Full Replication vs. Sampling: How Do Index Funds Track Their Benchmark?
Tracking an index sounds like it should mean owning exactly what the index owns, but for many funds that’s not practical, and the workaround they use instead has real consequences for how closely they follow their benchmark.
The short answer
Full replication means a fund buys every single security in its target index, in matching proportions, to mirror it as closely as possible. Sampling means a fund instead buys a representative subset of securities chosen to approximate the index’s overall behavior, without owning every constituent. Full replication is generally more precise but isn’t always practical, especially for indexes with very large numbers of holdings.
When full replication makes sense
Full replication works well when an index has a manageable number of constituents and those securities are reasonably liquid and easy to trade in the necessary proportions. In these cases, a fund can buy each holding directly, which tends to produce very close alignment between the fund’s performance and the index’s performance, since there’s no approximation involved. This is the simplest and most direct way for an index fund to track its benchmark.
Why sampling gets used instead
Some indexes include thousands of individual securities, some of which trade infrequently or in small volumes, making it costly or impractical to hold every single one in exact proportion. In these situations, a fund manager may use sampling: selecting a subset of securities, chosen and weighted using statistical techniques, intended to behave similarly to the full index without literally owning every constituent. This approach can reduce trading costs and improve practicality, particularly for indexes covering harder-to-trade segments of the market, such as certain bond indexes. Sampling is a different kind of adjustment than smart beta weighting — it’s a practical tracking technique, not a deliberate tilt toward particular investment factors.
The trade-off: tracking error
- What it is. The difference between a fund’s actual performance and its target index’s performance is generally referred to as tracking error, and sampling tends to introduce more of it than full replication.
- Why it happens. Because a sampled portfolio doesn’t hold every constituent in exact proportion, its returns can diverge somewhat from the index, especially during periods when the excluded or underweighted securities behave very differently from the sample.
- How it’s managed. Fund managers using sampling typically monitor and adjust the sample over time to try to keep tracking error within an acceptable range, though some divergence from the benchmark is generally unavoidable with this method.
What this means when comparing funds
Two funds tracking the same index can produce slightly different returns over time depending on whether they use full replication or sampling, along with other factors like a fund’s expense ratio and trading costs. Comparing a fund’s actual historical tracking record against its benchmark, rather than just assuming any fund tracking a given index will perform identically to it, can reveal how well a particular fund executes its stated tracking approach.
The bottom line
Full replication and sampling are two different tools for the same underlying goal: mirroring an index as closely as practical. Neither guarantees a perfect match to the benchmark, and understanding which approach a fund uses helps explain why its returns might diverge slightly, even modestly, from the index it’s designed to track.