Crypto Reference Rates and Benchmarks: What Makes a Price a Reference Rate
When a US spot Bitcoin ETF calculates its daily value, it does not look up the price on an exchange. It uses a reference rate: a single figure computed from trades across several exchanges over a fixed window, by a method published in advance, administered by a company regulated for that purpose.
The difference between a price and a reference rate is the difference between a number and a process. This article explains what that process involves, how the most widely used crypto reference rate is calculated, what regulation adds, and how the various rates in the market differ from one another and from a live aggregated price.
Price, reference rate and benchmark: the difference
A price is what an asset traded at, or was quoted at, at some moment on some venue: a fact about a transaction.
A reference rate is a price constructed to be used by others: to settle a contract, value a fund, mark a portfolio, or measure performance. Its design gives it properties a raw price does not have. It should be representative of the whole market, resistant to manipulation, reproducible by anyone who knows the method, and available at a defined time.
CF Benchmarks, the administrator of the CME CF Bitcoin Reference Rate, frames these as three requirements in its suitability analysis: representation of the underlying market, resistance to manipulation, and replicability by third parties. A price that fails any one of them is unsuitable as a reference, however accurate it happens to be on a given day.
A benchmark is a reference rate that has been formally designated as such under a regulatory framework, with obligations on its administrator. Every benchmark is a reference rate. Only some reference rates are regulated benchmarks. The distinction matters and returns below.
How the CME CF Bitcoin Reference Rate is calculated
The most influential crypto reference rate is the CME CF Bitcoin Reference Rate, usually abbreviated BRR. It settles CME's Bitcoin futures and options, and its New York variant, struck at 4 p.m. Eastern instead of 4 p.m. London, is used by most US spot Bitcoin ETFs to calculate NAV. Its construction shows what a reference rate actually requires.
The BRR is calculated once a day from trades between 15:00 and 16:00 London time; nothing outside that hour counts. Only trades from named constituent exchanges are included. The list has included Bitstamp, Coinbase, Gemini, itBit, Kraken and LMAX Digital, with Bullish added at the end of 2024, and exchanges are selected against published criteria and reviewed periodically. The hour is split into twelve five-minute intervals. Within each interval, all trades from all constituent exchanges are pooled and a volume-weighted median price is calculated, the price at which half the volume traded above and half below, so that a single anomalous trade, however large, cannot move the interval's figure. The twelve interval medians are then averaged with equal weight to produce the BRR.
Each design choice defends against a specific attack. The one-hour window means a manipulator would have to sustain a distorted price for an hour, not a second. Partitioning means they would have to do it across twelve intervals. Medians mean they would have to move more than half the volume in each. Multiple exchanges mean they would have to do it on several venues at once. The result is a rate that is expensive to manipulate, which is the point of the design.
The BRR is published once daily, as a settlement rate. A live price was never its purpose.
What a regulated benchmark administrator is
In 2019, CF Benchmarks became, by its own account, the first crypto index provider authorised by the UK Financial Conduct Authority as a benchmark administrator under the EU Benchmarks Regulation, ahead of the regulation taking full effect in 2020. Its compliance is independently assured by KPMG. The company is a subsidiary of the exchange Kraken.
Kaiko, through its indices business, is a BMR-registered benchmark administrator supervised by France's financial regulator, the AMF. CoinDesk Data, formerly CryptoCompare and CCData, received FCA authorisation as a benchmark administrator in November 2021.
What that authorisation means in practice is set out in the Benchmarks Regulation, Regulation (EU) 2016/1011, which grew out of the IOSCO Principles for Financial Benchmarks written in 2013 in the wake of the LIBOR manipulation scandal. An authorised administrator must maintain governance and conflict-of-interest controls, ensure the quality of input data, oversee contributors, publish its methodology, and submit to supervision. Under Article 29 of the regulation, supervised entities in the EU may only use benchmarks provided by authorised or registered administrators, which is why regulated products such as ETFs and listed derivatives settle against regulated benchmarks.
On this specific dimension, Blocksize and most other data providers are not comparable to CF Benchmarks or Kaiko Indices. Blocksize is not a registered benchmark administrator and does not claim to be. What it provides is a reference price with a published methodology, which serves different purposes.
Types of crypto reference rates: daily to real-time
Not every reference rate is built like the BRR, and the differences reflect what each is for.
Daily settlement rates follow the BRR model: one window, one figure per day, maximum manipulation resistance, designed for contract settlement and fund NAV. They are slow by design.
High-frequency reference rates serve uses that need a market-wide figure more often than once a day, such as intraday marking, risk monitoring and execution benchmarking, while retaining a documented methodology. Kaiko publishes reference rates at a five-second cadence, using a volume-weighted median combined with time-weighting. Coin Metrics publishes rates updating as often as every 200 milliseconds.
Real-time aggregated prices sit at the fastest end of the spectrum. A continuously updated market-wide price such as Blocksize's real-time VWAP is recalculated as trades arrive across 70+ exchanges, with statistical filtering in place of a fixed constituent list, and serves live use cases such as pre-trade checks, real-time risk and oracle inputs, where a periodic rate would be stale. How that filtering works is covered in Real-Time Crypto Data Aggregation: How It Works.
Fixed-timestamp closing prices sit between the two: a price struck on a fixed schedule from a defined window, far more frequently than daily. Blocksize's 30-minute VWAP is calculated every half-hour from the preceding thirty minutes of trades, giving a reproducible closing price forty-eight times a day.
The right choice depends on the use. Settling a monthly futures contract needs the BRR's manipulation resistance and does not need speed. Striking a daily NAV needs a fixed timestamp and a defensible method, which a 30-minute or daily VWAP provides. Checking whether a quote is fair before executing needs a live price, for which a daily rate is useless. The full family of price types, and what each is suited to, is set out in Crypto Prices Explained.
What every reference rate needs
Across the spectrum, the properties that make a rate usable are the same, and they are the properties a raw exchange price lacks.
The first is a published methodology, so that anyone can read how the rate is calculated: which sources, what window, what weighting, what filtering, what happens when data is missing. Without it, a rate can be neither replicated nor audited. The second is defined inputs: either a named list of constituent venues, or documented criteria for inclusion and exclusion. A rate that draws on "the market" without saying which part of it is not a reference rate. The third is manipulation resistance, through some combination of multiple sources, volume weighting, medians, time averaging and outlier filtering, so that no single venue or single trade can move the result cheaply. The fourth is a timestamp convention, so that every published value corresponds to a defined moment or window and can be reproduced and compared. The fifth is a set of contingency rules, written in advance, for what the rate does when an exchange is offline, when volume is too thin, or when inputs disagree.
Blocksize publishes methodology documents for each of its products covering exactly these points: calculation, constituent selection, filtering, contingency rules and governance. That documentation is what allows a Blocksize price to serve as a reference and not merely a quote.
Reference rates for NAV and fair value
Fund valuation is the use case where the distinction between price and reference rate matters most. A fund that strikes NAV against a raw exchange price has a number. A fund that strikes against a reference rate has a number it can explain.
For regulated products such as US spot Bitcoin ETFs, the reference rate is typically a regulated benchmark. According to CF Benchmarks' own release, six of the eleven original US spot Bitcoin ETFs use its rates. For private funds, hedge funds and family offices, the requirement is a reasonable, consistent, documented method, and a reference rate with a published methodology from any credible provider meets it. The fund side is covered in detail in Net Asset Value in Crypto.
One point of accounting precision: an aggregated reference rate is generally a Level 2 input under fair value standards, because Level 1 requires a quote from a single principal market. Funds commonly use a reference rate for NAV while classifying separately for financial statements. The accounting treatment is covered in Crypto Asset Valuation for Audit and Reporting, and the reason single-venue quotes stray from market-wide value is measured in Rethinking Crypto Data.
Blocksize reference prices
Blocksize produces reference prices. It does not produce regulated benchmarks. Its real-time VWAP, 24-hour VWAP, 30-minute VWAP and consolidated bid/ask are each calculated from trades or quotes aggregated across more than 70 centralised and decentralised exchanges, with a published methodology covering calculation, filtering, contingency handling and governance by a data committee.
That places it in the middle of the spectrum above: faster than a daily settlement rate, more structured than a raw exchange feed, and documented in the way a reference rate has to be. The 30-minute VWAP serves as a closing price for valuation. The real-time VWAP serves live use cases where periodic rates are too slow.
The same feeds are published into decentralised oracle networks, where they are validated against consensus before being accepted on-chain, an independent check on representativeness that no single-venue price receives. How that validation works is covered in How Oracle Price Feeds Work.
For products that require a regulated benchmark, the answer is a regulated benchmark administrator. For valuation, benchmarking, risk and execution, a documented reference price is what the use case actually needs.
Frequently asked questions
What is a crypto reference rate? A price constructed for use by others, to settle contracts, value funds or measure performance, designed to be representative of the whole market, resistant to manipulation, and reproducible from a published methodology. It differs from a raw exchange price, which is simply what one venue traded at.
How is the CME CF Bitcoin Reference Rate calculated? From trades on named constituent exchanges during a one-hour window, 15:00 to 16:00 London time, split into twelve five-minute intervals. A volume-weighted median is calculated for each interval, and the twelve medians are averaged with equal weight. The rate is published once daily.
What is the difference between a reference rate and a benchmark? A benchmark is a reference rate formally designated under a regulatory framework such as the EU Benchmarks Regulation, with obligations on its administrator around governance, input data and transparency. All benchmarks are reference rates; only those from authorised administrators are regulated benchmarks.
Which crypto data providers are regulated benchmark administrators? CF Benchmarks (UK FCA-authorised, 2019), Kaiko Indices (BMR-registered, supervised by France's AMF) and CoinDesk Data (UK FCA-authorised, 2021). Most other data providers, including Blocksize, publish reference prices with documented methodology but are not registered benchmark administrators.
Why do Bitcoin ETFs use a reference rate instead of an exchange price? Because a reference rate is representative across venues, resistant to manipulation, reproducible and regulated, properties a single exchange's price lacks. In the EU, supervised entities may only use benchmarks from authorised or registered administrators.
Can a reference rate be real-time? Yes. Reference rates range from once-daily settlement rates like the BRR to rates updating every few seconds or continuously. Faster rates trade some manipulation resistance for timeliness, and suit live use cases such as pre-trade checks and risk monitoring.
What makes a price suitable as a reference rate? A published methodology, defined inputs, manipulation resistance through multiple sources and robust statistics, a timestamp convention, and written contingency rules for missing or anomalous data.