Portfolio Management Pathway · Index-Based Equity Strategies

Indexing Methods: Full Replication, Sampling and Optimization

Updated 8 October 2026 · Fact-checked

Indexing methods are the ways a manager builds a portfolio to track an index. Full replication holds every constituent at index weight. Stratified sampling holds a subset matched on key characteristics. Optimization uses a risk model to minimize expected tracking error. Choose based on index size, liquidity, costs and tracking tolerance.

Understand Indexing Methods: Full Replication, Sampling, Optimization

An index fund tries to match the return of a benchmark index. The manager must decide which securities to hold and in what amounts. There are three main methods. They differ in how closely the portfolio copies the index and what that costs.

Full replication holds every security in the index at its index weight. It gives the lowest tracking error before costs, because the portfolio is the index. It works best when the index is made of fewer, large, liquid stocks. It is costly for broad indexes with hundreds or thousands of names, because many small positions are expensive to buy and rebalance. Some constituents may be illiquid, so the manager pays wide bid-ask spreads. Small positions also make fixed trading costs large relative to the position.

Stratified sampling divides the index into cells (strata) using characteristics such as industry, country, size, style or other risk factors. The manager then buys a representative sample of securities from each cell so the portfolio's weight in each cell matches the index. It holds fewer securities than the index, so costs are lower and illiquid names can be avoided. The trade-off is higher tracking error, because the sample will not match the index exactly. The match is only as good as the cells chosen. Characteristics not used to build the cells can differ from the index.

Optimization uses a quantitative risk model. The manager chooses holdings and weights that minimize expected tracking error against the index, subject to constraints such as number of holdings and trading costs. It uses historical relationships among securities (factor exposures, correlations). It can hold fewer securities than sampling by hand and can handle many factors at once. The trade-offs are model dependence and the use of past data. If the relationships change, tracking error can be higher than the model predicted. The method also needs specialist inputs and may produce holdings that are hard to explain.

The choice is a trade-off between tracking error and cost. Large, liquid, concentrated indexes favor full replication. Broad, less liquid indexes favor sampling or optimization. Always link your choice to the facts in the question, such as the number of index constituents, liquidity, fund size and the client's tolerance for tracking error.

Key rules to remember

Tracking error (ex post): Tracking error = standard deviation of (portfolio return − index return) over time; Also called tracking risk. Measures how closely the fund follows the index. Lower is better for an index fund.
Tracking error drivers: Portfolio return − index return = (cost drag) + (sampling/holding differences) + (cash drag and timing) + (securities lending and other income); A qualitative breakdown, not a numeric rule. Use it to explain why a fund's return differs from its index.
Method selection rule: Fewer, liquid constituents → full replication; many or illiquid constituents → sampling or optimization; A guide, not an absolute. Full replication gives the lowest tracking error but the highest trading and holding cost for broad indexes.

How to solve Indexing Methods: Full Replication, Sampling, Optimization questions

Use this sequence for any question that asks you to choose, compare or critique an indexing method.

1Read the index facts: number of constituents, liquidity of the smallest names, and how often the index changes.
2Read the fund facts: size of assets, cost sensitivity and the client's tolerance for tracking error.
3Identify the method that fits: full replication, stratified sampling or optimization.
4State how the method works in one short sentence, using the command word asked (identify, explain, justify).
5Give the main advantage and the main disadvantage tied to the case facts, for example cost versus tracking error.
6Name the extra risk if relevant: cell design for sampling, or model and historical-data risk for optimization.
7Finish with a clear recommendation or conclusion that answers the question asked.

Quickest way: Index size and liquidity test

When to use it: Use for multiple-choice items and short essay parts that ask which method is most appropriate.

Few large liquid stocks: pick full replication.
Hundreds or thousands of names, or illiquid ones: pick sampling or optimization.
Question mentions risk factors, a risk model or minimizing expected tracking error: pick optimization.
Question mentions cells, industries, sectors or matching characteristics: pick stratified sampling.
Check the stated trade-off: lower cost brings higher tracking error.

Common mistakes in Indexing Methods: Full Replication, Sampling, Optimization

Saying full replication always has the lowest total cost.
Students remember that it has the lowest tracking error before costs and assume it is best on every measure.
Fix: Say it has the lowest tracking error from holdings, but high trading and rebalancing costs for broad or illiquid indexes. Costs can raise its net tracking error.
Confusing stratified sampling with optimization.
Both hold a subset of the index.
Fix: Sampling builds cells and picks securities by hand to match cell weights. Optimization uses a risk model to minimize expected tracking error mathematically.
Claiming sampling or optimization removes tracking error.
The methods are described as matching the index's risk.
Fix: Both accept some tracking error in return for lower cost. Optimization depends on historical data that may not hold in future.
Choosing a method without using the case facts.
Students write general pros and cons from memory.
Fix: Quote the index size, liquidity or fund size from the vignette and link it to your choice.
Ignoring the cells in sampling.
Students focus on holding fewer securities.
Fix: Remember that tracking is only good for the characteristics used to form the cells. Other characteristics may differ from the index.

Worked examples

Example 1

A manager runs a fund tracking a global equity index with about 3,000 constituents. Many small-cap constituents trade infrequently with wide bid-ask spreads. The client wants low cost and accepts a small amount of tracking error. Identify the most appropriate indexing method and justify it in two points.

Show the solution

Index facts: about 3,000 constituents and many illiquid small caps.
Full replication would need thousands of small positions and expensive trading in illiquid names, so cost would be high.
Client accepts small tracking error, so a method that holds fewer securities is suitable.
Sampling or optimization fits. Optimization can use a risk model to hold a smaller set of liquid securities while matching factor exposures.

Answer: Use optimization (stratified sampling is also acceptable). Justification: (1) holding a subset avoids the high trading costs and wide spreads of the illiquid small caps; (2) the client tolerates a small tracking error, which is the trade-off for lower cost. Full replication is not suitable because of cost and illiquidity.

Example 2

An index has 40 large, highly liquid stocks. The fund is large and the client wants the tightest possible tracking. Which method should the manager use, and what is the main disadvantage the manager should monitor?

Show the solution

Index facts: only 40 constituents, all liquid.
Client priority: tightest tracking.
Full replication holds all 40 at index weights, so tracking error from holdings is minimal.
Holding all 40 is cheap and practical because the names are liquid and few.
The remaining disadvantage is that the fund must trade whenever the index changes, and transaction costs and cash drag still cause some tracking error.

Answer: Use full replication. It matches the index exactly in holdings and suits a small, liquid index. The main disadvantage to monitor is transaction cost and cash drag when rebalancing for index changes, which still create some tracking error.

Exam tips

Link the method to the vignette facts: number of constituents, liquidity, fund size and tracking tolerance.
When the command word is justify, give the method plus one benefit and one cost, not a long list.
Remember the order of tracking error versus cost: full replication lowest tracking error, sampling and optimization lower cost.
For optimization, mention reliance on historical data and a risk model as the key weakness.
For sampling, mention that the cells chosen decide how well the portfolio tracks.

Indexing Methods: Full Replication, Sampling, Optimization: frequently asked questions

What is the difference between full replication and stratified sampling?

Full replication holds every index security at its index weight. Stratified sampling splits the index into cells by characteristics and holds a representative subset in each cell so cell weights match the index. Sampling costs less but has higher tracking error.

When is optimization better than stratified sampling?

Optimization is useful when many risk factors matter and a model can capture them. It minimizes expected tracking error with fewer holdings. Its weakness is dependence on historical relationships that may change.

Does full replication always have zero tracking error?

No. Even if holdings match, costs, cash drag, trading delays at index changes and fees cause the fund's return to differ from the index.

How do I choose an indexing method in an exam answer?

Use the case facts. Few liquid constituents point to full replication. Many or illiquid constituents point to sampling or optimization. State the cost versus tracking error trade-off in your answer.