ETF Sampling vs. Full Replication: How Funds Track Indexes

Advanced7 min readUpdated March 17, 2026
ETF Sampling vs. Full Replication: How Funds Track Indexes

Key Takeaways

  • Full replication means the ETF holds every security in the index at the exact weight.
  • Sampling (or optimization) holds a representative subset of index securities to reduce costs.
  • Sampling is common for bond ETFs and broad international ETFs with thousands of constituents.
  • Full replication generally produces tighter tracking but costs more for large, illiquid indexes.

When an ETF tracks an index, it faces a fundamental choice: hold every single security in the index (full replication) or hold a carefully chosen subset (sampling). This decision affects tracking quality, cost, and the practical mechanics of how the fund operates.

Full Replication

A fully replicating ETF holds every security in its target index at the exact index weight. SPY holds all 500 stocks in the S&P 500. VOO does the same. For indexes with a manageable number of liquid securities, full replication is the gold standard because it minimizes tracking error.

The advantages are straightforward: the portfolio perfectly mirrors the index, eliminating any selection-based tracking deviation. The main cost is that the fund must hold and trade every constituent, including small, illiquid positions at the bottom of the index that may be expensive to trade.

Stratified Sampling

For indexes with thousands of constituents — like the Bloomberg US Aggregate Bond Index with over 10,000 bonds — full replication is impractical. Many of those bonds trade infrequently, making them expensive or impossible to acquire in the right quantities. Sampling solves this problem.

The portfolio manager divides the index into cells based on characteristics like sector, credit quality, maturity, and duration. They then select representative securities from each cell to construct a portfolio that matches the index's overall risk and return profile without holding every single bond.

Optimization Techniques

Modern sampling uses sophisticated mathematical optimization to minimize expected tracking error. The optimizer selects holdings that, in combination, most closely replicate the index's factor exposures, sector weights, and risk characteristics. This can produce surprisingly tight tracking with far fewer holdings than the index contains.

When Each Approach Is Used

US large-cap equity ETFs almost universally use full replication. US total market funds may use near-full replication for large and mid-caps but sample micro-caps. International equity ETFs often sample to avoid illiquid foreign stocks. Bond ETFs nearly always use sampling due to the vastness and illiquidity of bond markets.

Impact on Investors

For most investors, the replication method is transparent — you only see the end result in tracking difference. The key takeaway is that a well-managed sampling approach can track nearly as well as full replication while keeping costs lower. Check the fund's prospectus to understand which method your ETF uses and evaluate it based on actual tracking results, not the method itself.

Frequently Asked Questions

Which method is better?
Full replication is preferred when practical because it minimizes tracking error. For indexes with a manageable number of liquid securities (like the S&P 500), full replication is standard. For indexes with thousands of securities or illiquid components (like global bond indexes), sampling is more practical and cost-effective.
How does sampling work?
Portfolio managers use statistical optimization to select a subset of index securities that closely matches the index's risk, return, and factor characteristics. The sample is designed to minimize tracking error while avoiding the transaction costs of holding every security. Typically, the largest and most liquid holdings are fully replicated while smaller ones are sampled.
Can I tell if my ETF uses sampling?
Check the fund's prospectus or annual report, which must disclose the tracking methodology. If the number of holdings is significantly less than the index constituents, the fund uses sampling. For example, an ETF tracking a 10,000-bond index that holds 3,000 bonds is clearly using sampling.

Related Articles

More in Advanced

View all →

Ready to explore ETFs?

Use our free tools to research, compare, and find the right ETFs for your portfolio.

Explore ETFs on ETF Beacon