We're building the most comprehensive crypto data API we can, and it's still growing — there's always a coin, a field or an endpoint we haven't got to yet.
Please tell us about any data you think Crypto Data API should nail but doesn't. We're data nerds who love to geek out about market data, trading bots, AI agents, or how to make your trading stack work in general. Prefer a form? Send feedback.
— Sam and the CDA team
Join us on DiscordArticles on crypto market intelligence, trading APIs, AI agents, derivatives data, and building algorithmic trading systems.
Glassnode's API sits behind a Professional plan that starts at $584 a month. The ten BTC cycle indicators most bots actually read are daily series from Coin Metrics' public history, and we serve them on a free key.
Read more →
Farside's table fills fund by fund, biggest issuers last. Read the newest row and you publish a partial sum as the day's flow. We did, on 43 of 56 days, until we rebuilt the series on settled trade dates.
Read more →
clearinghouseState tells you one address's book, exactly. It will not tell you which addresses to ask. Every whale feed on Hyperliquid starts by solving that, and most never say how.
Read more →
A flat chart with flat open interest is a quiet market. A flat chart with OI up 6% in four hours is a loaded spring. The two look identical on a candle, which is why the gap between them is the signal.
Read more →
Candles are final. Health scores, regime labels and breadth counts are not, unless someone froze them on the day. The daily snapshot archive did, at 20:00 UTC, and never rewrote them.
Read more →
A strategy playbook says 'enter when funding is crowded'. An agent needs the route, the field and the threshold. Two catalogues and the AlgoBrain page gates do that translation for all 317 strategies.
Read more →
A rate series is not carry paid. Hyperliquid settles every hour on the oracle price, its history API never returns that price, and a 5-minute sample summed naively reads 12x too high. Here is the series to use and the Python to page it.
Read more →
A stock agent on Robinhood gets 18 market-data tools. A crypto agent gets one: a quote. Here's what it can trade, what it can't, and the second MCP server that gives it something to decide with.
Read more →
Liquidations, funding, open interest, long/short, ETF flows, Fear & Greed: each CoinGlass product mapped to an endpoint, with the Hyperliquid per-fill tape CoinGlass does not have, and no tier marked personal use only.
Read more →
A CEX heatmap guesses where leverage sits from OI changes and assumed leverage buckets. On Hyperliquid the positions are on-chain, so a liquidation ladder is a sum, not a model. Here is the difference, field by field.
Read more →
Six crypto data MCP servers an agent can actually connect to today, what their tools return, which run locally and which are remote, and the one line that adds each to Claude Code.
Read more →
A 25-coin candle poller spends 500 weight a sweep against a 1,200-per-minute budget. Two sweeps and you are rate limited. Here is the weight table from Hyperliquid's docs, and the batch route that spends it once.
Read more →