Global liquid equities and exchange-listed ETFs. For a GBP investor, ETFs are taken as LSE-listed GBP UCITS lines to avoid currency conversion; single stocks are bought in their native currency. AI/semis, power, defence, gold and crypto proxies are the active 2026 themes.

Every Sunday the books are repriced off the previous Friday's closes, and Claude works in a fixed order: the benchmark first, then the market overview, then positions, then the hold-or-trade decision, then the watchlist and the week ahead. Taking the benchmark first is deliberate, so the write-up cannot be framed to flatter the portfolios. Two parallel books test the same themes through different instruments (ETFs vs single stocks), so the wrapper question gets a clean answer.

No derivatives, leverage or shorting. Cash is permitted and accrues interest. The default is to hold: a three-to-five-year horizon does not trade weekly noise. The 10% position cap applies at purchase rather than to drift, with a mechanical trim only above 12%. Valuation updates publish automatically, but any discretionary buy, sell or rebalance is proposed and held for human approval before it appears here. All three start level at inception; dealing friction (0.10% commission, 0.5% UK stamp on UK shares, 0.25% FX spread each way on non-GBP assets) applies to trades made thereafter.

Indexed against Vanguard's S&P 500 UCITS ETF (VUAG) in GBP, a real, GBP-denominated, total-return tracker. Both books are reported separately so the ETF vs single-stock question is settled by the data, not the narrative.

Prices are taken as dated closing prices for each exchange, so the US, London and Xetra lines are all marked on the same Friday. Each holding is converted using its own quoted currency rather than an assumed one, which matters because one line quotes in pence and another quotes in dollars despite its London listing. Percentages are computed from the underlying figures rather than entered by hand. Every update must reconcile, with holdings plus cash equalling NAV, and anything that looks wrong, a move beyond 25% in a week or a price that suggests a currency-unit error, halts publication for review rather than going live.