How crypto ETFs actually work — spot, futures, and why the difference matters
Spot Bitcoin ETFs hold actual BTC in custody. Futures ETFs hold rolling CME contracts. The basis between them costs investors 4-8% a year. Here is how the plumbing works.
On 10 January 2024, the US Securities and Exchange Commission approved eleven spot Bitcoin ETFs in a single order, ending a decade-long standoff that began with the first Winklevoss filing in 2013. Twenty-three months later, those funds collectively hold roughly $115 billion in assets, with BlackRock’s IBIT alone above $60 billion — the fastest ETF launch in the product’s 33-year history. A spot Ether complex followed in July 2024, and Hong Kong, Brazil, and Canada now run parallel regimes. But “crypto ETF” is doing a lot of work in that sentence: a spot product and a futures product look identical on a brokerage screen and behave nothing alike in a portfolio.
What is at stake is the difference between owning Bitcoin and owning a wrapper that pays 4-8% per year in roll cost to approximate it. The first futures-based BTC ETF, ProShares BITO, launched in October 2021. Over the following two years it underperformed spot BTC by roughly 18 percentage points on a cumulative basis — not because the manager did anything wrong, but because the structure of a CME futures roll guarantees a drag in contango markets. Investors who held BITO thinking it was “Bitcoin exposure” found out the hard way. The same trap exists today in Ether futures ETFs, in any leveraged crypto product, and in most of the international vehicles still trading without spot approval. The structure is the product.
The two structures, mechanically
A spot ETF is a 1940-Act-style wrapper (technically a grantor trust for the US Bitcoin products) that holds the underlying asset itself. When an authorised participant — typically a market maker like Jane Street, Cantor, or Virtu — creates new shares, they deliver cash to the issuer, who instructs the custodian (Coinbase Custody for most US issuers, BNY Mellon for some) to acquire spot BTC. The shares track NAV minus expenses because the trust literally owns the coin. Redemptions reverse the process. A futures ETF, by contrast, holds front-month CME Bitcoin or Ether futures contracts and rolls them forward as expiry approaches, typically over a five-day window. The fund never touches the underlying coin; it tracks an index of futures prices, not the spot index.
The roll cost, in basis points
Crypto futures have spent the bulk of their history in contango — the back month trades above the front month — because the natural carry on BTC is positive (you can lend spot and earn interest, so the futures price must reflect that). When a futures ETF rolls from the expiring contract to the next, it is mechanically selling low and buying high. The annualised basis on CME BTC futures has averaged 8-12% during bull regimes and 2-4% in flat markets, occasionally inverting (backwardation) during sharp drawdowns. That basis, minus management fees, is the structural drag on any futures-based product. CME Group publishes the term structure daily.
| Fund | Ticker | Type | Expense ratio | AUM (approx.) |
|---|---|---|---|---|
| iShares Bitcoin Trust | IBIT | Spot BTC | 0.25% (0.12% waiver) | $62B |
| Fidelity Wise Origin Bitcoin Fund | FBTC | Spot BTC | 0.25% | $19B |
| Bitwise Bitcoin ETF | BITB | Spot BTC | 0.20% | $4B |
| ARK 21Shares Bitcoin | ARKB | Spot BTC | 0.21% | $3B |
| Grayscale Bitcoin Trust | GBTC | Spot BTC (converted) | 1.50% | $20B |
| ProShares Bitcoin Strategy | BITO | BTC Futures | 0.95% | $2.5B |
| iShares Ethereum Trust | ETHA | Spot ETH | 0.25% (0.12% waiver) | $5B |
Why spot took eleven years and futures took six months
The SEC’s stated objection to spot products from 2017 to 2023 was market manipulation: it argued the underlying spot exchanges (Coinbase, Kraken, Binance, etc.) were not subject to surveillance-sharing agreements of the kind the CME provides. Futures products, regulated under the 1940 Investment Company Act with the CME as the reference market, sidestepped that objection. The wall fell in Grayscale v. SEC (August 2023), when the DC Circuit ruled the SEC’s distinction between spot and futures products was “arbitrary and capricious” because both ultimately track the same underlying price. The agency had ninety days to either appeal or capitulate. It capitulated.
Tracking error: where the rubber meets the road
For spot products, tracking error against the underlying CF Benchmarks BRRNY index is generally under 30 basis points annualised — the difference is essentially the expense ratio plus minor execution slippage. For futures products, tracking error against spot is dominated by roll cost and can exceed 800 basis points in steeply contangoed markets. The other dimension is intraday: spot ETFs trade at NAV plus or minus a few bps thanks to tight AP arbitrage, while thinly-traded futures products can dislocate 50-100 bps during volatile sessions. If you are sizing a position that you intend to hold beyond a year, the structural drag of futures dwarfs every other cost. If you are running a short-term tactical trade, intraday liquidity and bid-ask spread matter more. Either way, read the prospectus.
The custody question
Roughly 90% of US spot Bitcoin ETF AUM is custodied at Coinbase Custody Trust, a New York-chartered trust company. That is a single point of failure most institutional allocators have, by now, written into their risk memos. BlackRock added BNY Mellon as a secondary custodian for IBIT in late 2025, partly to address that concentration. The custodian holds private keys in deep cold storage, with insurance via Lloyd’s syndicates; segregation of customer assets is bankruptcy-remote under New York banking law, in theory. Whether that holds up in a hypothetical Coinbase insolvency has not been tested in court. For Ether products, staking is currently prohibited by the SEC’s approval order — meaning ETH-ETF holders forgo the roughly 3-4% native staking yield that direct holders earn. That gap will likely close in 2026-27 as the agency revisits the rule.
International parallels worth knowing
- Canada: Spot BTC ETFs (BTCC, EBIT) live since February 2021, two years ahead of the US. Tracking error has been textbook.
- Hong Kong: Spot BTC and ETH ETFs approved April 2024 under the SFC framework, with in-kind creation/redemption — which the US currently disallows for tax-structuring reasons.
- Brazil: B3 lists multiple crypto ETFs including baskets; QBTC11 is the bellwether.
- Europe: ETPs (debt instruments, not ETFs) dominate via 21Shares, CoinShares, and ETC Group; MiCA may open the door to true UCITS-eligible crypto ETFs in 2026-27.
- UK: The FCA permitted crypto ETNs for professional investors in March 2024 and is reviewing retail access.
The fragmentation matters for arbitrage. Hong Kong spot ETFs trade during Asian hours, US spot ETFs during US hours, and the underlying BTC trades 24/7 — which means the NAV gap between the closing print of one venue and the open of the next is regularly bridged by APs running cross-venue books. The result is tighter average tracking over a 24-hour cycle than any single venue achieves in isolation. For non-US allocators, that fragmentation also means meaningfully different fee schedules: the cheapest UCITS-eligible BTC ETP currently runs around 19 bps versus 25 bps on US spot products, and Canadian spot products cluster around 30-40 bps. Tax wrapper, not headline TER, usually drives the final choice — talk to your accountant before you talk to your broker. Forward listings and approvals are tracked on our events calendar.
How to read flow data without fooling yourself
Daily ETF flow numbers are the most over-interpreted data series in crypto. A $200m net inflow day in IBIT does not mean $200m of new BTC demand; it means net creations exceeded net redemptions by that amount, after authorised participants have already hedged in spot or futures markets. The true marginal demand signal is the cumulative trailing 30-day flow, not the daily print, and even that is muddied by basis-trade arbitrage where institutions are long ETF and short CME futures to harvest the basis spread. Farside Investors publishes the cleanest aggregator. Cross-reference with CME futures open interest before drawing conclusions; if both rise together, you are watching the basis trade expand, not directional buying. We aggregate the relevant flows on our market dashboard.
Leveraged and inverse: the daily-rebalance trap
2x and -1x crypto ETFs exist (BITX from Volatility Shares, BITI for inverse, ETHU for leveraged Ether) and they are uniformly worse than retail investors understand. Daily-rebalanced leverage products achieve their stated multiplier only over a single trading day; over any longer horizon, volatility decay eats the returns. The arithmetic is unforgiving: an underlying that swings -10% then +11.1% returns to its starting value, but a 2x daily product would print -20% then +22.2%, ending at 97.76% of where it started — a 2.24% drag from a single round-trip move. Compound that across the 220 trading days in a year and 2x products can underperform a static 2x position by 15-25 percentage points in flat-but-volatile regimes. For crypto, which prints multiple 5%+ days per month, the drag is structural and large. The use case is intraday tactical exposure only; everything else is a tax on misunderstanding.
In-kind vs. cash creation: a technical detail with real consequences
The US spot Bitcoin ETF approvals required cash-only creation and redemption — authorised participants deliver dollars to create new shares, the issuer buys BTC on the open market, and reverse for redemptions. Most other ETF asset classes (equities, bonds, commodities) allow in-kind creation, where the AP delivers the basket of underlying securities directly. The cash-only constraint exists because the SEC was unwilling, in early 2024, to permit broker-dealers to handle crypto directly. The consequence is roughly 5-15 bps of additional execution slippage during creation/redemption and a tax inefficiency — every creation forces a taxable event at the trust level. In late 2024 the SEC began consulting on relaxing the rule; the Hong Kong SFC framework already permits in-kind, and tracks 8-12 bps tighter as a result. Watch for US in-kind approval in 2026.
The bottom line is that “ETF” is a wrapper, not an asset class. Spot products give you something close to pure BTC or ETH exposure for 12-25 basis points a year, with regulated custody and brokerage-account convenience. Futures products give you a path-dependent approximation that bleeds in flat or rising markets. Leveraged and inverse products bleed faster. If you can articulate why you own the wrapper rather than the asset — tax wrapper, IRA eligibility, internal mandate, hedging precision — the choice will make itself. If you cannot, you probably want spot, and the cheapest available expense ratio you can hold inside your account type.