Core-satellite with a broad core and targeted satellites: costs, tracking error and rules

Basics9 min read · updated 3 October 2026 · 8 sources

Core-satellite splits the portfolio into two roles: a large, broadly diversified and low-cost core, usually a world ETF, and a few smaller satellites for factors, themes, individual stocks or active funds. The core determines risk and return, the satellites the deviation from the market. Here we calculate the total costs as a weighted TER and the active risk as the tracking error of the whole portfolio (with correlations), show how a risk budget limits the size of the satellites, and collect rules for rebalancing and behavior.

The idea: a broad core and a few satellites

The core should track the market as closely and cheaply as possible: an ETF on the MSCI ACWI or FTSE All-World, or a combination of World and emerging markets (see MSCI World vs ACWI vs FTSE All-World). The satellites are deliberate deviations: a factor ETF on value or small caps, a thematic ETF, individual stocks, an active ETF or an actively managed fund. Because the satellites are small, the portfolio stays close to the market, and mistakes in selection cost little.

Core-satellite is not an answer to the question of how much equities, bonds and money market belong in the portfolio. That question is answered by the strategic asset allocation. It is a question of implementation: within an asset class, how much is replicated passively and how much deviates actively? Waring et al. (2000) treat exactly this as an optimization: for a given budget of active risk there is a best mix of managers, and the passive core follows in inverse proportion to this risk budget. The less deviation you allow yourself, the larger the core.

Total costs: the weighted TER

The weighted TER shows what the satellites cost the whole portfolio. Expensive satellites carry little weight if they are small. That is the appeal of the structure. A fairer comparison is with what the satellites deliver in deviation: extra costs divided by tracking error gives the information ratio the satellites must at least achieve to earn their costs.

Active risk: the tracking error of the whole portfolio

Because the core tracks the market, the portfolio's active return arises solely in the satellites, measured by their deviation from the core, not from their own index. The portfolio's tracking error follows from this with the correlations of the active returns:

A tracking error of 1.86% roughly means: in about one year in three, the portfolio is more than 1.86 percentage points above or below the core alone. The correlation of the satellites matters a great deal here: if their deviations move independently of each other, 1.48% remains; if they move in step, it is 2.55%. Two satellites that both bet on technology and growth hardly diversify the active risk.

How big can the satellites be? A risk budget

Instead of sizing the satellites by gut feeling, you can set a budget for the tracking error, for example 2% relative to the core, and work backwards from it. A single satellite with a 15% tracking error may then make up at most 2 / 15 = 13.3% of the portfolio; two uncorrelated satellites of this kind together √2 times as much, 9.4% each. How much deviation you can stand is a personal question: if you get annoyed after two years of trailing the market by three percentage points and then throw everything over, your budget is too large. Large investors use the same principle for their tactical deviations (see active risk budget).

Rebalancing rules

1. Set target weights and bands: for example, core 80% with a band from 75 to 85%, each satellite at most 10%. Only rebalance once a band has been breached (see rebalancing).
2. Direct your contributions: new money goes into the part that is below its target, usually the core. This saves order fees and taxes.
3. Trim the winners: a successful satellite grows beyond its budget on its own. If you do not cap it, an add-on turns into a concentration risk.
4. Review losers rather than automatically buying more: with a factor ETF, buying more can be part of the rule; with an individual stock whose story has broken, probably not. Write the rule down in advance.
5. Mind the taxes: in a taxable account, every sale at a gain can trigger tax. Bands instead of fixed dates keep the number of sales low.

Amenc et al. (2004) describe a dynamic variant: the share of the satellites is varied following the pattern of portfolio insurance (CPPI), so that the tracking error stays below a threshold. After good phases there is more satellite, after bad ones less. For private investors, a fixed upper limit with bands is simpler and sufficient in most cases.

Behavior: a fenced-in playground

Core-satellite fits an observation from behavioral economics: many investors think in separate pots. Shefrin and Statman (2000) describe portfolios as layers of mental accounts: a safe layer against poverty and a risky layer for a shot at riches. The core is the safe layer, the satellites are the layer for ideas. This can help: if you channel the urge to trade into a small part of the portfolio, you leave the core alone.

But it can also do harm. Private investors who trade a lot achieve worse results on average (Barber and Odean 2000), and they prefer to buy stocks that are currently attracting attention (Barber and Odean 2008). Satellites that are constantly swapped cost fees and taxes and rarely deliver excess return. It helps to measure the satellites separately: what has each one delivered relative to the core, after costs? A simple form of performance attribution shows this.

For advanced readers: alpha, risk budget and the fundamental law

According to Waring and Siegel (2003), what an active manager delivers can be broken down into market factors and pure alpha. Market factors can be had almost for free through index funds; pure alpha costs active risk that is uncorrelated with the risk of the basic allocation. That is why the basic allocation and the selection of the active building blocks can be optimized separately, and the optimal structure consists mainly of index funds and risk-controlled active funds, with little weight for highly concentrated funds. For private investors this means: a factor ETF as a satellite delivers mainly factor risk, which you can also buy directly and cheaply. Genuine alpha is rare.

The fundamental law of Grinold and Kahn (2000) shows why a few individual stocks rarely achieve a high information ratio: ten decisions a year with a skill of 0.05 give IR ≈ 0.05 × √10 = 0.16. With a 1.86% tracking error, that would be an expected 0.29 percentage points a year before costs, and only if the skill is genuinely positive. For the average of all active investors, the rule of Sharpe (1991) applies in any case: the market return before costs, below it after costs.

What the research shows

Research treats core-satellite as a question of risk budgeting: Waring et al. (2000) optimize the mix of passive and active building blocks under a budget for active risk, while Waring and Siegel (2003) separate pure alpha from factor risk and arrive at structures with a lot of index and few concentrated funds. Amenc et al. (2004) argue that tracking error is not bad per se (upward deviations are welcome) and propose a dynamic allocation that limits the downside risk. On the behavioral side, Shefrin and Statman (2000) show why investors think in layers at all, and Barber and Odean (2000) what excessive trading costs. The arithmetic of Sharpe (1991) is a reminder that satellites must lose after costs on average across all investors; if you hold them, you are betting on being better than average.

Limitations and typical mistakes

• Too many satellites: eight add-ons of 2.5% each are hard to keep track of and often overlap heavily. That means more funds, not more diversification.
• Tracking error against the wrong benchmark: for the portfolio, what counts is each satellite's deviation from the core, not from its own index.
• Ignoring correlations: three satellites on technology, growth and AI are essentially one bet.
• Thinking of costs only in terms of TER: order fees for individual stocks, spreads and taxes on reallocations belong in the calculation.
• No fence: without an upper limit, successful satellites grow into the core, and unsuccessful ones are topped up out of stubbornness. Not investment advice.

How Portfolio Lab calculates it

In the Backtest you choose the core as the benchmark and see the tracking error, information ratio, beta and upside and downside capture for your portfolio with satellites; you can set rebalancing by calendar or tolerance band, contributions into underweight positions and trading costs. On the Risk page, the app breaks down the tracking error from the active weights relative to a benchmark into contributions per position using the Euler decomposition. This is exactly the calculation from the example, with estimated covariances. You will find the ongoing costs of the funds under Securities; the app does not calculate a weighted TER for the portfolio, as the costs are contained in the fund prices. How much core and satellites overlap is shown by the Look-through. The metrics are described in the metrics methodology.

This article explains a method; it is not investment or tax advice. Worked examples are simplified and rounded. All terms in the glossary.