How few holdings are enough for an index?
Index replication with single stocks picks, from the holdings of an index fund, the securities and weights that track the index best: for stocks with the lowest tracking error, for bonds with duration, spread, maturities and key rates as in the index. You set the number of holdings, the maximum weight and limits per sector or country.
Free during the beta, no ads and no subscription. Once you sign in, this page opens the tool.
How it works
An index often holds hundreds or thousands of securities. Anyone who wants to replicate it with a few, for their own portfolio or a mandate, has to select and weight them. Portfolio Lab takes the holdings list of an index fund as the target, for US funds from its filing with the Securities and Exchange Commission (SEC), or a list of your own.
For stocks, the tool estimates from daily or weekly prices how the securities and the index move together and searches for the weights with the lowest expected tracking error. The number of holdings is not a simple limit: a search with several starting points drops, adds and swaps securities until the objective stops falling. All limits are checked again at the end.
How to read the result: the tracking error says how much the return of the replica is likely to vary around that of the index. The active share shows how much of the portfolio differs from the index. The test period after the estimate shows whether the replica also holds up outside the sample.
What a result looks like
Sample dataMade-up sample data, no market prices and no recommendation. Calculated with the same methods as the tool itself.
Inputs in this example
Locked: you can't change the inputs in the preview.
Run your own numbers with a free account Your own inputs, up-to-date data and every setting.
Create accountReplica versus index
Estimation period 30 Jun 2023 to 30 Jun 2025, test period afterward 30 Jun 2025 to 31 Dec 2025. In the estimation period the calculation knows the prices, so the result looks too good there.
| Metric | Estimation period | Test period |
|---|---|---|
| trading days | 521 | 132 |
| Return of the replica | 6.7% | -13.5% |
| Return of the index | 11.3% | -12.9% |
| Difference p.a. | −2.16% | −1.16% |
| Tracking error (measured) | 1.56% | 1.66% |
In the estimation period (start = 100)
Buy-and-hold with the weights as of the holdings date, calculated back.
In the test period (start = 100)
Held afterward, outside the sample.
Shares by Sector
Index versus replica; the index weight ± 1% is allowed.
Shares by Country
Index versus replica; the index weight ± 2% is allowed.
Largest holdings of the replica
Fictitious companies, weight in the index and in the replica.
| Ticker | Sector | Country | Index weight | Weight |
|---|---|---|---|---|
| BX129 | Financials | Germany | 7.71% | 8.00% |
| BX190 | Information technology | Canada | 5.22% | 6.31% |
| BX199 | Information technology | United Kingdom | 4.16% | 5.60% |
| BX074 | Information technology | USA | 4.71% | 4.54% |
| BX114 | Information technology | USA | 3.03% | 3.97% |
| BX115 | Consumer discretionary | Germany | 2.62% | 3.60% |
| BX141 | Industrials | USA | 2.27% | 3.17% |
| BX108 | Energy | USA | 3.29% | 3.12% |
| BX110 | Financials | Germany | 2.24% | 3.05% |
| BX062 | Consumer discretionary | Switzerland | 3.02% | 2.95% |
| BX131 | Utilities | France | 2.20% | 2.93% |
| BX092 | Health care | Japan | 2.22% | 2.73% |
Assumptions: A fictitious index with 200 holdings, weights by market capitalization and daily prices from a factor model with market, sector, country and specific risk; the index ETF holds all securities by market capitalization. Calculated with the same method as the tool.
Frequently asked questions
How many holdings do you need to replicate an index?
It depends on the index. For broad equity indices, the tracking error falls clearly with the first few dozen holdings and ever more slowly after that. Limits per sector and country keep the replica close to the index in sub-markets as well.
What is tracking error?
The volatility of the return difference between replica and index, annualized. At 1.5%, the difference stayed within about 1.5 percentage points in roughly two out of three years.
What is active share?
The share of the portfolio that differs from the index: half the sum of the absolute weight differences. 0% means exactly the index, 100% means no holding in common.
Where does the data come from?
With an account, the tool works with the holdings lists of US index funds from their SEC filings, with lists of your own and with the prices of the securities. On this page, the index, the holdings and the prices are fictitious.
Learn more in the knowledge base
Run it with your own portfolios
Create an account, confirm your email and use this tool with up-to-date prices and your own ETFs.
- Price
- Free during the beta
- Data
- Licensed prices; interest rates, inflation and factors from public sources
- Privacy
- Servers in the EU, no ads, tracking only with your consent
- Devices
- Computer and phone, everything saved in your account