Skip to main content
ETF mechanics deep dive

Synthetic vs physical ETF replication

An ETF can track an index by holding the underlying shares (physical) or by signing a contract with a bank that promises the index return (synthetic). Both achieve similar headline outcomes. The trade-off is counterparty risk against tracking efficiency and tax. Here's how each actually works, why the industry shifted to physical after 2008, and where synthetic still wins.

Educational only. Both structures are legal UCITS structures with regulator-imposed limits. The choice between them is one input in your ETF selection, not the whole story.

Three replication flavours UK retail will encounter

  1. Full physical replication — the fund holds every constituent of the index in roughly index weights. Used for funds tracking liquid indices with manageable membership (FTSE 100, S&P 500, MSCI World).
  2. Sampled physical replication — the fund holds a representative subset of the index members. Used for indices with thousands of constituents (FTSE All-World ~4,000 stocks, MSCI ACWI ~3,000) where holding every name is uneconomic.
  3. Synthetic (swap-based) replication — the fund holds a basket of unrelated securities and enters into a "total return swap" with a counterparty bank. The bank pays the fund the index return in exchange for the return on the basket. The fund effectively gets the index return without owning the index constituents directly.

You'll also see "unfunded swap" and "funded swap" variants of synthetic, and "physical with optimisation" or "stratified sampling" variants of physical. These are nuances on the three main flavours.

How synthetic replication actually works

A simplified walkthrough of a synthetic S&P 500 ETF:

  1. Investor buys £1m of the ETF. Cash arrives at the fund.
  2. The fund buys a "substitute basket" of, say, European large-cap stocks worth £1m. This basket is the fund's actual physical assets. It bears no relation to the S&P 500.
  3. The fund enters into a total return swap with, say, Société Générale. The terms: the fund pays SG the total return of the European basket; SG pays the fund the total return of the S&P 500.
  4. Daily net settlement: if the S&P went up 1% and the basket went up 0.7%, SG pays the fund 0.3% × £1m = £3,000. If both went the other way, the fund pays SG.
  5. NAV is calculated daily using the S&P 500 return (which the swap delivers), so the fund's NAV moves with the S&P even though the underlying assets are different.

From the investor's perspective: the ETF tracks the S&P 500. From the fund's perspective: it owns European stocks and has a contractual claim on SG for the difference.

Counterparty risk — the synthetic-specific issue

The investor's exposure to SG's creditworthiness is real but bounded. UCITS rules cap the "swap exposure" at 10% of the fund's NAV per counterparty. So at any given moment, the most the fund can be exposed to SG via the swap is 10% of fund value — about £100k on a £1m fund.

How the 10% cap works in practice: the swap is rebalanced ("reset") periodically — typically when the swap exposure approaches the cap, the counterparties square up via a cash payment. Between resets, exposure can range from 0% (just after a reset) to 10% (just before the next reset). If the counterparty fails between resets, the fund's recovery is limited to what's pledged as collateral.

For UCITS synthetic ETFs the collateral is typically held by a trustee at a triple-A-rated custodian, and the collateral basket is required to be diversified. In a failure scenario, the fund would seize the collateral and have a claim against the counterparty's estate for the rest.

The worst-case retail loss from a synthetic ETF counterparty failure is usually estimated at 1–3% of fund value, not 100%. This is meaningfully different from "you could lose everything", which is the common misconception.

Synthetic's structural advantages

Despite the counterparty risk, synthetic has real upsides:

The WHT advantage was historically big enough that synthetic S&P 500 ETFs (db x-trackers, Lyxor in the early 2010s) had tracking differences 0.15–0.25% better than physical equivalents. Some of that gap has narrowed as physical funds have improved their WHT optimisation.

The 2008–2011 industry shift

European ETFs were predominantly synthetic in 2008. After the financial crisis, two factors drove a shift to physical:

By 2026, the major European ETF providers (iShares, Vanguard) are mostly physical. Synthetic remains a meaningful minority — Invesco's S&P 500 swap-based ETF (ticker SPXS, OCF 0.05%) is a frequently-mentioned example, and Xtrackers retains synthetic share classes for some indices.

When each makes sense for UK retail in 2026/27

Index / asset class Recommended replication Reasoning
S&P 500Physical (default) or synthetic if hunting trackingDefault to physical for simplicity (VUSA / CSPX). Synthetic Invesco SPXS offers tighter tracking due to WHT optimisation if you accept the counterparty risk.
FTSE 100 / FTSE All-SharePhysicalNo WHT advantage; physical replication is cheap and clean
MSCI World / FTSE All-WorldSampled physicalStandard. Synthetic alternatives exist (Xtrackers) but most retail picks physical.
Emerging marketsSampled physical or syntheticSynthetic historically had tracking advantage; physical has caught up. Pick on TD numbers, not on structure.
Commodities (Gold, oil)Physical (gold) or synthetic (oil)Gold ETCs hold real bars (IGLN, SGLN). Oil and most non-precious metals are synthetic because storing barrels is impractical.
Single-country EM (Vietnam, Saudi etc.)Often syntheticAccess restrictions can make physical replication impossible or very expensive

How to check the replication method on any ETF

  1. Fund factsheet: usually states "Replication method: Physical / Synthetic / Optimised" on the front page
  2. Fund KIID: required disclosure for UCITS, found in the "investment policy" section
  3. justETF.com filter: lets you filter by replication method directly
  4. The fund manager's investor info page: usually has a dropdown showing replication, sometimes with the swap counterparty name(s)

Frequently asked questions

Is synthetic dangerous for retail investors?

"Dangerous" is too strong. The UCITS counterparty cap, the collateral requirements, and the diversified counterparty arrangements mean the worst-case loss is bounded at low single-digit percentages of fund value, even in a counterparty failure. The relevant question is whether the tracking advantage is worth the additional complexity and tail risk — for most retail buyers picking between similar funds, physical is the simpler default.

Can a physical ETF have any counterparty risk?

Yes, through securities lending. Most physical ETFs lend portfolio holdings overnight to short-sellers in exchange for fee income. The borrower pledges collateral (typically over 100%). If the borrower defaults and the collateral falls short, the fund takes a small loss. This counterparty risk is usually smaller than swap counterparty risk but it's not zero.

What's an "unfunded" vs "funded" swap?

Unfunded swap: the fund holds the substitute basket and exchanges only the performance via the swap (the more common European structure). Funded swap: the fund gives cash to the counterparty in exchange for a delivery promise (more common in US synthetic structures). Unfunded has lower counterparty exposure on average.

Do synthetic ETFs have Reporting Fund status?

Most popular synthetic UCITS ETFs marketed to UK retail (Invesco SPXS, Xtrackers etc.) have UK Reporting Fund status. Always verify on the specific ISIN before buying — see our Reporting Fund status guide.

Are leveraged and inverse ETFs synthetic?

Yes, almost always. Daily-resetting 2x or 3x leveraged ETFs can only be built via swaps or futures — physical replication of leverage isn't practical. This is also why leveraged ETFs are usually a poor long-term holding (compounding decay), regardless of the replication method.

Editorial accountability
Open Trust Centre →

Every page is reviewed against the editorial standards, written from primary sources, sourced openly, and corrected publicly. No affiliate revenue. No sponsored content. No paid placements.

Editorial standards Editorial process Corrections policy How we make money Editorial team Methodology
Cookie settings