Quick Navigation
General
What metric naming conventions does Coin Metrics use?
What metric naming conventions does Coin Metrics use?
volume_reported_future_perpetual_usd_1d is ordered such that the volume term is first and all subsequent terms are modifiers to what type of volume the metric represents.Some metrics are naturally represented as an aggregation (such as a sum or mean) over a time interval (such as a block, an hour, or day). If the metric represents an aggregation over a time interval, the interval is appended as a suffix to the metric name. If the metric represents a value at a point in time, there is no suffix. Please see the frequently asked question “What timestamp conventions does Coin Metrics use?”.The first exception to this convention is that all Network Data Pro metrics use upper camel case (ex: CamelCase) in which names omit spaces and the separation of words is indicated by a single capitalized letter. The first word is also capitalized. Network Data Pro metrics used the upper camel case naming convention prior to our adoption of the snake case naming convention for all other metrics, so we maintain the upper camel case naming convention for Network Data Pro metrics for consistency and backwards compatibility.The second exception is our Reference Rate metrics (ReferenceRate, ReferenceRateUSD, `ReferenceRateEUR, etc) which use a point-in-time convention, but for similar historic reasons are named in camel case for backwards compatibility.If a frequency or time interval is added to the metric name, it follows the following conventions:s: secondsb: blocksm: minutesh: hoursd: daysw: weeksmo: monthsq: quartersy: years
What timestamp conventions does Coin Metrics use?
What timestamp conventions does Coin Metrics use?
time for a trade is 2021-08-04 23:56:00.356749000, that means that the trade was executed exactly at that timestamp.For any data type that represents a summary statistic over an interval of time, we set the timestamp to the beginning of the time interval. This is referred to as the “beginning-of-interval” timestamp convention. Summary statistics can include transformations such as sum, mean, median, and count. Our candles is an example of a data type that follows this convention because it represents the open, high, low, close, and volume over an interval of time such as a day. For instance, if the time for a daily candle is 2021-08-04 00:00:00.000000000, that means the candle represents the daily interval from 2021-08-04 00:00:00.000000000 to 2021-08-04 23:59:59.999999999, inclusive. Here we represent 00:00:00.000000000 as the beginning of the day according to the ISO 8601 standard.The following API endpoints serve data using the point-in-time convention:/timeseries/market-trades/timeseries/market-openinterest/timeseries/market-liquidations/timeseries/market-funding-rates/timeseries/market-funding-rates-predicted/timeseries/market-orderbooks/timeseries/market-quotes/timeseries/market-contract-prices/timeseries/market-implied-volatility/timeseries/market-greeks/timeseries/index-levels/timeseries/index-constituents- Any metric in
/timeseries/asset-metrics,/timeseries/exchange-metrics,/timeseries/exchange-asset-metrics,/timeseries/pair-metrics,/timeseries/institution-metrics,/timeseries/market-metricswithsnake_casenaming convention and without an interval suffix, such asopen_interest_reported_future_usd - The Reference Rate asset metrics, including
ReferenceRate,ReferenceRateUSD,ReferenceRateEUR, etc. which follow an older convention before we adopted snake case.
/timeseries/pair-candles/timeseries/market-candles- Any metric in
/timeseries/asset-metricswith upper camel case (ex:UpperCamelCase) naming convention - Any metric in
/timeseries/asset-metrics,/timeseries/exchange-metrics,/timeseries/exchange-asset-metrics,/timeseries/pair-metrics,/timeseries/institution-metrics,/timeseries/market-metricswithsnake_casenaming convention and with an interval suffix, such asvolume_reported_future_perpetual_usd_1d
What asset ticker naming conventions does Coin Metrics use?
What asset ticker naming conventions does Coin Metrics use?
parentasset[_fullname][_network][_chain], where the fullname, network, and chain are optional. Market data and aggregated network data are assigned to the parentasset ticker, where aggregated network data consists of data from individual network or chain-specific forms of an asset.To understand our naming convention, we first introduce some important context surrounding the two primary considerations regarding unique ticker symbols.First, what is thought as a singular asset may actually exist in various forms across multiple layer one and layer two blockchains. From the perspective of the blockchain ledger, each form resides on a separate blockchain, and Coin Metrics collects and produces data for each form independently. To differentiate between the specific form, Coin Metrics appends the blockchain ticker as a suffix to the asset ticker. For example, Tether (usdt) exists on Tron, Ethereum, Solana, and several other blockchains. To track the network activity of each form, Coin Metrics uses usdt_tron, usdt_eth, and usdt_sol tickers, respectively.Centralized exchanges, however, do not typically differentiate between different forms of an asset. They will allow users to deposit several forms of an asset and credit users internally with a generic parent form of the asset. Trading then occurs using the generic form of the asset, and all market data collected by Coin Metrics is assigned to the parent ticker.Returning to the Tether example, a centralized exchange may allow a user to deposit the ERC-20 form of Tether, which resides on Ethereum, as well as the TRC-20 form of Tether, which resides on Tron. Regardless of which form a user deposits, the user is credited with a generic parent form of Tether which is traded on all markets on the centralized exchange. Therefore, market data such as trades are assigned to the parent asset usdt.Coin Metrics also aggregates data from individual forms of an asset to the parent asset for certain metrics. Therefore, metrics under the parent asset usdt represent an aggregation across usdt_tron, usdt_eth, and the other individual forms of Tether.Second, some assets may share the same display ticker as another asset. To resolve these ticker conflicts, Coin Metrics ensures that each asset ticker in our coverage universe is unique by appending the full name of the asset as a suffix to the asset ticker. For example, both Starcoin and Starchain share the same display ticker of stc. Coin Metrics resolves this ticker conflict by assigning the tickers stc_starcoin and stc_starchain, respectively.What market naming conventions does Coin Metrics use?
What market naming conventions does Coin Metrics use?
{exchange}-{base}-{quote}-spot naming convention like coinbase-btc-usd-spot. The base and quote both use our harmonized asset tickers.For futures markets, we use the {exchange}-{symbol}-future naming convention like binance-BTCUSDT-future. The symbol is the exchange-reported symbol.For option markets, we also use the {exchange}-{symbol}-option naming convention like deribit-BTC-15OCT21-60000-C-option. The symbol is the exchange-reported symbol.We use exchange-reported symbol for our derivative markets because of the difficulty in establishing a standard naming convention. Exchanges may list multiple different contracts that trade the same underlying but with different contract specifications. And exchanges may list exotic derivative contracts that do not conform to standard contracts. In general, exchanges are constantly experimenting with different contract specifications. To prevent any confusion in which contract we are referencing, we adopt the exchange-reported symbol. More metadata about our markets such as the contract specifications, price and amount precision, and fees can be found through our /reference-data/markets endpoint.For decentralized exchange liquidity pools, we use the the {exchange}-{pool_config_id}-{base}-{quote}-spot naming convention.The pool_config_id is needed since for some decentralized exchanges like Uniswap v3, multiple pools can exist for a single base and quote asset pair with different pool configurations such as different fees. More metadata about our liquidity pools can also be found through our /catalog/markets endpoint.The pool_config_id usually takes an integer value that represents the order in which a pool was created. For convenience, we also create a synthetic aggregated market where we set the pool_config_id to agg so that users can pull candles data that is aggregated across all pools.Network Data
How do I see full Stablecoin Universe Market Cap?
How do I see full Stablecoin Universe Market Cap?
Why don't you have Coinbase deposits/withdrawals (on-chain flows) supply data?
Why don't you have Coinbase deposits/withdrawals (on-chain flows) supply data?
Do your aggregate exchange deposit/withdrawal and supply (on-chain flows) metrics include any Coinbase flows?
Do your aggregate exchange deposit/withdrawal and supply (on-chain flows) metrics include any Coinbase flows?
Why is your Current Supply (SplyCur) of XLM different from the supply reported on Stellar's Dashboard? (https://dashboard.stellar.org)
Why is your Current Supply (SplyCur) of XLM different from the supply reported on Stellar's Dashboard? (https://dashboard.stellar.org)
What is the latency of block-by-block network data?
What is the latency of block-by-block network data?
Why does your Market Capitalization metrics differ so heavily from other data providers?
Why does your Market Capitalization metrics differ so heavily from other data providers?
Do you have metrics for total blockchain size?
Do you have metrics for total blockchain size?
Do you have metrics for total transactions?
Do you have metrics for total transactions?
How can you calculate total transfer value on the Ethereum Blockchain (ETH + other ERC20s)?
How can you calculate total transfer value on the Ethereum Blockchain (ETH + other ERC20s)?
What is the best source for daily volume for Tether-Omni, Tether-ERC20, USDC and DAI?
What is the best source for daily volume for Tether-Omni, Tether-ERC20, USDC and DAI?
Is there a way to approximate the number of users for a specific blockchain?
Is there a way to approximate the number of users for a specific blockchain?
Do you have a metric for the number of days BTC is held between transactions?
Do you have a metric for the number of days BTC is held between transactions?
Why does your FeeTotNtv metric for BTC slightly differ from other data providers?
Why does your FeeTotNtv metric for BTC slightly differ from other data providers?
How are your aggregated Exchange Flows calculated?
How are your aggregated Exchange Flows calculated?
How are your aggregated Miner Flows calculated?
How are your aggregated Miner Flows calculated?
What are the exchanges that serve as constituents for your Trusted Volume metric?
What are the exchanges that serve as constituents for your Trusted Volume metric?
Is there a way to calculate ETH staking yield metrics with your Consensus Layer metrics?
Is there a way to calculate ETH staking yield metrics with your Consensus Layer metrics?
What asset ticker naming conventions does Coin Metrics use?
What asset ticker naming conventions does Coin Metrics use?
- An asset may be represented by tokens on more than one layer one network. For instance, Tether exists on 16 different networks, such as Ethereum, Solana, and Tron. Similarly, an asset may also exist on multiple versions of a layer two network deployed across multiple layer one networks.
- An asset may undergo a contract upgrade to enable new features or address security flaws. This involves deploying a new smart contract to create a new token. Existing holders of the old token can exchange their old tokens for new tokens. Both new and old versions of an asset can exist simultaneously.
- An asset may begin its existence as a token represented as a smart contract on a network but then migrate to being represented as a native asset on another network or as a smart contract on another network. Both forms of the asset can exist simultaneously.
- Centralized exchanges allow users to deposit various forms of an asset. Users are then credited with a generic form of the asset and are able to trade the generic form of the asset on the exchange. For instance, an exchange may allow users to deposit Tether tokens on the Ethereum network and Tether tokens on the Solana network. Users would be credited with generic Tether and trade generic Tether. Centralized exchanges do not have one set of markets for Tether tokens on Ethereum and a separate set of markets for Tether tokens on Solana.
- Units of an asset may be bridged from a Layer 1 network (L1) to a Layer 2 network (L2) or another Layer 1 network through a blockchain bridge. A blockchain bridge can be a centralized entity or a smart contract that allows units of an asset on a Layer 1 network to be locked and an equivalent amount of the asset to be issued on a Layer 1 or 2 network. Sometimes the asset that is bridged is referred to as a wrapped version of the asset.
- For some assets, there is no industry-wide convention on the ticker. A centralized exchange may decide to use a ticker to refer to a specific asset that differs from another centralized exchange. Multiple assets may also share the same ticker.
- Project: Represents the asset across all networks and contract instances. This is the broadest classification. Data collected from centralized exchanges are assigned to assets at the project level.
- Layer One Network: Represents the asset on a specific layer one network, encompassing all contract instances and layer two networks on that network. Network data is assigned to this level.
- Layer Two Network: Represents the asset on a specific layer two network, encompassing all the contract instances on the layer two network. Network data is also assigned to this level.
- Contract: Represents the asset at the individual smart contract level on a specific network. This is the lowest level of classification. Data from decentralized exchanges is assigned to this level.
[_fullName]: If a ticker conflict exists, we append the full name of the asset immediately after the asset, to make the ticker unique.[_contractVersion]: If an asset is at the contract level and has multiple smart contract implementations, we append the contract version to uniquely identify each contract level asset. The contract version is a sequential integer starting with 1.[>bridgedFromNetwork|>wrappedFromNetwork]: If an asset is a bridged asset, we append the network from which the asset is bridged from using>bridgedFromNetwork. If an asset is a wrapped asset, we append the network from which the asset originally existed using>wrappedFromNetwork.[#bridgeUsed|#wrappedContractAddress]: If the asset is a bridged asset, we append the bridge that was used using#bridgeUsed. If the asset is a wrapped asset, then we append the wrapped contract address using#wrappedContractAddress.[_network|_layer2Network.layer1Network|_bridgedToNetwork]: If the asset represents the asset on a specific network, we append the network using_network. If the asset represents a bridged asset using_bridgedToNetwork. Please note that_bridgedToNetworktakes the naming convention oflayer2Network.layer1Networklikebase.ethif the bridged to network is a layer 2 network.
Why are their multiple versions of the same asset, such as usdt and usdteth and usdtsol?
Why are their multiple versions of the same asset, such as usdt and usdteth and usdtsol?
usdt_eth and usdt_sol to these tokens, respectively.Using Tether as an example, we refer to the ticker usdt as the project level asset, and we refer to the tickers usdt_eth and usdt_sol as network level assets.Network data that is specific to an asset on a specific network, such as a swap that occurs on a decentralized exchange, is assigned to network level asset tickers. Network data that can be aggregated across multiple networks is assigned to the project level asset. Market data is also assigned to the project level asset.For a more complete description of our asset ticker naming conventions, please see What asset ticker naming conventions does Coin Metrics use?Can the daily number of unique buyers be derived from DEX swaps data?
Can the daily number of unique buyers be derived from DEX swaps data?
- Initiator is the Ethereum address which submitted the transaction, as a result of which the swap occurred
- Sender is the Ethereum address that invoked the liquidity pool smart contract’s function for swapping
- Beneficiary is the Ethereum address credited with the output tokens upon the completion of a swap
Market Data
How do I see what exchanges are supported?
How do I see what exchanges are supported?
/reference-data/exchanges endpoint and the metrics available for each exchange can be found by querying our /catalog-v2/exchange-metrics and /catalog-v2/exchange-asset-metrics API endpoints.Market data for these exchanges is served through our market data endpoints.Can you explain your historical data coverage?
Can you explain your historical data coverage?
How does Coin Metrics ensure high levels of data quality and data integrity?
How does Coin Metrics ensure high levels of data quality and data integrity?
- Exchange coverage universe: While there are over 800 digital asset exchanges in existence, Coin Metrics has curated our exchange coverage universe to include only high quality exchanges with legitimate trading activity. The presence of fake volume and wash trading is widely acknowledged in the industry, and Coin Metrics has independently confirmed the findings of several prominent researchers who have studied this problem. When deciding whether to include an exchange in our coverage universe, we consult a series of qualitative and quantitative features that are described in our Candidate Market Guidelines and the Talos Exchange Scorecard. We also consult feedback from our institutional user base. The exchanges in our coverage universe are widely recognized by market participants and researchers who have studied the fake volume problem to be of high quality.
- Market data collection system: Our market data collection system is engineered to have high levels of redundancy and resiliency. We collect data from exchanges using two instances of each application each located in an independent data center. For certain data types, we collect data from an exchange’s HTTP endpoint and websocket endpoint simultaneously as an added redundancy measure. CM utilizes multiple proxy servers to ensure that rate limits imposed by some exchanges do not impact data collection. Each server that hosts our market data collection system has local database storage as a fault tolerant measure in case of a failure in our primary database. These measures ensure high levels of availability for our market data collection applications and that no observations are missed.
- Multiple data centers: Coin Metrics utilizes two geographically-separated and vendor-independent data centers. Each data center contains an independent and complete collection of the infrastructure and applications needed to collect, process, and serve our data. In the case of failure of one of the data centers, our API will automatically failover to use our secondary data center with no action needed to be taken by our users.
- Internal monitoring: A dedicated internal team of data quality and site reliability engineers monitor logs and telemetry from our servers, databases, and applications in real-time using a suite of dashboards and automated alerts. We also have dedicated monitoring to detect interruptions of service from an exchange, incidents reported by an exchange, or breaking changes to their API. This monitoring allows us to take swift corrective or mitigating action if necessary.
- External monitoring: Coin Metrics maintains a dedicated external monitoring application that continuously polls our API endpoints to detect for interruptions in the data or unexpected responses from our API. This monitoring application allows us to continuously test the health of the entire pipeline of our systems from the perspective of our users. It is deployed in an external environment that is independent from our data centers so that it would not be affected in case of any degraded performance in our systems. Our internal team of data quality and site reliability engineers monitor the alerts generated by this application and take swift corrective or mitigating action if necessary.
- Deployment process: Our deployment process is governed by a series of SOC 2-compliant policies that include code reviews, extensive testing, manual review and quality control of historical values, and approvals prior to release. We received our SOC 2 Type 1 certification from Deloitte in August 2021 in the areas of security, availability, and processing integrity and have maintained this certification for every year.
- Human review: For certain critical data types, we employ regular human review to detect anomalies and assess the quality of our data. For instance, our Reference Rates are reviewed by a dedicated staff member every day, 365 times a year, at 16:00 New York time. Our Reference Rates and other critical data types are checked for several issues, including timeliness, data anomalies, sufficient data inputs, and a comparison against external sources.
Is there a way to pull data for multiple markets (such as all the markets for a particular exchange) in one API call?
Is there a way to pull data for multiple markets (such as all the markets for a particular exchange) in one API call?
markets parameter will accept wildcards like exchange-* or exchange-*-spot or *-future or *-option. The wildcards will match any market which fits this pattern so users do not need to specify every individual market when querying data for multiple markets. The markets parameter will also accept a comma-separated string of individual markets.Is there a way to pull data for trading volume across a specific exchange or asset?
Is there a way to pull data for trading volume across a specific exchange or asset?
How should I avoid look-ahead bias when running simulated backtests?
How should I avoid look-ahead bias when running simulated backtests?
time is when an event happened on the exchange, generally using the exchange-reported timestamp, not when the observation became available to consume. We also offer collect_time (in websocket API endpoints) or database_time (served from HTTP API endpoints) to reconstruct what was knowable at a given instant. A few practical rules:- Gate on collection time, not event time. Admit an observation into the simulation only if its
collect_time(stream) ordatabase_time(HTTP) is at or before the simulated decision time, not merely itstime. - Match your research input to your production input. If production consumes the real-time stream, research against the stream (or against history filtered to real-time availability), so the backtest does not benefit from trades or corrections that only historical collection captured.
- Treat recent candles as provisional. Coin Metrics produces candles from collected trades. While more than 99 percent of trades are collected in real-time with minimal latency, a small number of trades can be backfilled via a delayed process. Coin Metrics calculates candles in real-time and recalculates candles based on a fixed interval to include any delayed trades. Assume any candle within roughly three hours of your simulated “now” (about twenty minutes for options and decentralized-exchange spot) may still be recalculated, and do not rely on the finalized value being available at interval close.
- Allow for revisions and reorgs. Do not assume a very recent DeFi swap is final, and be aware that reference rates and indexes can be revised in limited historical windows.
- Account for latency, not just collection order. Even a live trade is not actionable at its exchange
time. it is actionable atcollect_timeplus your own downstream processing delay.
How do time, collecttime, and databasetime differ, and how do I tell when Coin Metrics became aware of an observation?
How do time, collecttime, and databasetime differ, and how do I tell when Coin Metrics became aware of an observation?
time— the exchange-reported event time: when the trade executed, the quote changed, or the snapshot was taken, according to the venue. This is the timestamp you index on for analysis, but it says nothing about when the observation became available.collect_time— the feed-handler host clock at the instant the message was received from the exchange. Present on the real-time websocket feed. This is the earliest moment Coin Metrics was aware of the observation.database_time— when Coin Metrics persisted the observation to its database. Returned by the HTTP (historical) endpoints. This is when the observation became queryable through the API.
time is the wrong field to gate on. Take the appropriate collection timestamp instead: for anything queried over HTTP, the observation was knowable at database_time; for anything consumed from the real-time stream, Coin Metrics received it at collect_time, typically milliseconds after the exchange’s time. A point-in-time-correct backtest for decision time T admits only observations whose collection timestamp is at or before T, not those whose time is at or before T. The difference between the two is exactly the window in which look-ahead bias hides.Latency varies substantially by venue: the fastest exchanges deliver in single-digit milliseconds, most sit in the tens to low hundreds of milliseconds, and the slowest reach a few hundred milliseconds at the median. For venues that do not stamp their own messages, time is Coin Metrics’ receive time, so there is no independent event time to compare against. See the Market Trades and Market Order Books pages for per-data-type latency figures.Why can the real-time trades stream contain fewer trades than the historical dataset?
Why can the real-time trades stream contain fewer trades than the historical dataset?
- Real-time collection — feed handlers connected directly to each exchange’s websocket record every trade as it happens.
- Historical collection — where an exchange permits querying past trades, Coin Metrics backfills that history.
collect_time), or accept that the live feed will occasionally be marginally thinner than history. The gap is small but non-zero, and it is largest exactly when it matters most — during volatile periods when disconnects and late arrivals cluster.How does candle recalculation create look-ahead bias?
How does candle recalculation create look-ahead bias?
What other Coin Metrics behaviors can introduce look-ahead bias?
What other Coin Metrics behaviors can introduce look-ahead bias?
- Backfilled market history. When a new market lists, or when Coin Metrics extends a market’s history from an exchange’s historical endpoint, that data lands in the dataset stamped with its original
timebut was not available in real time. A market’s early history can therefore appear “as if” it had always been there. Coverage and backfill depth vary by exchange. - Near-tip DeFi trades and reorgs. For decentralized-exchange swaps, a very recent trade near the chain tip can be removed by a block reorganization. The
min_confirmationsparameter trades latency against this risk. A backtest that consumes swaps at the chain tip assumes finality that did not yet exist. - Order book snapshot timing. Snapshots are stamped with the exchange event
timeof the most recent update captured before the round 10-second or hourly mark, whiledatabase_timeis stamped when the snapshot is persisted just after the mark. Gate on the collection timestamp, not the snapshottime, when reconstructing availability. - Derived and aggregated metrics run on delayed inputs. Metrics built on top of candles or reference rates inherit their delays. For example, basis metrics are recalculated from delayed candle data, so an observation reflects the market as of a short delay before its timestamp rather than the live tick.
- Reference rate and index revisions. Reference rates and single-asset indexes have published revision policies under which Coin Metrics occasionally recalculates limited portions of history to improve quality. A revised historical value differs from the value originally printed at that timestamp. See the Coin Metrics Prices Policies.
- Reference-data changes applied retroactively. Asset and market metadata (for example, an asset ticker rename) can change and be reflected across historical records. An analysis keyed on a current identifier may not correspond to how that market was labeled at the historical moment being studied.
- Gapless candles. A flat, zero-volume candle is filled forward from the prior close to give an unbroken series; it represents no trading in the interval, not a live observation. Do not treat the carried-forward price as a fresh market print.
How do I interpret the value of your volume metrics like volumereportedspotusd1d?
How do I interpret the value of your volume metrics like volumereportedspotusd1d?
coinbase-btc-usd-spot.- For our volume metrics served through
/timeseries/asset-metrics, the volume for a given asset (likebtc) represents the sum of the volume from markets where the asset is either the base or quote. - For our volume metrics served through
/timeseries/pair-metrics, the volume for a given pair (likebtc-usd) represents the sum of the volume from markets that contain the given pair. - For our volume metrics served through
/timeseries/exchange, the volume for a given exchange (likecoinbase) represents the sum of the volume from all markets on the given exchange. - For our volume metrics served through
/timeseries/exchange-assets, the volume for a given exchange-asset (likecoinbase-btc) represents the sum of the volume from all markets on the given exchange where the given asset is either the base or quote.
binance_aggregate) futures markets to prevent double-counting, as its data is identical to Binance’s futures except it is reported at a different level of aggregation. We also exclude any Uniswap v3 Ethereum aggregate markets (any market with the agg from our volume metrics, such as uniswap_v3_eth-agg-weth-usdt_eth-spot) to prevent double-counting, as these markets represent an aggregation of individual pools containing the same pair of assets.Are the volume metrics like volmereportedspotusd1d double counted?
Are the volume metrics like volmereportedspotusd1d double counted?
coinbase-btc-usdt-spot will be included in both the volume for btc and usdt because btc is the base asset of the market and usdt is the quote asset of the market. Some users have asked whether this represents double counting of volume.This is a convention that we use that is widely adopted by other data providers. The reasoning behind our choice can be best illustrated using a simple example.Suppose the world consists of only one market, coinbase-btc-usdt-spot and two assets, btc and usdt. A trader purchases btcby selling usdt in a transaction worth $100 U.S. dollars. If we only included markets where the given asset is the base currency, then the volume for btc would be $100 but the volume for usdt would be $0.In reality, both $100 worth of btc and $100 worth of usdt were exchanged. So we report the volume for both btc and usdt to be $100.Why do pair candles not include volume?
Why do pair candles not include volume?
Why are pair candles calculated from CM Reference Rates instead of trades?
Why are pair candles calculated from CM Reference Rates instead of trades?
Why are some of the values null for market metrics liquiditydepthpercentvolume?
Why are some of the values null for market metrics liquiditydepthpercentvolume?
liquidity_depth_1_percent_bid_volume_usd and liquidity_depth_1_percent_ask_volume_units are designed to measure the sum of all orders on a given side of the order book for a given percent away from the midprice in units of U.S. dollars or in native units.Exchanges differ in the amount of order book depth provided through their API. Some exchanges offer the full book depth while others only expose part of the order book such as the top 20 levels. The approximate maximum depth observed for each venue is tabulated under Limitations on the Market Order Books page.In calculating our liquidity depth metrics, we were forced to decide how to represent metric values when the reported order book depth is insufficient to calculate the depth for a given percent away from midprice. In these situations, we decided to represent this as a null value so that it is transparent to the user.Please note that for markets with high liquidity, even exchanges with relatively large order book depth of say 5,000 levels will only consist of prices less than 1 percent away from midprice. Therefore, it is common for many exchanges to have metrics with null values unless the exchange reports full order book depth.How do I interpret the volume for futures markets and convert volume to U.S. dollars?
How do I interpret the volume for futures markets and convert volume to U.S. dollars?
/timeseries/market-trades endpoint for market cme-BTCN4-future as an example."amount": "2" means that two contracts of cme-BTCN4-future were exchanged in this trade. The amount for other data types such as quotes, order books, candles, open interest, and liquidations are similarly in contract units.According to the contract specifications for this futures market, one contract is equal to 5 BTC of notional value. This can be seen in our /reference-data/markets endpoint."contract_size": "5" and "size_asset": "btc" define the contract size. The contract size is unique to each futures market and other markets may have different contract size.To convert a futures market volume in contract units to U.S. dollars, the following formula can be used: [amount in contract units] * [contract size] * [U.S. dollar price of contract size asset].How do I reproduce the CME volume figures published on CME's website?
How do I reproduce the CME volume figures published on CME's website?
/timeseries/market-trades endpoint.CME’s published volume figures include the futures contract itself, such as cme-BTCM5-future, as well as any calendar spread where the future is the short leg or long leg, such as cme-BTCM5-BTCN5-future. Therefore, users should query all three types of futures contracts using a markets parameter such as markets=cme-BTCN5*-future,cme-*BTCN5-future.The time period that CME uses is from 17:00 to 16:00 America/Chicago time, and the timestamp they assign to the volume observation is the end-of-day timestamp. Therefore, to reproduce the published volume figure on 2025-06-05 , users should query all trades from start_time=2025-06-04T17:00:00Z, end_time=2025-06-05T16:00:00Z with timezone=America/Chicago.Finally, CME’s published volume figures include both the CME Globex venue, their electronic trading platform, and PNT Clearport, which represents OTC transactions that are negotiated outside of CME’s electronic order book and later submitted to CME for clearing. Coin Metrics collects data from Globex only.What is the difference between the closing price of a CME futures contract from Coin Metrics's candles and the figure published on CME's website?
What is the difference between the closing price of a CME futures contract from Coin Metrics's candles and the figure published on CME's website?
When is the data for Data Quality collected and reviewed?
When is the data for Data Quality collected and reviewed?
Would you consider including as part of the criteria?
Would you consider including as part of the criteria?
Does the Talos Exchange Scorecard include futures exchanges?
Does the Talos Exchange Scorecard include futures exchanges?
Is there a certain score level that separates trustworthy from not?
Is there a certain score level that separates trustworthy from not?
Why are certain subscores assigned different weights?
Why are certain subscores assigned different weights?
How should we interpret the scores?
How should we interpret the scores?
Can the list of trusted exchanges be accessed via the API?
Can the list of trusted exchanges be accessed via the API?
What's the difference between constituent market selection and the Talos Exchange Scorecard?
What's the difference between constituent market selection and the Talos Exchange Scorecard?
Why was exchange given this score?
Why was exchange given this score?
Prices
How do you calculate the CM Reference Rates?
How do you calculate the CM Reference Rates?
Is there a difference between the metrics ReferenceRate and ReferenceRateUSD?
Is there a difference between the metrics ReferenceRate and ReferenceRateUSD?
ReferenceRate and ReferenceRateUSD are identical. When we added reference rates quoted in other currencies like ReferenceRateEUR, we also added ReferenceRateUSD and preserved ReferenceRate for backward compatibility.Why do you use non-USDT-USD pairs to calculate the USDT-USD price?
Why do you use non-USDT-USD pairs to calculate the USDT-USD price?
What is the expected latency for the 1s frequency of Reference Rates served through your websocket API?
What is the expected latency for the 1s frequency of Reference Rates served through your websocket API?
When pulling ReferenceRateUSD and ReferenceRateEUR along with other metrics (like PriceUSD or FeeMeanNTV), the reference rates are updated with data up to the current date while other metrics are only updated to yesterday's date. Why are these data fields assigned different dates?
When pulling ReferenceRateUSD and ReferenceRateEUR along with other metrics (like PriceUSD or FeeMeanNTV), the reference rates are updated with data up to the current date while other metrics are only updated to yesterday's date. Why are these data fields assigned different dates?
PriceUSD, use the “start-of-interval” timestamp, which represents the beginning of a time interval. The majority of our network data metrics represent a summary statistic over a daily time interval.For example, suppose our FeeMeanNTV metric, which represents mean fees over an interval of a day, has a timestamp of 2020-12-10 00:00:00. The FeeMeanNtv value represents the mean of fees that occurred from 2020-12-10 00:00:00 to 2020-12-10 23:59:99.999999.Other metrics, like ReferenceRate (and many other data types like trades, open interest and order book), use the “point-in-time” timestamp convention, where timestamp is set to the specific timestamp of the measurement or event. This is because we want to generate a price at a specific point in time.When you compare something that uses the “point-in-time” convention with something that uses the “start-of-interval” convention, it can seem like the “start-of-interval” timeseries is lagged or is potentially missing data. I can assure that this is not the case, but we certainly understand your confusion here since the different values are delivered via the same timeseries/asset-metrics endpoint.For more information, please see What timestamp conventions does Coin Metrics use?What is the difference between your Reference Rates and a volume-weighted average price?
What is the difference between your Reference Rates and a volume-weighted average price?
What do the frequency parameters 1d and 1d-ny-close mean?
What do the frequency parameters 1d and 1d-ny-close mean?
1d and 1d-ny-close.The 1d frequency represents a daily frequency that ends at 00:00:00 in the UTC timezone and the 1d-ny-close represents a daily frequency that ends at 16:00:00 in the America/New_York timezone. These timestamps are not altered for weekends or holidays.Please also see our FAQ on What timestamp conventions does Coin Metrics use? for more information.Why does the candles closing price differ from the ReferenceRate metric or PriceUSD metric or an index value?
Why does the candles closing price differ from the ReferenceRate metric or PriceUSD metric or an index value?
PriceUSD use the beginning-of-interval timestamp convention while ReferenceRate and index values use the point-in-time timestamp convention.For more discussion on these timestamp conventions, please see the frequently asked question What timestamp conventions does Coin Metrics use?.Indexes
How are the CMBI Single Asset Index and CMBI Multi Asset Index levels calculated?
How are the CMBI Single Asset Index and CMBI Multi Asset Index levels calculated?
What's the difference between your reference rate and a single-asset index for the same asset?
What's the difference between your reference rate and a single-asset index for the same asset?
Is there an index equivalent of the metric AssetEODCompletionTime? When should I expect an end-of-day index value to be published?
Is there an index equivalent of the metric AssetEODCompletionTime? When should I expect an end-of-day index value to be published?
What are the constituent exchanges included in the CMBI Bitcoin Index, along with details such as the domicile, regulation and legal compliance?
What are the constituent exchanges included in the CMBI Bitcoin Index, along with details such as the domicile, regulation and legal compliance?
What are the criteria considered for an exchange to become a constituent exchange?
What are the criteria considered for an exchange to become a constituent exchange?
The methodology states that the index does not utilize data from over-the-counter markets or derivatives platforms but may do so in the future. What are the factors in determining whether to utilize data from the over-the-counter markets or derivative platforms?
The methodology states that the index does not utilize data from over-the-counter markets or derivatives platforms but may do so in the future. What are the factors in determining whether to utilize data from the over-the-counter markets or derivative platforms?
Reference Data
How do you classify the assets?
How do you classify the assets?
How often is the taxonomy structure reviewed?
How often is the taxonomy structure reviewed?
What is the datonomy Advisory Board and who is on it?
What is the datonomy Advisory Board and who is on it?
Token Migrations and Rebrandings
How does Coin Metrics handle token migrations, swaps, and rebrandings?
How does Coin Metrics handle token migrations, swaps, and rebrandings?
usdt_eth and usdt_sol existing as separate assets — different on-chain forms are tracked independently.Case 2 — A pure rebranding (name and/or ticker change only). If there is no on-chain migration and the change is purely cosmetic — the project renames the asset and/or its ticker, but the underlying token, contract, blockchain, and units are unchanged — then Coin Metrics does not create a second asset. We simply relabel the existing asset: the display name and ticker are updated, and all data that has ever been produced for that asset remains available without interruption. Units map 1:1 across the change because it is the same token.To avoid breaking existing integrations, the legacy ticker remains an accepted alias and continues to resolve in API queries after a rebranding.If you are unsure which case applies to a specific event, we publish the details for each migration or rebranding to our Status Page, and you can always reach out to us at info@coinmetrics.io.When an asset is renamed or rebranded, what happens to my historical data across each data type?
When an asset is renamed or rebranded, what happens to my historical data across each data type?
- Reference Data (assets, markets, pairs) — ✅ Seamless. Relabeled to the new ticker; identifiers move from the old ticker to the new one. Legacy identifiers continue to resolve.
- Market Data (spot trades, candles, quotes, order books) — ✅ Seamless. Continuous history; queryable under both the new and legacy identifiers.
- Principal Market Price — ✅ Seamless. Continuous history.
- Indexes (incl. CMBI Single- and Multi-Asset Indexes) — ✅ Seamless. Index levels continue uninterrupted; the renamed asset replaces its prior name as a constituent.
- Reference Rates — ⚠️ Split at cutover. The new series begins at the effective time; earlier history remains available under the legacy ticker until a later unification.
- Asset & Market Metrics — ⚠️ Split at cutover. The new ticker applies from the effective time onward; pre-rebrand history remains under the legacy ticker.
- Derivatives (futures, perpetuals) — 🆕 New series. New contracts appear as venues relist them under the new name; existing contracts remain as separate historical series.
- Update any display labels for the asset in your applications.
- For a continuous Reference Rate or Metrics series that spans the rebrand, query the legacy ticker for data before the effective time and the new ticker from that time forward. We will notify you once the two series are unified.

