Amberdata's Reference Rates provide benchmark prices for BTC and ETH across qualified exchanges. Amberdata’s hourly and daily reference rates are published once per hour and once per day. They are SOC I and II compliant, GAAP-aligned, and adhere to the IOSCO Principles for Financial Benchmarks.
Reference rates play an important role for financial institutions, who use benchmark reference prices for reporting, making informed trading decisions, and settling contracts. Amberdata’s Reference Rates are produced using trade data from exchanges that meet a selection of quantitative and qualitative criteria. The trade data is processed via several statistical techniques to produce a highly representative United States dollar price for a given digital asset. The price is denominated in U.S. dollars because the U.S. dollar is the most widely used currency in international transactions.
The daily rate is an hourly reference rate marked with a timezone relevant to the user’s geographical location. For example, an hourly reference rate calculated at 8 p.m. UTC is the 4 p.m. EST daily reference rate and will be marked as such when the rate is produced.
BTC and ETH reference rates are currently available via REST API, with delivery in AWS S3 coming soon.
The algorithm for hourly and daily reference rates of an Asset, A, at Delivery Time, T, can be summarized in the following steps:
- At the top of the hour, retrieve all Qualified Transactions for Asset, A, within the Lookback Window, L
- Convert all of the Qualified Transactions to U.S. dollar prices where necessary
- Split L into K partitions of size L
- Within each partition, calculate the volume weight per unique price level per exchange
- Within each partition, calculate the price variance weight per exchange
- Within each partition, calculate the transaction weight per exchange
- Combine the aforementioned weights to generate an aggregate weight per price level per exchange within each partition
- For each partition calculate the weighted median price using the price and aggregate weight pairs
- The outcome of the weighted median calculation will be K prices, one for each partition of L
- Using the K prices, compute a Hadamard product with exponentially decreasing time weights defined for each partition
- The sum of the elements in the Hadamard product vector is the hourly reference rate
- For the supported timezones and geographies, mark the hourly reference rate as a daily reference rate when appropriate
Reference Rates White Paper
Exchanges are qualified using a rigorous methodology. A venue is eligible to be a Qualified Exchange if it offers a spot trading market for any of the Qualified Transactions of the Supported Assets. All venues that are in consideration to be included in the reference rate calculation must fulfill the following quantitative measures:
- For a given asset in Supported Assets, over the previous 180 days, the mean daily volume of the asset in the venue under evaluation must be at least 3% of the combined mean daily volume of the asset across all venues over the same period.
- 3% is set as the threshold because 2.35% is the cutoff for two standard deviations. Hence we round up to the nearest integer to not be below the two standard deviation threshold
- Volume is measured in units of the given asset
Every 90 days, all currently included exchanges and any not included are evaluated against the latest Eligibility Criteria. In extraordinary circumstances, exchanges that do not meet the Eligibility Criteria may be excluded from the methodology temporarily. The temporary suspension may result in permanent removal if warranted. Please refer to the Reference Rates White Paper for more information on exchange qualification.
The current list of exchanges whose transactions (spot trades) are included in the real-time reference rates calculation is listed below.
|Removed in Version
For the query parameter dailyTime the following timezones are supported:
|4 p.m. Relative to UTC
|Singapore & Hong Kong
What makes your reference rates manipulation-resistant?
- Our reference rate algorithm is highly immune to manipulation through the use of weights and medians. The combination of price-level volume weight, exchange price dispersion weight, and exchange transaction weight ensures that exchanges with high volume and low price dispersion are treated favorably.
Why do you use exponential time weighting?
- Staying relevant is crucial in volatile crypto markets, where prices can and do change extremely rapidly. Amberdata uses exponential time weighting in our calculations by assigning more weight to recent transactions. This makes our reference rate highly responsive to the latest market conditions. By emphasizing recent data, our approach minimizes the impact of earlier noise and any outliers and reduces lag effects from news, regulatory events, or large trades that affect other methods.
How does your methodology adapt to various market conditions?
- Our methodology considers how price dispersion is important during period of high volatility and also penalizes illiquid and highly volatile exchanges in the calculation. For time-sensitive and regulation-bound use cases (such as a benchmark for financial instruments, tax, accounting, compliance, etc), our comprehensive and responsive approach ensures precision, relevance and accuracy.
Why do you have two London and New York timezones supported for daily rates?
- There are two options for New York and London to account for Daylight Saving Time and British Summer Time respectively.
Updated 3 months ago