The Minotaur Global Opportunities Fund gained 1.1% in October, though this missed our benchmark by 2.4ppts. Over the past twelve months, the Fund delivered +31.5% vs. the benchmark at +22.7%.

October’s headwind came from our US underweight, particularly the Mag7. We actively reshaped exposure: we cut Alphabet (a mistake in hindsight) and trimmed Meta (a good call given positioning/risk), kept an under-index weight in Nvidia, and broadened the AI infrastructure sleeve by initiating new positions in AMD and Micron (both positive contributors). We reduced Megachips on position sizing discipline and increased HCA and Dollar Tree on idiosyncratic catalysts and resilient earnings power. We also initiated Coway to diversify factor and regional drivers with a quality compounder outside the US mega-cap complex.

Another detractor for us this month was European Defence. The sector pulled back on de-escalation hopes around Trump/Putin headlines and profit-taking after a strong run. We’ve been steadily trimming into strength; the sleeve now sits near ~7% of the portfolio (down from a peak of 25% in March this year). Thesis discipline remains focused on solid backlogs and budget support but we’ll keep sizing the position in line with payoff asymmetry and event risk.

AI Infrastructure: Taking Stock
The latter part of October was characterised by a slew of AI results leading us to try to cement our thinking on where we sit on this thematic. Big Tech prints emphasised monetised AI, not just demos. A few examples include Amazon spotlighting agents atop Bedrock/SageMaker with a swelling backlog; Meta showing AI-driven ad/recs at real run-rates; Alphabet pointing to broad Gemini usage across Search/Cloud; Apple leaned into on-device AI; Tesla pushing autonomy/embodied AI; and Reddit discussing product and ads lifts from AI. It can now be seen how distribution and integrated stacks are capturing economics, but the cash-flow share across the stack remains uneven. That’s why we prefer a barbell approach of durable platform beneficiaries and “picks & shovels” (e.g., AMD/Micron) rather than paying any price for the narrowest cohort of winners. Our medium-term caution remains, however. What worries us looking 2-3 years out is tracking the rate of adoption — it’s hard to build systems that do this well, and stocks have re-rated so much that any hiccup there could be quite negative.

Context is Everything: Modular AI Skills
When working with AI, models often advertise large token limits, but the reality is that performance degrades quickly as the context window fills up. We've handled this in the past by being precise with the context we hand to the AI. For example, we'd point coding agents to different instruction sets — framework conventions, database patterns, deployment scripts — depending on what they were working on.

When Claude launched their Skills system in October, we recognised it as a clean way to standardise what we'd been doing ad hoc. We've since implemented our own version across Taurient, building ways for agents to learn skills on demand, like Neo downloading kung fu in the Matrix.

This approach scales elegantly — each new skill is a self-contained directory, so we can expand capabilities without loading irrelevant context into unrelated tasks. We're already seeing improved coding and research outputs from adding more specialised skills.

Positioning & Process
Whilst not the best result this month, we remain comfortable with our positioning. We remain anchored to our process owning mispriced opportunities at rational prices, diversifying across regions and themes, and investing heavily in the systems that scale our judgment. Our process prizes breadth over concentration, fundamental valuation over momentum, and asymmetry over index replication. That discipline has supported performance over time and, in our view, leaves us better placed if leadership rotates, if dispersion widens, or if earnings/positioning risk in the biggest names normalises.

To borrow Seneca, “Luck is what happens when preparation meets opportunity.” Our preparation is continuous, so when the opportunity set evolves, we shall meet it.