Gold ETFs: The Complete Guide

A generation underwater

In January 1980, gold hit roughly $850 an ounce. By April 2001, it was trading around $255. An investor who bought at the peak and held for 21 years lost about two-thirds of their money in nominal terms, and considerably more after inflation. Equities over the same window roughly tenfolded.

That stretch is the part of gold's history that almost no "best gold ETF" listicle mentions. It matters because gold doesn't behave like a stock or a bond. There are no earnings and no cash flow. The price is set entirely by what someone else will pay for it, which can drift in one direction for a very long time.

So why do roughly $400 billion in physical-gold ETF assets sit on global exchanges in 2026? Because gold does specific jobs that other assets don't, in specific moments. The trick is being honest about which moments those are. The wrapper choice and position sizing flow from there.

This guide walks through the gold ETFs that matter. Physical bullion trusts, mining stock funds, plus the UCITS equivalents that European and UK investors actually buy. How they differ in practice, and where a metal sleeve belongs in a real portfolio.

What gold actually does in a portfolio

The cleanest academic framing comes from Baur and Lucey's 2010 paper Is Gold a Hedge or a Safe Haven? (Financial Review 45). They distinguish three roles an asset can play:

• A diversifier has low average correlation to stocks but moves with them in normal markets.
• A hedge is uncorrelated or negatively correlated on average.
• A safe haven is uncorrelated or negatively correlated specifically during extreme stock-market stress.

Their finding: gold is both a hedge against equities on average and a safe haven during extreme drawdowns. With one important caveat. The safe-haven property only persists for roughly 15 trading days after a shock. After that, gold's behavior reverts to its long-run average, which is mildly diversifying but not protective.

The headline cases support the framing. In 2008, the S&P 500 fell about 37% while gold rose about 6% on the year. The COVID episode was less clean: gold initially sold off with everything else in March 2020 as investors raised cash, then recovered to finish 2020 up roughly 25%. By 2022, with US inflation at a 40-year high, gold was roughly flat in dollar terms because real interest rates rose sharply at the same time. The "inflation hedge" story is true over centuries (Erb and Harvey, The Golden Dilemma, 2013) but unreliable over investor-relevant horizons of five to twenty years.

The takeaway: gold's diversification is real but moderate, and it tends to help against panic more than it helps against the slow grind of rising real rates.

Physical gold ETFs: the wrapper that holds metal

A physical gold ETF behaves more like a warehouse receipt than a fund. It's a grantor trust holding bars of bullion in a vault on behalf of shareholders. Each share represents a fractional ounce. There's no manager picking anything; the trust's only activity is storage and the occasional redemption by an authorized participant. Individual shareholders sell on exchange like any other ETF rather than redeeming for bullion. Total return comes entirely from the gold price minus the expense ratio.

Six funds, broadly the same exposure. What separates them is cost, plus liquidity at scale and (for some buyers) the choice of vault location.

GLD vs IAU vs GLDM: when each one wins

The 30-basis-point gap between GLD at 0.40% and GLDM at 0.10% looks small. Compounded on $50,000 held for 20 years, with gold returning a hypothetical 5% per year, the cost difference is roughly $7,000.

For a buy-and-hold position, GLDM (or IAUM at 0.09%) is the cheaper default. The metal and the custodian network match GLD's, with a fee that's an order of magnitude lower.

GLD still has reasons to exist. It's the most liquid gold ETF in the world, with a deep options market and tight spreads on large block trades. For an institution rolling a multi-million-dollar position weekly, the spread savings and options availability outweigh the higher TER. For an individual buying once and holding, they don't.

IAU sits in the middle. The fee is lower than GLD's, the AUM is larger than GLDM's, and the track record runs slightly longer than the minis. A defensible choice for someone who wants liquidity without paying GLD's premium and doesn't mind giving up 16 basis points on the mini.

SGOL exists for a narrower reason. The metal is vaulted in Zurich rather than London or New York, which appeals to investors who want geographic separation between their gold and the financial centers their other assets sit in. The case for it is sovereign-risk diversification rather than better returns.

Gold miners are not gold

Miners are equities of companies that dig gold out of the ground. Their share prices are influenced by the gold price, but they're also influenced by labor costs, energy prices, country risk in jurisdictions like South Africa or Burkina Faso, management quality, and balance-sheet leverage. The link to spot gold is indirect.

Two funds dominate the category:

Over the past decade, GDX has delivered cumulative returns roughly comparable to spot gold, but the path was nothing alike. GDX experienced about double the volatility of GLD and a peak-to-trough drawdown near 43%, against roughly 20% for the metal itself (Quantpedia, Gold's Rally and the Gold Mining Stocks Trap, 2024). Same destination, much rougher ride.

The mental model: miners are a leveraged option on gold combined with an equity bet on the mining industry. When gold rises sharply, miners often rise faster, because their cost of production is fixed in the short run and incremental revenue drops to the bottom line. When gold falls or moves sideways, miners can underperform sharply, because operational and financing costs don't fall with the price.

GDXJ extends this to small-cap and exploration-stage miners. Beta to gold runs higher and historical drawdowns are deeper. The fee matches GDX at 0.51%.

If the goal is gold price exposure, a physical trust delivers it more cleanly at a quarter of the fee. Miners belong in a portfolio for a different reason: a deliberate equity bet on the mining sector, sized like any other thematic exposure. The same confusion shows up with structurally flagged ETFs like covered calls and buffer products, where the wrapper's mechanics quietly differ from what the marketing implies.

Tax: the one place geography matters

In the United States, physical gold ETFs structured as grantor trusts (GLD, IAU, GLDM, IAUM, SGOL, BAR) are taxed as collectibles. The maximum federal long-term capital gains rate on collectibles is 28%, against 20% for stocks held more than a year. The rule applies regardless of how the metal is wrapped. A $50,000 gain on GLDM held for five years in a US taxable account can owe up to 28% federal tax, plus state tax where applicable. The SPDR GLD tax FAQ on ssga.com walks through the mechanics in detail.

Gold mining ETFs are equity funds and don't trigger the collectibles rule. GDX and GDXJ are taxed as ordinary equities at the long-term capital gains rate.

In Europe and the United Kingdom, none of this applies. Most non-US gold products are structured as exchange-traded commodities, or ETCs, rather than ETFs. They're treated as ordinary investment products under local capital gains rules, and many are eligible for tax-advantaged wrappers like the UK Stocks and Shares ISA. The result is that for the same metal exposure, after-tax outcomes can diverge sharply by jurisdiction.

In few corners of the ETF world does the right pick depend so heavily on which passport the buyer holds and which account the position will sit in.

Non-US investors: the UCITS-and-ETC family

UCITS rules forbid single-commodity funds, so European gold products are structured as ETCs (debt securities backed by physical metal). For practical purposes they behave like US grantor trusts: each note represents a fractional ounce, no income, total return tracks gold minus the fee.

At 0.12%, SGLN and SGLD are slightly cheaper than even the lowest-cost US grantor trust. They list on multiple European exchanges in EUR, GBP, and CHF share classes, which removes the need to convert into USD before buying. PHAU is the older, pricier WisdomTree product; WGLD is the cheaper successor and almost always the better pick of the two.

Picking a UCITS ETC also avoids US estate-tax exposure on US-domiciled holdings, which can apply above $60,000 in US-situs assets for non-resident aliens. A separate point from day-to-day tax treatment, and worth checking with an adviser if the position is meaningful.

Where gold fits, and where it doesn't

Most allocation studies put a sensible gold sleeve at 5% to 10% of a diversified portfolio. Below 5% the contribution to portfolio behavior is too small to matter; above 10%, the lack of yield drags on long-run expected return. Ray Dalio's All-Weather uses 7.5%; Harry Browne's Permanent Portfolio uses 25% and accepts a lower expected return in exchange for stability.

Where adding gold has historically made sense. A portfolio heavy in stocks (especially in one country or sector) gets a moderate diversification benefit from a small gold sleeve, mostly visible in extreme drawdowns. The same logic that makes international diversification worthwhile applies to non-equity diversifiers. The argument strengthens when sovereign or currency anxiety is part of the picture: gold has no counterparty, which is a feature with no cheap substitute for investors in jurisdictions with weak currencies or capital controls. SGOL or a Swiss-vaulted UCITS ETC compounds the geographic separation. The same logic puts gold alongside cash equivalents and low-volatility equity in a portfolio whose explicit job is "lose less in the next five years."

Where it's the wrong tool. Long accumulation horizons with limited capital lose out: someone with 30+ years to compound is better served by maximum equity exposure, since gold's expected real return is roughly zero over multi-decade windows against roughly 5% for global equities. The opposite error is treating gold as a fixed income substitute. Bonds offer income and rate-sensitivity; gold offers neither, and replacing a bond sleeve with gold removes the part of the portfolio designed to appreciate when growth slows.

A different question for each wrapper

The decision flattens out once the role is clear. For buy-and-hold gold price exposure, GLDM or IAUM at roughly 0.10% wins on cost in a US account; for the same job in a European account, SGLN or SGLD at 0.12% wins on cost and tax. Geographic vaulting separation pushes the choice toward SGOL at a small premium for Zurich. Active trading with an options overlay is GLD's case: the higher fee buys depth and tight spreads. The full gold ETF universe is filterable by structure and expense ratio.

Miners are a separate decision: a deliberate equity bet on the mining industry, with gold-price beta as the headline factor and operational leverage as the volatility cost. They earn a portfolio slot only when the equity bet itself is what's wanted.

Central banks bought roughly 1,000 tonnes of gold in each of the past three years (World Gold Council, Gold Demand Trends 2025), a structural shift from the past two decades and one likely reason gold pushed through $4,000 an ounce in late 2025. None of that changes the long-run case. Gold still throws off no income, and within living memory it spent a generation underwater. The case for owning some isn't that those facts stopped being true. It's that when they don't matter, gold does something no other asset does.