Strategies · Long-term investing
Core-satellite portfolio: the rules and a 22-year test
How a core-satellite portfolio works, the arithmetic that limits what satellites can add or cost, and a test of 80% SPY with 20% in four satellites from 2004 to 2026.
On this page
A core-satellite portfolio has two parts. The core is most of the money, held in one or two broad, cheap index funds and left alone. The satellites are a small share spread over narrower holdings: a region, a sector, an asset such as gold, or individual stocks.
It is less a strategy than a way of containing one. The core fixes what most of the portfolio will do. The satellites are where an investor's own views go, with a hard limit on how much they can matter. This lesson shows the arithmetic of that limit, states the rules of a plain 80/20 version, and tests it on real funds from December 2004 to October 2026.
The idea
Two facts sit behind the design. First, most investors who pick funds or stocks end up behind a plain index fund after costs, for reasons covered in Why most active funds underperform: costs, arithmetic and luck. Second, many investors want to act on a view anyway: that small companies are cheap, that another region will catch up, that gold belongs in a portfolio.
Core-satellite accepts both. It does not argue about whether the view is right. It caps the size of the bet, so that being wrong is affordable and being right still shows.
The arithmetic of the cap
If the core returns c and the satellites return s, a portfolio with a satellite share w returns about:
With w = 20%, the portfolio gets one-fifth of the gap, in either direction.
| Satellites against the core, per year | Effect on a portfolio with 20% in satellites |
|---|---|
| 10 points behind | About 2 points behind |
| 5 points behind | About 1 point behind |
| Level | None |
| 5 points ahead | About 1 point ahead |
The formula is approximate because the two parts compound separately and are reset each year, but the test below matches it closely. EFA, the developed-markets fund, returned 3.1% a year while SPY returned 8.8%, a gap of 5.7 points. A portfolio of 80% SPY and 20% EFA returned 7.7%, which is 1.1 points behind.
The rules
An educational outline of a core-satellite portfolio with fixed parts:
- Choose the core: one broad stock index fund. Here it is SPY, at 80% of the portfolio.
- Choose the satellites and write them down. Here there are four, at 5% each: IWM (small US companies), EFA (developed markets outside the US), EEM (emerging markets) and GLD (gold).
- On the last trading day of every December, bring every holding back to its target: 80%, and four times 5%.
- Between Decembers, trade nothing. A satellite is not added to because it fell, or sold because it rose, until the reset.
- A satellite may be replaced only at a December reset, and the 20% cap on all satellites together never changes.
Rules 4 and 5 are the ones people break. Without them the satellite share drifts upward after a good run, which is exactly when the cap matters most.
What the backtest shows
The test, stated in full so it can be checked:
- Data. Month-end closing prices of the five funds from December 2004, GLD's first full month, to October 2026: 263 months, nearly 22 years.
- Trading. The first purchase and each December reset happen at that month's closes.
- Costs. 0.10% of the amount on every buy and every sell. Fund fees are inside the prices: about 0.09% a year for SPY, 0.19% for IWM, 0.32% for EFA, 0.72% for EEM and 0.40% for GLD (issuers' and fund-data pages, November 2026).
- Left out. Dividends and taxes. SPY, IWM, EFA and EEM all pay dividends and GLD pays none, so every line in the figure is lower than a real holding would have been, the stock funds by something like 1.5 to 3 points a year.
| Measure | 80% core, 20% satellites | SPY alone |
|---|---|---|
| Total return | 489.3% | 525.2% |
| Return per year (CAGR) | 8.5% | 8.8% |
| Deepest fall from a peak | −50.4% | −52.2% |
| Longest time below a peak | 63 months | 64 months |
| Trading costs paid | 0.3% in 22 years | none |
One dollar became $5.89 in the mix and $6.25 in SPY alone. The two lines are hard to tell apart in the figure, and that is the result: with a 20% cap, four satellites that together returned 6.8% a year pulled the portfolio 0.3 points below its core.
The yearly resets moved about 36% of the portfolio in total over the whole period, under 2% a year, which is why trading costs came to only 0.3%.
The two parts of the period
| Period | 80% core, 20% satellites | SPY alone |
|---|---|---|
| December 2004 – December 2012 | 3.4% a year | 2.1% a year |
| December 2012 – October 2026 | 11.5% a year | 12.8% a year |
The satellites helped in the first eight years and hurt in the next fourteen. From 2004 to 2012, emerging markets and gold both far outran US stocks. After 2012 the US market led almost everything else. Someone who chose these satellites in 2012 because of how they had just done got the second row.
Other satellites, one at a time
Each row is 80% SPY and 20% in a single fund, reset each December, over the same period.
| Satellite (20%) | The satellite alone | The 80/20 portfolio | Its deepest fall |
|---|---|---|---|
| None (SPY alone) | 8.8% a year | −52.2% | |
| QQQ, Nasdaq-100 | 13.9% a year | 9.9% | −51.7% |
| GLD, gold | 10.1% | 9.4% | −40.8% |
| XLE, energy companies | 6.4% | 8.9% | −50.9% |
| IWM, small US companies | 6.9% | 8.4% | −52.2% |
| EEM, emerging markets | 4.7% | 8.2% | −54.3% |
| EFA, developed markets | 3.1% | 7.7% | −53.8% |
Two things stand out. The range of outcomes is narrow: from 7.7% to 9.9% a year, against a range of 3.1% to 13.9% for the satellites themselves. And two rows did better than the formula suggests. XLE alone returned less than SPY, yet the portfolio with XLE finished slightly ahead, and GLD lifted the portfolio by 0.6 points while cutting its worst fall by more than 11 points. Both moved differently enough from SPY that the yearly reset kept selling them high and buying them low (rebalancing explains the mechanism). That effect is real, but small, and it cannot be counted on in advance.
Other sizes for the satellites
| Core share | Return per year | Deepest fall |
|---|---|---|
| 100% | 8.8% | −52.2% |
| 90% | 8.6% | −51.3% |
| 80% | 8.5% | −50.4% |
| 70% | 8.3% | −49.5% |
| 60% | 8.1% | −48.6% |
| 50% | 7.9% | −47.7% |
Every step is about 0.2 points of return and one point of the worst fall. Nothing special happens at 80%. It is a convention, and the table shows what moving it does. Leaving the 80/20 mix alone with no resets at all returned 8.4% a year with a deepest fall of 49.0%, so the reset was not where the result came from either.
Where it fails: the losing periods
- It does not protect in a crash. From October 2007 to February 2009 the mix fell 50.4% and SPY 52.2%. Three of the four satellites were stock funds, and they fell further than the core: EFA by 59.7% and EEM by 61.9%. Only a satellite that is not a stock fund changed the worst fall, as the gold row shows.
- 2013 and 2014. The mix gained 24.8% and 8.3% while SPY gained 29.7% and 11.3%. Gold fell and markets outside the US lagged.
- 2021. The mix gained 22.2% against 27.0%.
- A satellite can be wrong for a very long time. EFA and EEM spent more than 17 years below their 2007 highs at month-end prices. XLE fell 71.3% from June 2014 to October 2020. At 5% or 20% of a portfolio that is survivable. It is the whole case for the cap.
Other ways it fails or misleads:
- Drift by neglect. A satellite that triples becomes a large position. Without the December reset the "small" bet stops being small, usually just before it reverses.
- Many satellites add up to an expensive index fund. Ten satellites at 2% each, each a diversified fund, mostly reproduce the core at a higher fee.
- Taxes. In a US taxable account each reset that sells a winner realizes a gain (IRS Topic 409). The test ignores this.
- The comfort is partly false. "Only 20%" sounds safe, but the core is 100% stocks here. A cap on satellites says nothing about how much risk the core carries; asset allocation is the lesson for that decision.
Compared with its neighbours
| Approach | What decides the result | In this test | What it asks of the investor |
|---|---|---|---|
| One index fund | The market | 8.8% a year | Nothing |
| Core-satellite, 80/20 | Mostly the market, a fifth by you | 7.7% to 9.9% a year | Keeping to the cap and the reset date |
| Satellites only | Your choices | 3.1% to 13.9% a year | Being right, and enduring long droughts |
| A mix of stocks and bonds | The share in stocks | Not tested here | See the asset allocation lesson |
How Finabase does it
Finabase does not choose a core or satellites and does not rebalance for you. These tools cover the bookkeeping and the checking.
- Portfolios. A Finabase portfolio is a record of trades with live values for each holding (Track portfolios and trades on finabase.io). Keeping the core and the satellites in one portfolio shows what each is worth today, from which the current shares follow. The target split and the reset date are yours to keep; the app does not store targets or trade.
- Comparison. The link above draws SPY and four possible satellites from the same starting value over two years (Compare stocks side by side on finabase.io), which shows quickly whether a satellite is a different thing from the core or the same thing with a higher fee.
- Backtest and Strategy. The Backtest tool trades indicator signals on one symbol with $10,000 per signal and no trading cost; it does not test a portfolio. The Strategy tool can split an amount across sibling branches by weight, such as 80% SPY and 5% in each of four funds, and replay it, but only over two years and without costs. Two years holds one or two resets, far too few to judge a mix; the test on this page covers 22.
Frequently asked questions
What is a good core-to-satellite split?
There is no correct number. In this test each 10 points moved from core to satellites changed the yearly return by about 0.2 points and the worst fall by about one point. The split decides how much a wrong view can cost, so the question is how much the investor is prepared to lose to their own ideas.
Did the satellites improve the portfolio in the test?
Slightly not. With IWM, EFA, EEM and GLD the 80/20 mix returned 8.5% a year against 8.8% for SPY alone. It was ahead through 2012 and behind afterwards.
Can individual stocks be satellites?
Yes, and the arithmetic is the same: a 5% position passes on one-twentieth of its difference from the core. A single stock can fall much further than a fund, which is the argument for keeping each one small.
How often are the satellites reset?
Once a year in this lesson, at the end of December. Resetting more often trades more and was not tested here; never resetting let the split drift and returned 8.4% a year against 8.5%.
Summary
A core-satellite portfolio keeps most of the money in a broad index fund and confines everything else to a small, fixed share that is restored once a year. The cap is the point: with 20% in satellites the portfolio receives about one-fifth of whatever the satellites do differently from the core. From 2004 to 2026 an 80/20 mix with small-company, foreign and gold satellites returned 8.5% a year against 8.8% for SPY alone and fell just as far in 2008. Across six satellites tried one at a time the result stayed between 7.7% and 9.9%. The structure made wrong views cheap and right views modest, which is all it claims to do.
Check your understanding
4 questions on this lesson. Each answer comes with the reason, and nothing is scored.
A portfolio is 80% in a core index fund and 20% in satellites. Over ten years the satellites return 5 percentage points a year less than the core. Roughly how far behind the core is the whole portfolio?
- About 5 points a year
- About 1 point a year
- About 0.2 points a year
Show the answer
B. About 1 point a year The portfolio receives the satellite share of the gap: 20% of 5 points is about 1 point. In the test, EFA trailed SPY by 5.7 points a year and the 80/20 portfolio with EFA trailed by 1.1.
In the 2007 to 2009 crash the 80/20 mix fell 50.4% and SPY alone fell 52.2%. Why did the satellites help so little?
- Because the satellite share was reset in December
- Because three of the four satellites were stock funds that fell with the core, some of them further
- Because trading costs used up the benefit
Show the answer
B. Because three of the four satellites were stock funds that fell with the core, some of them further EFA fell 59.7% and EEM 61.9% in the same stretch. A satellite softens a crash only if it is a different kind of asset: with gold as the only satellite the worst fall was 40.8%.
A satellite doubles in a year and is now 9% of the portfolio where its target is 5%. What do the rules say?
- Leave it, because it is working
- Add to it, because the trend is strong
- Sell it back to 5% at the December reset
Show the answer
C. Sell it back to 5% at the December reset The cap is the whole design. Letting a winner run turns a small bet into a large one, usually near its high. The rule restores the split once a year whatever the satellite has done.
The mix beat SPY from 2004 to 2012 and trailed it from 2012 to 2026. What does that say about choosing satellites?
- Satellites picked for their recent results can lead for years and then lag for longer
- Satellites should be changed every year
- The first period is the reliable one because it came first
Show the answer
A. Satellites picked for their recent results can lead for years and then lag for longer Emerging markets and gold led before 2012 and lagged after. Someone who chose them in 2012 on that record got the second period. The cap is what made that mistake cost 1.3 points a year in that stretch and not far more.
Sources and further reading
- Jack L. Treynor and Fischer Black, “How to Use Security Analysis to Improve Portfolio Selection,” Journal of Business 46(1), 1973, pp. 66–86 — a passive portfolio combined with a small active one
- William F. Sharpe, “The Arithmetic of Active Management,” Financial Analysts Journal 47(1), 1991, pp. 7–9
- iShares — MSCI EAFE ETF (EFA) fund page, expense ratio 0.32%
- iShares — MSCI Emerging Markets ETF (EEM) fund page, expense ratio 0.72%
- SEC, Investor.gov — Beginners’ Guide to Asset Allocation, Diversification, and Rebalancing
- IRS — Topic no. 409, Capital gains and losses
- Finabase chart data — SPY, IWM, EFA, EEM, GLD, QQQ and XLE month-end closes, December 2004 to October 2026; portfolios and results computed by Finabase
