ETFReplicationBasics

Physical vs synthetic replication: how ETFs track an index

person Axiomly Team
•
calendar_today October 8, 2026
•
schedule 5 min read

bolt Key Takeaways

  • check Physical replication: the ETF actually buys and holds the shares in the index
  • check With optimised sampling, the ETF buys only the largest and most liquid shares in an index to keep costs down
  • check Synthetic replication tracks the index through a swap agreement with a bank or other financial counterparty
  • check In a UCITS ETF, a depositary holds the fund's assets separately from the management company's own assets, whatever the replication method. That does not remove the counterparty risk of a swap
  • check For UK investors, an overseas fund such as an EU-domiciled UCITS ETF has to be recognised by the FCA to be promoted to retail investors, and HMRC reporting fund status is a separate question that applies to either method

An ETF can track an index in two ways: it buys the shares itself (physical replication) or it uses a swap agreement (synthetic replication). The difference looks small, but it decides what sits inside the fund and which extra risk comes with it.

Physical replication: the ETF holds what the index holds

With physical replication, the ETF buys the shares that make up the index and holds them in the fund. When a company's weight in the index rises, its weight in the ETF portfolio rises with it. The ETF owns what it tracks.

There are two variants. With full replication, the ETF buys every share in the index at its exact weighting. That works for indices with a manageable number of constituents, such as the FTSE 100, which covers 100 of the largest companies listed on the London Stock Exchange that pass FTSE Russell's size and liquidity screens.

With optimised sampling, the ETF buys only a representative selection of the largest and most liquid shares. This applies to very broad indices with a very large number of constituents, such as global indices like the MSCI World. Buying every tiny position would push up dealing costs without noticeably improving accuracy. The aim is a return close to the full index at a lower cost.

Synthetic replication: the swap delivers the index return

With synthetic (or swap-based) replication, the ETF does not buy the index shares directly. The fund manager agrees a swap with a financial counterparty, usually a bank. The counterparty commits to pay the ETF the return of the index. In exchange, the ETF holds its own portfolio of securities, which need not match the index shares but has to be diversified and liquid.

The swap is adjusted regularly so that the securities portfolio and the swap together follow the index. Under European UCITS rules, exposure to a swap counterparty may not exceed 10% of fund assets if the counterparty is a credit institution, and 5% otherwise. The cap applies to the exposure after collateral received, not to the value of the swap itself. Most of the fund's assets are the securities portfolio it holds itself.

Is synthetic replication riskier?

Synthetic ETFs add one risk: counterparty risk. If the swap provider fails to pay, the fund has to work out how to cover the gap. The legal cap limits this risk, and in practice swap providers are usually large, international banks.

Physically replicating ETFs are not automatically risk-free. Many lend out part of their shares for a fee (securities lending). That is a separate risk, usually covered by collateral, and it is similar in structure to swap counterparty risk.

For a UCITS ETF, whatever the method, a depositary holds the fund's assets separately from the management company's own assets. That does not protect against the counterparty risk of a swap.

What this means for UK investors

UK investors have two separate checks to make: whether the fund is recognised by the FCA, and whether it has HMRC reporting fund status. Most ETFs that UK investors buy on the London Stock Exchange are UCITS funds, and many of them are domiciled in the EU, often in Ireland. Since Brexit, the FCA says a fund established outside the UK must be recognised by the FCA before it can be promoted to UK retail investors. EEA UCITS can apply for that recognition under the Overseas Funds Regime, and some funds still rely on temporary permissions. The FCA explains these routes on its website.

Tax is a separate question. For UK investors, an overseas ETF falls under HMRC's offshore funds rules, which distinguish between reporting and non-reporting funds. The status belongs to the fund, not to its replication method, so a physical and a synthetic ETF can each have it or lack it. HMRC publishes a list of approved reporting funds and updates it monthly. Check a fund's status before you buy. Tax treatment depends on your own situation, and this article is not tax advice.

How much the choice of method matters

For most private investors, the replication method plays a minor role with well-known, broadly diversified indices. Both approaches have worked in practice and are used by large, established fund providers. If you prefer maximum simplicity and transparency, physical replication is the more common pick.

Whether one method is cheaper depends on the individual product. The ongoing charges figure and the tracking difference are in the factsheet. For most investors, it matters more that an ETF is set up under the UCITS framework than how it tracks its index technically.

The replication method does not change your position size

Both methods follow the same index return. What differs is the side risk: swap counterparty risk on one side, securities lending on the other. Neither answers how large a single position in your portfolio should be. Work that out with the free Axiomly position size calculator, whatever replication method your ETF uses.

help_outline Frequently Asked Questions

Is synthetic replication riskier than physical?

Synthetic ETFs carry an added risk: counterparty risk against the swap provider. UCITS rules cap it: exposure to a single swap counterparty may not exceed 10% of fund assets if the counterparty is a credit institution, and 5% otherwise. The limit applies to the exposure left after collateral, not to the value of the swap itself. In practice swap providers are usually large, established banks. Physically replicating ETFs are not risk-free either. Many lend out part of their shares, which carries its own risk, usually covered by collateral.

Why doesn't an ETF just buy every share in its index?

For very broad indices with a very large number of constituents, buying every position would be costly and would push up dealing fees. With optimised sampling, the ETF buys a representative selection of the largest and most liquid shares, which aims to follow the full index closely.

How do I know which replication method an ETF uses?

Check the fund's factsheet, key information document or prospectus. They usually describe the method with terms such as "physical replication", "optimised sampling" or "synthetic" or "swap-based replication". Many broker and comparison sites show it in the product overview too.

Does the replication method affect how a UK investor is taxed?

As a general rule, not directly. HMRC's offshore funds rules look at whether a fund has reporting fund status, which is listed on GOV.UK and updated monthly. For a fund with that status, a gain on disposal is generally taxed as a capital gain, and the investor reports their share of the fund's income even if it is not paid out. For a non-reporting fund, gains on disposal are generally taxed as income. Your own circumstances matter, so check HMRC's guidance or ask a tax adviser.