Core-Satellite Portfolio Construction

The structure

The core (80–95%). One to three broad funds (Chapter 27's classics): global equity plus a stabilizer sized by Chapter 19's allocation. Requirements: cap-weighted or genuinely broad, cheap, well tracked (Chapter 35), and dull. The core is not where you express anything. It is the part you would keep if you lost interest in investing forever.

The satellites (5–20% in total). Each must clear four gates before it exists:

1. A written thesis that is not "it has been going up."
2. A benchmark chosen in advance and matched to the thesis: a European luxury index for a luxury stock-picking satellite, a US tech index for a tech sleeve, the world index when the thesis is the tilt itself. What matters is that it is written down before the first purchase.
3. A size, inside the total satellite budget, at which being wrong is annoying rather than plan-altering.
4. An exit condition tied to the thesis rather than the price: what would have to become true for you to conclude you were wrong?

A position without all four is not a satellite. It is a stray purchase inside a portfolio, which is the thing the framework exists to prevent.

Why cap the satellites so hard

Three arguments, in increasing order of importance.

The evidence argument. Everything in Chapters 6, 45 and 47 says the expected value of active bets, after costs, is around zero or below for most people most of the time. A framework that lets those bets run to 40% of a portfolio has not read its own book.

The arithmetic argument. A satellite at 10% that trails the core by 5% a year costs 0.5% annually: real, survivable, roughly the price of the education. The same satellite at 40% costs 2% a year, which across two decades is the difference between comfortable and constrained. The cap is what converts a mistake into tuition.

The behavioural argument, which is the actual one. Most investors will not hold a pure index portfolio for thirty years without touching it. The urge to do something is not a character flaw to be lectured away; it is a constant, and a plan that assumes it away fails at its first bear market. Core-satellite works because it is realistic about humans: it gives the urge a legitimate outlet, capped in advance, so the 90% doing the compounding is never the thing being fiddled with. A portfolio that survives its owner is worth more than an optimal one that does not.

What belongs in a satellite

The legitimate cargo runs from pure risk management to pure conviction, and every point on that range has a place:

• Risk-management positions: offsetting a concentration you already carry (employer stock, a home-country pension, a business in one sector). Chapter 42's best retail use, and arguably not a bet at all.
• Structural insurance: gold at 0–10% (Chapter 23), judged on decades.
• Factor tilts with a written persistence story and famine tolerance (Chapter 45).
• Sector or regional views with a real thesis (Chapter 42).
• Themes that survive Chapter 47's five questions, which most do not.
• Individual stocks: the home of direct conviction. If you know an industry, believe in a company, or want the ownership itself, this is where that view lives: chosen deliberately, and sized with Chapter 18's skewness arithmetic in mind.

What must never be a satellite

• Anything you cannot bring yourself to cap. That feeling is information: the position is a conviction wearing an allocation's clothes.
• Leveraged or inverse products held beyond days (Chapter 72): the decay clock makes "it's only a small position" false over time.
• Money you will need. Satellites are funded from the long-horizon slice, never from the near-term rungs of Chapter 19's ladder.
• A satellite that has grown into the core by winning. Which is the next section, and the part everyone skips.

Maintenance: the rules that make it real

Trim winners back to their caps. A satellite that doubles has become a bigger bet than you authorised, and the discipline that felt smart when you wrote it is exactly what feels wrong now. Trim on your normal rebalancing date (Chapter 63), on schedule, without forming a new opinion.

Judge each satellite against its written benchmark, annually. Not against zero, and not against your memory. Two questions, in order: did it beat its own benchmark (was the selection good within the thesis?), and did the whole idea beat holding more core (was the thesis worth having?). A luxury-stock satellite can win the first and lose the second, or the reverse; knowing which is the entire point of writing the benchmark down.

Retire on failed theses, not falling prices. A thesis-based exit is a decision; a price-based exit is a panic. If neither the thesis nor the exit condition has triggered, the correct action is nothing.

Cap the count, not only the size. Five satellites at 3% each dilutes every conviction to background noise. Two or three positions you genuinely believe in, each large enough inside the budget to matter, is where the sleeve does its best work.

The honest limits of the framework

Stated plainly, because core-satellite is usually sold as having no downside:

• It does not improve expected returns. The expected effect of the satellite sleeve is slightly negative after costs. What it buys is behavioural, and that is a real purchase, not a consolation prize.
• It can be a rationalisation. "It's just a satellite" is an easy way to smuggle in a position that fails the four gates. The gates are the framework; the label is not.
• The pure indexer is not doing it wrong. A 100% core portfolio is the evidence's own answer, and anyone who can hold it without fidgeting needs none of this chapter. Core-satellite is for everyone else: the investor with views, knowledge or interests that deserve better than being suppressed until they erupt.

A worked example

Henrik, €200,000, 85/15 core-satellite.

• Core (€170,000): an all-world equity fund and an aggregate bond fund at his Chapter 19 allocation. Untouched except for annual rebalancing.
• Satellite 1, €12,000 (6%): gold ETC as insurance. Thesis: crisis and inflation diversification. Benchmark: none, judged on decades. Exit: none, it is structural.
• Satellite 2, €10,000 (5%): a healthcare fund. Thesis written; benchmark the world index; exit if the valuation premise breaks. Reviewed each January.
• Satellite 3, €8,000 (4%): three individual stocks in an industry he knows from work. Thesis: he understands these businesses and wants to own them directly. Benchmark: the sector index, and the world index for the sleeve as a whole. Honest label: genuine conviction, on a defined budget.

Year three: the stocks double to €16,000, now 7.4% of the portfolio against a 4% cap. The rule says trim, and he does it in January, hating it. Year five: one of them falls 60% and the trim looks like foresight, which it was not. It was a rule doing its job, written before anyone knew anything. That sequence is the whole argument for making rules while calm.