Sampling Strategy

A sampling strategy (also called optimization or stratified sampling) is used by index funds and ETFs that don't hold every single security in their benchmark index. Instead, the fund holds a carefully selected subset of securities that is designed to match the index's risk and return characteristics.

Why Use Sampling Instead of Full Replication:

1. Illiquid securities — some index constituents are hard to buy/sell
2. Too many holdings — indexes with thousands of stocks (Russell 3000, FTSE Global)
3. Transaction costs — buying every stock is expensive for smaller funds
4. Foreign market access — some markets restrict foreign ownership
5. Minimal impact — tiny index constituents barely affect returns

Full Replication vs Sampling:

Approach	Holdings	Tracking Error	Cost	Use Case
Full replication	All index stocks	Very low	Higher trading costs	S&P 500, large indexes
Sampling	Representative subset	Slightly higher	Lower trading costs	Broad/international indexes

How Sampling Works:

1. Stratify — divide the index into cells by sector, size, country, etc.
2. Select — choose representative securities from each cell
3. Weight — assign weights to match index characteristics
4. Optimize — use statistical models to minimize tracking error
5. Monitor — continuously check and adjust the sample

Example: An emerging markets ETF tracking a 3,000-stock index might hold only 1,500 stocks, selected to match the index's sector weightings, country allocation, and risk characteristics.

Tracking Error from Sampling: Well-executed sampling typically adds only 0.05%–0.20% tracking error — a worthwhile tradeoff against the cost savings.