Crypto indices and benchmarks: how the number a product tracks is made

An allocator who buys a spot bitcoin ETF, a European crypto ETP or a structured note rarely asks what number the product actually tracks. They assume "the bitcoin price" is a single, self-evident fact. It is not. Every one of these products points at an index, and that index is a set of rules, run by a named administrator, over a chosen set of trading venues, at a chosen moment in time. Two providers pricing the same asset on the same day can, and do, publish different numbers. Understanding why starts with the rulebook, not the price.

Who gets into the basket

An index administrator does not price every token that trades. It screens a universe down to a set of "eligible" assets and venues, and the screen is the first place methodologies diverge. S&P Dow Jones Indices, for its digital asset series, takes spot prices from Lukka, a third-party data vendor whose Lukka Prime methodology identifies a "primary market" for each asset from a pool of roughly 18 to 20 screened exchanges, using factors the vendor describes only in general terms, such as jurisdiction, controls and trading activity, rather than a fully public checklist. CF Benchmarks, which administers the CME CF Bitcoin Reference Rate (BRR) that underpins CME's bitcoin futures, works the other way round: it names its constituent exchanges individually and publishes the criteria, covering security, KYC and regulatory standing, that a venue must meet to be added.

Both approaches are legitimate. "The exchanges behind this number" is a published, changeable list, not a fixed fact about the asset, and a small named panel versus a larger vendor-screened pool carries different exposure to any single venue's liquidity and any single venue's problems.

Which venues set the price, and how they are weighted

Once a venue is eligible, its trades still have to be turned into one number. A reference rate built on a small panel typically takes a volume-weighted calculation across those venues over a defined window, which limits how far one thin, manipulated print can move the fix. A fair-value methodology built on a wider vendor pool instead nominates a "primary market" per asset and leans more heavily on that single source at any given moment. Neither is inherently safer; they fail in different conditions. A narrow panel is more exposed if two of its six or seven venues go quiet at once. A vendor-selected primary market is more exposed to a change in that one venue's own liquidity or a change in the vendor's own internal model, which the index user cannot see happen in real time.

This is a different problem from the one covered in our piece on how on-chain price oracles work and fail: an oracle feeds a smart contract in near real time from whichever venues its owner configured, usually with no external administrator at all. An index sits one layer up, with a named, accountable administrator and, for the more heavily used series, a published methodology and a regulator that can ask it questions.

The fix: pricing windows and calculation

A daily reference rate is not a snapshot; it is an average, or a median, taken across an observation window on that venue panel, calculated once a day at a stated time. CF Benchmarks' BRR, launched on 14 November 2016 and administered under a published methodology overseen by a CME CF Oversight Committee, is exactly this kind of daily fix, distinct from the same provider's real-time index, which recalculates continuously rather than once. The same distinction runs through fund administration: Fidelity's own ETF documentation, cited in our piece on how ETF creation and redemption actually works, describes "a volume-weighted median methodology using eligible bitcoin spot markets" struck once a day at 4:00pm Eastern Time to set net asset value, which is a different figure again from the fund's market price, its bid-and-offer quotes and any intraday indicative value.

The practical consequence: a "bitcoin index" quoted at 11:00am may be a live, continuously updating figure, or a number fixed at 4:00pm the previous day that has not moved since. A methodology document says, precisely, which one a given series is.

When a venue goes dark

Constituent panels change, and the change itself is instructive. On 27 July 2026, CF Benchmarks suspended itBit Paxos as a constituent exchange across its bitcoin, ether and litecoin reference rates and real-time indices, following the retirement of the itBit trading platform; the suspension was formalised as a permanent removal roughly a month later. The mechanics matter more than the single event: a well-run reference rate has a pre-defined process for suspending a venue immediately when it stops producing reliable data, ahead of any formal consultation on removing it permanently, so that a defunct exchange cannot go on contaminating a live price. A methodology that instead has to be hand-patched after the fact, or that keeps a dead venue in its calculation because no one updated the list, is a worse methodology regardless of how sound its weighting formula looks on paper.

The same question applies to forks and delistings: what happens to the index the day a constituent asset forks or a venue delists the token, and how long does that decision take to reach the published number.

Rebalancing and capping

Multi-asset crypto indices rebalance on a set schedule, commonly quarterly, and most cap the weight of any single asset, typically bitcoin, so it cannot dominate the basket. Uncapped market-cap weighting would make most "diversified" crypto indices a leveraged bet on whichever token is largest that quarter; the cap level is a policy choice the administrator makes and can change, not a law of the market.

Who can change the rules

Every methodology names a body that can amend it, and the honest question is how much friction stands between "the administrator wants to change the rule" and "the rule changes." Under the EU Benchmarks Regulation, an in-scope administrator is legally required to publish "the key elements of the methodology," details of its internal review and approval process, and, specifically, "the procedures for consulting on any proposed material change in the administrator's methodology and the rationale for such changes, including a definition of what constitutes a material change," with advance notice and public disclosure of the comments received and the administrator's response. That obligation, Article 13 of Regulation (EU) 2016/1011, is a genuine constraint on an in-scope administrator changing course quietly. It is not, however, a constraint that reaches most crypto index providers, for a reason covered below.

The governance layer: what the rules actually require, and of whom

"Regulated benchmark" is doing a lot of work in marketing copy, and the scope has just narrowed sharply. The EU Benchmarks Regulation (Regulation (EU) 2016/1011) has applied since 1 January 2018, requiring administrators of in-scope benchmarks to be authorised or registered and to publish a methodology statement (Article 13) and a benchmark statement for each benchmark (Article 27), the latter setting out what the benchmark measures, how discretion is exercised, and what happens on cessation, reviewed at least every two years. But Regulation (EU) 2025/914, which entered into force on 8 June 2025 and has applied since 1 January 2026, narrowed the regime substantially: EU BMR now binds only "critical" and "significant" benchmarks, plus EU Paris-aligned and Climate Transition benchmarks, and non-significant benchmarks, along with most third-country benchmarks lacking an equivalence, recognition or endorsement decision, fall outside the regulation entirely and can be freely referenced by EU firms regardless. As of September 2026, no crypto reference rate or index has been designated critical or significant, though administrators already on ESMA's register at the end of 2025 have until 30 September 2026 for that designation exercise to complete, so today almost none of the benchmarks behind crypto ETPs, structured notes or fund NAVs sit inside mandatory EU BMR scope.

The UK took a different path. UK BMR, the onshored version of the same 2016 regime, still requires administrator authorisation through the FCA, and the FCA maintains a public UK Benchmarks Register recording who is authorised, registered or recognised. CF Benchmarks is one of the few named examples of an FCA-authorised crypto benchmark administrator, having received that authorisation in 2019; the firm states this lets regulated European and UK firms reference its indices in compliant products. That authorisation is separate from, and should not be confused with, the FCA's broader new cryptoasset regime for exchanges, custodians and stablecoin issuers, under which firms can apply from 30 September 2026 for a mandatory regime taking effect on 25 October 2027, which concerns firm conduct, not benchmark methodology.

Underneath both regimes sits IOSCO's 2013 Principles for Financial Benchmarks, the voluntary global template both draw on: governance, benchmark quality, methodology quality and accountability, with administrators expected to self-assess and publish a compliance statement. IOSCO has no crypto-specific benchmark principles; its 2023 crypto recommendations address exchange conduct, not index construction. Where a data vendor claims IOSCO alignment, as Lukka does in describing a Big Four review of its methodology against the principles, that is the vendor's own assurance exercise, not a regulator's authorisation, and the two should not be read as equivalent.

European crypto ETPs add a further wrinkle covered in our piece on how those products are structured: MiCA explicitly excludes ETPs from its scope, leaving them governed by the Prospectus Regulation and MiFID II rather than by any crypto-specific EU regime, and now, since January 2026, potentially by no Benchmarks Regulation obligation on their underlying index at all.

Why two indices on the same asset diverge

Put the pieces together and the divergence between, say, S&P's bitcoin index and the CME CF Bitcoin Reference Rate on any given day is not noise; it is the accumulated effect of different venue panels, different weighting logic, different fixing windows and different rebalancing and capping rules, each a deliberate methodology choice rather than an error. A volume-weighted median across six or seven named, actively monitored exchanges will diverge, usually by a small margin but sometimes materially during stressed markets, from a fair-value estimate drawn from a wider, vendor-screened pool using a different primary-market logic. Neither number is "wrong." A product's prospectus or fact sheet names which index it tracks for exactly this reason, and switching the reference index, something an administrator or an issuer can do, changes the product's return series without changing the underlying asset at all.

What to check before relying on a benchmark

• Whether the administrator is on the FCA's UK Benchmarks Register or, if the product is EU-distributed, whether the underlying benchmark is actually within scope of EU BMR after the 1 January 2026 narrowing, since "regulated" and "in scope" are no longer the same claim for most crypto indices.
• Whether the methodology document names its constituent venues and eligibility criteria, or defers to a vendor's own undisclosed screening process.
• What the fixing window is, and whether the number quoted to you is that daily fix or a continuously updating variant.
• What the published process is for suspending a venue that stops producing reliable data, and how long a formal removal takes to follow.
• Who can approve a material methodology change, whether that requires public consultation, and where past consultation responses are published.
• Whether independent assurance over the methodology, where claimed, was performed by an accounting firm engaged by the administrator or the data vendor, rather than imposed by a regulator.

None of this is investment, legal, tax or accounting advice; it is mechanics worth putting in front of a risk committee before a product referencing an index gets approved, the same scrutiny an allocator would give the vehicle itself, a step our piece on how institutions actually allocate to digital assets found is often skipped because a satellite crypto sleeve has no obvious benchmark to begin with. The index a product tracks is a policy document with a price attached, and reading the policy is the diligence step that gets skipped because the number on the screen looks like a fact.