Cloud Delivery

Cloud Delivery is in limited availability. To request access, reach out to your sales or account manager.

Overview

Cloud Delivery publishes Coin Metrics market data as Apache Parquet files in Amazon S3, refreshed daily, and exposes the same files as tables in Snowflake. Rather than paging through an API, you read whole datasets directly from object storage or query them in your warehouse. Quantitative researchers, data engineers, and analytics teams use it to load multi-year histories of trades, candles, quotes, and order books into their own environment once, then work with the data locally.

The same objects back both destinations. Choosing S3 or Snowflake changes how you reach the data, not what the data contains.

At a Glance

Delivery targets	Datasets	Update cadence	Format	Partitioning	Access model
Amazon S3 and Snowflake	Trades, candles, quotes, order books, reference rates, and reference data	Daily, covering the previous UTC day	Apache Parquet, Snappy compressed	Hive-style key=value prefixes by market type, exchange, and date	Cross-account IAM for S3. Secure Data Sharing for Snowflake

When to use Cloud Delivery

Cloud Delivery and the API serve different shapes of work. The API is the right tool for interactive queries, narrow slices, and anything real time. Cloud Delivery is the right tool when the volume of data makes repeated API calls impractical.

Use Cloud Delivery when	Use the API when
You want a full history of a dataset in your own storage or warehouse	You need a specific market and time range on demand
You are loading data into Spark, Snowflake, DuckDB, Athena, or a data lake	You are building an application that queries interactively
Your volumes make repeated API paging impractical	You need real-time data over websockets
You want columnar files you can scan and filter locally	You want the newest observation within seconds

Cloud Delivery does not replace API access. Most customers use both.

Datasets

Ten tables are published. Each maps to one prefix in S3 and one table in Snowflake. Field definitions live on the existing dataset pages, since the columns delivered here are the same ones the API returns.

Dataset	S3 prefix	Snowflake table	Field reference
Market trades	data-type=market-trades/	market_trades	Market Trades
Market candles	data-type=market-candles/	market_candles	Market Candles
Market quotes	data-type=market-quotes/	market_quotes	Market Quotes
Order book snapshots	data-type=market-orderbooks/…/dataset=snapshots/	market_orderbooks_snapshots	Market Order Books
Order book updates	data-type=market-orderbooks/…/dataset=updates/	market_orderbooks_updates	Market Order Books
Reference rates	data-type=reference-rates/	reference_rates	Reference Rate
Reference data: markets	data-type=reference-data/dataset=markets/	reference_data_markets	Market Reference Data
Reference data: assets	data-type=reference-data/dataset=assets/	reference_data_assets	See below
Reference data: exchanges	data-type=reference-data/dataset=exchanges/	reference_data_exchanges	See below
Reference data: pairs	data-type=reference-data/dataset=pairs/	reference_data_pairs	See below

Schema differences

Two tables carry fewer fields than the API returns for the same dataset. Everything else matches the linked pages field for field.

Table	Fields not delivered	Effect
market_trades	mark_price, index_price, iv_trade, liquidation	Futures and options trades arrive without their derivatives-specific fields. Spot and decentralized exchange trades are unaffected.
reference_data_markets	pool_config_id, fee, price_includes_fee, variable_fee, base_address, quote_address, multiplier_size	Decentralized exchange pool metadata is absent from the market catalog.

These fields remain available over the API if you need them.

Reference data schemas

The assets, exchanges, and pairs catalogs each carry two columns, matching the /reference-data/assets, /reference-data/exchanges, and /reference-data/pairs endpoints.

Table	Column	Type	Description
reference_data_assets	asset	string	Asset identifier, for example btc.
	full_name	string	Display name of the asset, for example Bitcoin.
reference_data_exchanges	exchange	string	Exchange identifier, for example coinbase.
	full_name	string	Display name of the exchange.
reference_data_pairs	pair	string	Pair identifier, for example btc-usd.
	full_name	string	Display name of the pair.

The markets catalog is much wider. See Market Reference Data for its fields.

Coverage

Cloud Delivery carries a curated subset of the Coin Metrics market data universe rather than every market on every venue. The scheduled datasets cover the following.

Dataset	Exchanges	Market types	Notes
Market trades	48	Spot, futures, options
Market candles	48	Spot, futures, options	Eight frequencies: 1m, 5m, 10m, 15m, 30m, 1h, 4h, 1d
Market quotes	29	Spot, futures	Raw granularity
Order book snapshots	29	Spot, futures	Full book depth, raw granularity
Order book updates	5	Spot, futures	Full book depth, raw granularity. Additional exchanges can be added on request
Reference rates	All covered assets		Frequencies 1s and 1h. Metrics ReferenceRateUSD, ReferenceRateEUR, ReferenceRateETH, ReferenceRateBTC
Reference data	All covered entities		Current catalog only, refreshed daily

Two of these are narrower than the corresponding product documented elsewhere in this knowledge base. Reference rates are delivered at two frequencies against four quote currencies, where the Reference Rate product covers more of both. Market quotes and order books cover spot and futures only, with no options.

For the market-level universe behind each dataset, see the coverage tool. If a market, frequency, or quote currency you need is not in the delivered set, your account manager can arrange a backfill, and the API carries the full set in the meantime.

Reference data has no history. The four reference-data catalogs are current snapshots that are overwritten each day. They carry no date partition, so you cannot reconstruct the catalog as it stood on a past date. Retain your own copies if you need point-in-time catalog history.

Update schedule

Every dataset runs once a day and covers the previous UTC day. Jobs start on a stagger to spread load.

Dataset	Start time (UTC)
Reference data	00:00
Reference rates	00:08
Market trades	00:10
Market quotes	00:20
Market candles	00:30
Order book snapshots	00:45

A file's dt partition is the UTC date of the data it contains, not the date the job ran. Data for dt=2026-07-28 is produced by the run that begins early on 2026-07-29. Larger datasets take longer to complete, so allow for processing time after the preceding start times before expecting a full day to be present.

File format

Every object is an Apache Parquet file with Snappy compression applied inside the file, which is the Parquet default and is handled transparently by every Parquet reader. There is no outer .gz wrapper, so files end in .parquet and can be opened directly.

Numeric fields are stored as native Parquet types rather than the JSON strings the API returns. Timestamps are stored as nanosecond epoch values in UTC.

Getting access

Access is provisioned per customer. To get started, reach out to your sales or account manager and tell them which destination you want.

1. Confirm your entitlements and the datasets you need.

2. Provide the details for your destination. For S3 this is your AWS account ID and, preferably, an IAM role ARN. For Snowflake this is your Snowflake organization and account identifier.

3. Coin Metrics grants read access and confirms the bucket, prefixes, or share name you have been given.

4. Validate the connection using the examples on the destination page below.

Choose your destination

Amazon S3Snowflake

Related

• Downloading Our Data: browser and spreadsheet tools for smaller extracts.

• API Conventions: the HTTP API, for interactive and real-time access.

• Status Page: incident and change notifications.