short in China via equity indices, reflecting a domestic economy whose gravity is, for now, winning against the innovation complex’s escape velocity.

Documents