Core–satellite in practice: sizing satellites with a tracking-error budget

What you will learn

• What tracking error means for a private portfolio and why it is the right budget for satellites
• How to size one or several satellites from a tracking-error budget
• How to use active share, overlap and the information ratio to test a satellite
• Three signs that a satellite adds nothing

Explanation

A core–satellite portfolio holds a broad, cheap index core and a few smaller satellite positions meant to add return, a factor tilt or a theme. This lesson is about sizing. The key measure is tracking error (TE): the annualised volatility of the difference between the portfolio's return and the core benchmark's return. It is also called active risk. Every satellite adds some.

For a single satellite with weight w and a TE of its own against the core of TE(s), the portfolio TE is approximately w × TE(s). For several satellites whose active returns are uncorrelated with each other, the contributions combine like volatilities: TE(portfolio) = √( Σ (w(i) × TE(i))² ). If their active returns are correlated (for example two technology-themed funds), add them almost linearly instead. A tracking-error budget is the maximum TE you accept; 1% to 3% per year is typical for private investors and already means trailing the core by 2–6 points in an unlucky year.

Three checks decide whether a satellite deserves its budget. Active share is the share of the satellite's holdings that differs from the core; a fund with 20% active share mostly duplicates the core. Overlap checks total exposure: if the core has 25% in technology and you add a 10% technology satellite, the portfolio holds 0.9 × 25% + 10% = 32.5% technology, which is concentration, not diversification. The information ratio (IR) is expected active return divided by TE; 0.3 is considered good for professional managers, and a satellite with an IR of 0.1 after costs adds very little.

A satellite adds nothing when (1) its correlation with the core is above about 0.95 and its expected return is not higher, (2) its extra cost (fee difference, spread, tax drag from turnover) exceeds its realistic expected outperformance, or (3) it doubles an exposure the core already holds. Then merge it back into the core.

Worked example

Hypothetical example: a €400,000 portfolio with a global equity core. The investor sets a tracking-error budget of 2% per year. Two candidates: a small-cap value ETF (assumed TE against the core 10%, expected outperformance 1.0% per year before extra costs, extra cost 0.4%) and an emerging-markets ETF (TE 8%, outperformance 0.5%, extra cost 0.3%). Active returns assumed uncorrelated. All figures are illustrative; core fees, spreads and taxes are excluded.

Step 1 – one satellite alone. Small-cap value only: w × 10% ≤ 2% → w ≤ 20%, at most €80,000.

Step 2 – two satellites. Try 10% and 12%. TE = √((0.10 × 10)² + (0.12 × 8)²) = √(1.00 + 0.92) ≈ 1.39%, inside the 2% budget.

Step 3 – is it worth it? Net expected outperformance: 1.0% − 0.4% = 0.6% and 0.5% − 0.3% = 0.2%. Portfolio add = 0.10 × 0.6% + 0.12 × 0.2% ≈ 0.084% per year, about €336 on €400,000, for a TE of 1.39%. IR ≈ 0.084 ÷ 1.39 ≈ 0.06.

Step 4 – decision. The emerging-markets satellite uses almost half the active risk for a quarter of the expected add, and many global cores already contain emerging markets at market weight, so it doubles an existing exposure. A cleaner portfolio keeps small-cap value at 15% (TE 1.5%, expected add 0.09%) and drops the other. The add is still small, and a 1.5% TE can mean trailing the core by 3 points or more in a bad year with no guarantee of catching up.

Common mistake

Practice activity

A €120,000 portfolio has a TE budget of 1.5%. Candidate satellites: a momentum ETF (TE 7%, net expected add 0.4%) and a sector ETF (TE 12%, net expected add 0.3%). Active returns are assumed uncorrelated. (a) What is the maximum weight of each if used alone? (b) With 10% momentum and 5% sector, what is the portfolio TE and the net expected add in euros per year? (c) Which one would you drop first, and why?

Key takeaways

• Size satellites from a written tracking-error budget: weight ≤ budget ÷ satellite TE.
• Uncorrelated satellites combine by root of summed squares; correlated ones add almost linearly.
• Check active share, overlap with the core and the information ratio before adding anything.
• A satellite that duplicates the core, costs more than it is expected to add, or correlates above 0.95 with the core should be merged back.

Next step

The next lesson leaves the static picture and asks how the allocation should change over a lifetime: glide paths, human capital as a bond and why the path should bend.