Patience
A durable moat the market has temporarily mispriced, held on a longer horizon than almost anyone else is willing to. Years to a decade. Broken when the moat erodes or returns on capital structurally fall — not when the price moves.
02
Investments
The book, open.
What you'll find: a hyper-concentrated book — 8 to 12 positions, long only, no leverage. Every holding with its weight, what it's done, and the single edge it relies on. If I can't name the edge, I don't own it.
€36,689
Book value
+€10,906
Total return since inception
7
Positions · mandate is 8–12
19.5%
Cash · target 3%
€36.7k
click a slice
€7,162Target 3%Overweight
Optionality — and right now, an overhang. Nineteen per cent against a three per cent target isn't a macro view, it's undeployed capital waiting on the next reverse-DCF. Cash is the correct hedge against a rate shock, and it's free.
€5,934Edge · PatienceTarget 13%
A durable moat the market re-prices on every AI headline. The edge is horizon, not insight: I am willing to hold it for a decade while others trade the quarter. Held until the moat erodes or returns on capital structurally fall — not until the story gets boring.
€5,411Edge · PatienceTarget 13%
Every credible AI build-out routes through one company's advanced-node capacity, and that capacity cannot be replicated on any horizon that matters. The concentration risk is real and geographic — which is the same risk I write about in section 01.
€5,200Edge · Catalyst — firedTarget 8%
Bought for the high-bandwidth memory cycle. The catalyst has fired and the re-rate happened, which under my own rules means the position is now living on momentum rather than thesis. Momentum is not one of my three edges. It gets trimmed to target.
€5,050Edge · CatalystTarget 11%
The obesity-drug franchise, held against dated pipeline readouts rather than a vague demographic story. A catalyst position has a defined window: when the events land, the position is reassessed on the evidence, not extended out of comfort.
€4,723Edge · CatalystTarget 9%
The accelerator cycle, sized as a catalyst rather than a forever-hold — which means it carries an exit condition, not just a narrative. Together with TSMC and SK Hynix this is one risk unit, not three positions; over half the equity book sits in semiconductors.
€1,948Target 0%Scheduled exit
A thesis that stopped working, kept only for one last dated window. The rule was written in advance: no credible growth path presented and it goes, regardless of price. Writing the exit down before the day arrives is the whole point — it's the only defence against talking myself back into it.
€1,263Edge · InformationalTarget 6%
Uranium and rare-earth processing — one sentence to explain, still under-followed by professional capital. The same theme as the popular pure-play at a fraction of the multiple. Currently under target: this is where cash should go before it goes anywhere else.
Live from Scalable Capital, 31 Aug 2026. Performance is unrealised, measured against the average cost of shares still held (FIFO); it excludes dividends and gains already realised on trims. Nothing here is investment advice.
Three edges, and only three
If I can't name one, the answer is noInformational
A business simple enough to explain in one sentence, still under-followed, narrative not yet arrived. Medium horizon. Broken the moment the narrative shows up and the price re-rates — which is a good outcome, not a reason to stay.
Catalyst
A specific, dated event that forces a re-rate inside a defined window. Short horizon by construction. Broken when the catalyst fires, or when the window slips indefinitely.
No kill-switch, no buy
Every position needs a checkable event that would end it — a margin below a number, a contract lost, a date passed. "If fundamentals deteriorate" is not a kill-switch. Written before the purchase, when I still have nothing to defend.
Cheap is not the signal
A reverse-DCF tells you what the price requires you to believe. It falsifies; it cannot rank. Cheap with returns on capital falling is a value trap. Cheap with a widening moat is the only thing worth owning.
The real constraint is holding period
My idea quality has been fine. The damage has come from selling things that were working. So the discipline isn't finding more names — it's owning the ones that survive the framework for as long as the edge lasts.