Skip to main content
The Coin Metrics Market Data Feed (MDF) provides historical and real-time data from the world’s leading centralized and decentralized spot and derivatives crypto exchanges. Coin Metrics collects data directly from each venue, harmonizes the differing formats into a single schema, and publishes it alongside a set of metrics calculated on top of it. This section documents every data type and metric family in the feed. If the feed is new to you, read the three orientation sections below first. They cover the one distinction (raw data versus metrics) and the one naming convention (entities) that together determine which endpoint you call and what you pass to it, followed by the discovery loop that almost every query against the API follows. After that, jump to the group that matches what you are trying to measure.

Raw Data and Metrics

Every page in this section documents one of two things, and the difference decides which endpoint you query. Raw data is what a venue reported, harmonized but not aggregated. A trade is one trade, an order book is one book, a funding rate is one venue’s rate for one contract. Raw data is always served per market, and it arrives either event-driven (one record per event, as with trades and liquidations) or as a snapshot on a fixed cadence (as with order books and open interest). It lives on the /timeseries/market-* endpoints, and most types also have a matching /timeseries-stream/market-* websocket for real-time delivery. Metrics are values Coin Metrics calculates from that raw data. They aggregate across markets, convert to a common unit (usually US dollars), and publish on a fixed frequency such as 1m, 1h, or 1d. Aggregation is what makes venues comparable: adding the activity of binance-BTCUSDT-future to that of coinbase-btc-usd-spot first requires converting contracts and base-asset amounts into one unit. Metrics live on the /timeseries/<entity>-metrics endpoints. Use raw data when you need to see or reconstruct exactly what happened on one venue. Use metrics when you need a number that is comparable across venues, assets, and time.

Entities

An entity is the thing a row of data describes, and its identifier is what you pass to an endpoint. Raw data is published for markets, and metrics are published at whichever levels the aggregation is meaningful for. A market identifier encodes the venue, the instrument, and the market type. Spot markets follow <exchange>-<base>-<quote>-spot, as in coinbase-btc-usd-spot. Futures and options append -future or -option to the exchange’s own symbol, as in binance-BTCUSDT-future and deribit-BTC-16MAY21-58000-C-option, so the middle segment is exactly what the venue publishes rather than a Coin Metrics rewriting of it. Prediction markets do the same with -event_contract, as in kalshi-KXFEDDECISION-27DEC-H25-event_contract. Exchange-asset and exchange-pair look alike and are not interchangeable. deribit-btc covers every Deribit market whose base asset is BTC. deribit-btc-usd covers the BTC-USD pair on Deribit specifically, which keeps option chains on the same underlying quoted in different currencies from being mixed together.

The Discovery Loop

Almost every query follows the same three steps. Find out what exists, check the time range, then pull the data. The steps below are the market data reading of that pattern, which holds across the whole API: The Discovery Loop documents it in general terms, including which endpoints sit outside it. 1. What exists? The /reference-data/* endpoints describe the coverage universe, and they come in two flavors. The entity endpoints (/reference-data/markets, plus assets, exchanges, pairs, and indexes) return which entities exist and what their attributes are, such as trading rules and contract terms. The metric endpoints (one per entity level, for example /reference-data/asset-metrics) return which metrics exist, along with their descriptions, units, and frequencies. 2. Over what time range? The /catalog-v2/* endpoints list every entity that supports a given data type together with the min_time and max_time of the data available for it. There are two variants. /catalog-v2/* reflects what your API key is entitled to, while /catalog-all-v2/* reflects the entire Coin Metrics dataset, so comparing the two shows what sits outside your subscription. 3. Pull the data. The /timeseries/* endpoints serve history over HTTP, and /timeseries-stream/* serves the same data as a real-time websocket feed for the data types that support one. Charting Coinbase BTC-USD trades runs the loop like this: The Coin Metrics Coverage Tool presents the same universe and time ranges visually, which is often faster than the first two calls when you are exploring rather than automating. API Conventions documents every endpoint family in full, including paging, response formats, and timestamp handling.
  • Market Reference Data: the identity, trading rules, and contract terms of every market Coin Metrics collects, including prediction market contracts, and the way to build or filter the universe of markets you want to query.
  • Prediction Market Event Reference Data: the title, series, status, and outcome structure of every Kalshi and Polymarket event, and the key that groups prediction markets under the question they answer.
Prediction markets from Kalshi and Polymarket are listed in Market Reference Data and grouped by event in Prediction Market Event Reference Data. Their trades, candles, and open interest are served by Market Trades, Market Candles, and Market Open Interest when requested with type=event_contract.

Trades, Candles, and Volume

What traded, at what price, and how much. Trades are the tick-level record, candles summarize them over fixed intervals, and volume metrics aggregate them into comparable US dollar totals.

Order Books, Quotes, and Liquidity

What is resting on the book and what it would cost to trade against it. These datasets describe available liquidity rather than executed activity, so they answer questions about execution cost that trades and candles cannot.

Futures and Perpetuals

Positioning, cost of carry, and forced deleveraging on derivatives venues. The raw per-market series a venue reports come first, followed by the metrics that aggregate them across markets and convert them to US dollars. Most concepts appear in both halves, so open interest, funding, and liquidations can each be read per contract or per asset.

Options and Volatility

Contract-level option data as venues report it, plus the derived surfaces that make it comparable through time. Realized volatility sits here as well, because it is most often read against implied volatility at a matching horizon.

Institutional Products

Fund-level data for institutional digital asset investment products, which sits outside the exchange coverage universe and is published on its own endpoint.

Exchange Coverage

Coin Metrics collects from dozens of the leading spot and derivatives venues, and the set changes as venues are added, renamed, or wound down. Rather than repeat a count here, consult the sources that track it directly.
  • The Coin Metrics Coverage Tool shows the exchanges covered, the number of spot, futures, and options markets on each, and the history available.
  • /reference-data/exchanges returns the same universe programmatically.
  • Each data type page carries its own coverage link, because not every venue supports every data type. Historical order book updates, for example, are served only for supported exchanges, while trades are collected from every venue in the universe.
  • CM Prices: the Coin Metrics Reference Rates and Principal Market Prices, the cross-exchange price series behind pair candles, realized volatility, and the US dollar conversions used throughout the metrics.
  • Talos Exchange Scorecard: the quantitative evaluation of exchange quality that defines the trusted exchange set used by volume metrics.
  • Decentralized Exchange Data: swap coverage on supported decentralized exchanges, which flows into market trades and market candles.
  • Access Our Data: the HTTP API, websocket API, Python client, and cloud delivery options for pulling everything in this section.
  • Tutorials and Examples: worked notebooks and guides that build on these datasets.