A spot bitcoin ETF holds actual bitcoin through a custodian; a futures bitcoin ETF holds CME bitcoin futures contracts that expire and must be rolled forward, which can add a cost a spot fund never incurs. That cost has ranged widely across the sources reviewed here: Bloomberg reported a roughly 2-percentage-point gap in ProShares’ BITO after its first two months of trading, WEEX Crypto News puts the drag at 5 to 10 percentage points a year in contango markets, and a 24/7 Wall St snapshot from March 2026 showed BITO’s one-year return nearly matching bitcoin’s own decline. The size of the gap depends on the shape of the futures curve at the time, not a fixed rate set in a prospectus.
What each fund actually holds
Per WEEX Crypto News, published June 4, 2026, spot bitcoin ETFs hold bitcoin directly through custodians such as Coinbase, while futures-based ETFs track CME bitcoin futures contracts rather than the coin itself. AlphaAI Capital’s research team, in an article published May 17, 2026, frames the same distinction: a spot product is designed to track the underlying asset by holding it, while a futures product gains exposure through a standardized contract tied to a future date; the article notes that futures contracts are time-limited, do not convey ownership of the underlying asset, and require the fund to settle or roll them forward as expiration approaches. CF Benchmarks, in a blog post published October 16, 2025, adds a detail specific to bitcoin: most US-listed spot bitcoin ETFs price their net asset value against CME’s BRRNY reference rate (the New York variant), while CME futures settle against the London variant of the same index, BRR. CF Benchmarks says this shared benchmark ties the regulated spot and futures markets together and keeps tracking error between them small at the index level – though, as shown below, that alignment at the index level does not eliminate the cost of running a futures-based fund through time.
The mechanism: contango, backwardation, and the roll
The cost of rolling depends on the shape of the futures curve. When longer-dated futures trade above near-term ones – a condition called contango – a fund rolling its position forward sells a cheaper contract and buys a more expensive one, creating a drag. When the curve inverts into backwardation, the roll can cost less or even help. The CFTC, as described by AlphaAI Capital, calls the contango case a “roll premium” that traders effectively pay when routinely rolling contracts forward. CF Benchmarks’ October 16, 2025 post states that Bitcoin markets have typically sat in contango, with backwardation occurring only during extreme bearish sentiment or market stress, and describes contango-based basis trading as the dominant strategy in the bitcoin derivatives market as a result.
A worked example of the basis
CF Benchmarks’ October 16, 2025 post lays out a simple contango scenario. Spot bitcoin trades at $100,000 and a near-expiry CME futures contract trades at $101,000, so the basis is $1,000. Annualized over a 30-day period, CF Benchmarks calculates that as roughly 12.2% (a 1% monthly basis, annualized as 1% × 365/30). A trader who bought spot bitcoin via an ETF at $100,000 and simultaneously sold the futures contract at $101,000 would, per CF Benchmarks’ example, see both prices converge to $105,000 by expiration. The spot leg gains $5,000; the short futures leg loses $4,000; the net result is the original $1,000 basis, locked in regardless of which direction bitcoin actually moved.
CF Benchmarks’ example describes a trader who is short the futures contract, profiting as the basis converges by expiration – the opposite side of the trade from a futures ETF, which stays long a rolling futures position and never lets a single contract reach expiration before replacing it with a new one. As analysis, not stated by CF Benchmarks: a fund positioned on the long, continuously-rolling side of a contango market would be the one paying away the basis the short side captures, each time it rolls. But CF Benchmarks’ own example does not model a rolling long position, and the evidence here does not show how closely the arbitrage economics it describes map onto any specific fund’s realized roll cost. That mapping is inference, not something CF Benchmarks’ data confirms.
What that has meant in practice, at two different points in time
The actual size of the roll-cost drag has looked very different depending on when it was measured. Bloomberg’s headline-only figures do not themselves state a year; the timing here is inferred from 24/7 Wall St’s separate reporting that ProShares’ BITO, the fund Bloomberg’s headline describes, launched in October 2021 as the first US bitcoin futures ETF. On that basis, Bloomberg reported that BITO trailed spot bitcoin by 2 percentage points after its first two months of trading, with a projection at the time of roughly 12 points of underperformance over its first year, attributed largely to the cost of rolling futures contracts. This figure is drawn from a source held only as a headline for this article, so it should be read as a contemporaneous estimate rather than a confirmed year-end result, and it is not known from the evidence available whether that 12-point projection actually materialized.
A very different snapshot comes from 24/7 Wall St, reporting on March 25, 2026, when bitcoin was trading near $71,000 and down roughly 19% year to date. At that point, 24/7 Wall St reported BITO’s one-year return of -19% as slightly underperforming bitcoin’s own roughly 19% one-year decline – a smaller gap than the 2021 figures suggest, but still a reported shortfall, not a match. The same article reported BITB, a spot bitcoin ETF, down 19.3% year to date against bitcoin’s own 19.3% year-to-date decline as of that date, and put BITB’s expense ratio at 0.2% against BITO’s 1%. 24/7 Wall St’s own read is that BITO’s futures structure adds roll costs and a higher expense ratio relative to a spot fund, and that the gap has been more pronounced in prior periods when contango was steeper – consistent with the larger gap Bloomberg reported and the smaller, but still present, gap in the 2026 snapshot.
WEEX Crypto News, in its June 4, 2026 article, offers a broader estimate: that rolling contracts monthly in contango markets can create a drag causing a futures ETF to underperform spot bitcoin by 5% to 10% annually, versus none for a spot ETF. WEEX also lists expense ratio ranges of 0.19% to 0.90% for spot bitcoin ETFs and 0.70% to 1.50% for futures ETFs, and works a separate example showing that a 1.31% fee difference on a $100,000 position costs $1,310 a year in fees alone, before any roll-cost effect. WEEX is a crypto exchange that competes for spot trading volume, which is worth keeping in mind when reading its comparison of ETF costs against direct exchange trading.
Why the gap moves instead of holding steady
The pattern across these three snapshots – a roughly 2-point gap after two months, a 12-point full-year projection made at that same time, and a smaller but still reported gap in the 2026 snapshot – is not a contradiction. It reflects that roll cost is a function of the futures curve at each roll date, not a fixed annual tax written into a prospectus. CF Benchmarks’ October 16, 2025 data shows the annualized front-month basis on CME bitcoin futures approaching 25% in February 2024 and exceeding 20% in November 2024, before briefly dipping below zero in March 2025 during a reversal in market momentum. Reading across from that pattern – which CF Benchmarks does not itself apply to fund-level tracking – a futures ETF rolling through a period of a 20%-plus annualized basis would be expected to pay away far more than one rolling through a period near zero or in backwardation. Expense ratios are fixed by prospectus; roll cost is not.
The demand side: what spot ETF flows show
Separately from tracking mechanics, spot bitcoin ETF demand itself has fluctuated. Per CoinDesk, citing data source SoSoValue, US-listed spot bitcoin ETFs had drawn $58.72 billion in cumulative net inflows as of early May 2026 since their January 2024 launch, still below the $61.19 billion peak reached in October 2025 – the year is not stated explicitly by CoinDesk for that figure but is inferable from the article’s own timeline, which places the peak before the November 2025-to-February 2026 outflow period it also reports. CoinDesk reported that investors pulled $6.38 billion out of the funds between November 2025 and February 2026, a period in which bitcoin fell from over $100,000 to nearly $60,000, before ETFs took in $3.29 billion over the following two months as of that same report, published May 4, 2026. None of this speaks to tracking or roll cost directly, but it is useful context: the futures-versus-spot cost comparison sits on top of a market where flows into the spot vehicles themselves have not been steady.
What this page does not tell you
Roll cost is not a rate a reader can apply going forward. Every figure above – the 2-point gap after two months, the 12-point full-year projection, the 5% to 10% annual estimate from WEEX, and the smaller gap in the 24/7 Wall St 2026 snapshot – is a reading taken at one point in time under one shape of the futures curve, not a constant a reader can plug into a future holding period.
The Bloomberg figures are cited from a source held only as a headline for this article, and the year attached to them is inferred from 24/7 Wall St’s separate reporting on BITO’s launch date, not stated by Bloomberg itself. This page states what that headline claims but cannot verify its underlying methodology, and it is not known from the evidence available whether the projected 12-point first-year underperformance for BITO actually materialized.
AUM, flow, and expense-ratio figures throughout this page are self-reported by ETF issuers or by data aggregators such as SoSoValue, as cited by CoinDesk and 24/7 Wall St, not independently audited by any source cited here.
The evidence used here does not include the underlying CME futures-curve data needed to reconcile precisely why the tracking gap Bloomberg reported was larger than the gap 24/7 Wall St reported for the one-year period ending March 2026. The general direction – contango costs more when it is steeper – is documented in CF Benchmarks’ data, but a month-by-month reconciliation between the two snapshots is not something this page can support.
WEEX’s 5% to 10% estimate comes from a crypto exchange that competes with ETFs for spot trading volume; that commercial interest does not make the figure wrong, but it is not an independent or audited estimate either.
Sources
Every fact above is attributed to one of these reports. Where they disagree, the article says so.
Read More
All News
August 16, 2026
What a seed phrase is, and how it can be stolen or lost
August 16, 2026
