Guide

Crypto historical data: funding, open interest and liquidation history

Where crypto funding rate, open interest and liquidation history comes from, why each dataset starts on a different day, and how to get it as CSV or API.

Blockmetric ResearchUpdated 6 min read

On this page
  1. Two kinds of history
  2. What history exists for each kind of data
  3. Point-in-time data and gaps
  4. Three ways to get the data
  5. What to check before using historical data
  6. Limits of the data

Crypto derivatives history comes from two kinds of sources: what a venue publishes about its own past, and what someone recorded live while it happened. Funding rates and price candles can usually be loaded afterwards; open interest, liquidations and positions often cannot, so their history starts when someone began to collect them. That is why every dataset on our historical data page has its own start date.

This guide explains which history exists for which data, why the starts differ, what "point-in-time" means, and the three ways to get the data: a CSV per coin, the API and daily archive files.

Two kinds of history

On the historical data page every dataset is labelled with one of two origins.

  • Reconstructed history is what was recorded in the past and loaded into our database afterwards. Funding rates, price candles and part of the open interest and long/short history come this way. The daily snapshot of every open Hyperliquid position, which is real chain state, is imported every day too.
  • Collected by us is what our collector recorded live: the positions of tracked wallets, whale moves and stops, the live liquidation heatmap and the liquidation streams of exchanges that publish no liquidation history.

The difference decides what can ever be filled in. If a venue serves its past, a missing stretch can be loaded later. If the data only exists as a live state, such as the positions a wallet held at a given minute, nobody can read it afterwards.

What history exists for each kind of data

DataWhere it comes fromWhy it starts where it starts
Funding ratesReconstructed funding history of each venue, then our own collectionVenues publish past funding, so the series reaches back into each venue's own history
Open interestReconstructed history where a venue serves it, then our own collectionBinance publishes only recent open interest history, so its series starts later; Bybit's reaches back years
Long/short ratiosEach venue's published ratios; Hyperliquid from positionsBinance publishes only recent ratios, Bybit's reach back years; Hyperliquid's ratio from tracked wallets starts with our collection
LiquidationsThe Hyperliquid fill record; exchange streams collected liveBinance and Bybit publish no liquidation history, only a live stream; the Hyperliquid fill record has no liquidation field before a certain day
Liquidation heatmap, dailyThe daily snapshot of every open Hyperliquid positionStarts with the first stored daily snapshot
Liquidation heatmap, intradayLive snapshots of tracked positionsStarts with our live collection on 2026-11-03 at 14:15 UTC
Positions and whale movesPolled livePast positions of a wallet cannot be loaded afterwards

The exact first day of each row, per venue where they differ, is read from the database and shown on the historical data page.1 We do not repeat those dates here, because some of them move back as more reconstructed history is imported.

For the heatmap, the two rows work together. Before the live collection, and in any gap after it, the daily snapshot of all positions shows where liquidation levels sat once a day. After it, intraday columns show the tracked wallets snapshot by snapshot. The guide How to read a liquidation heatmap explains the chart itself.

Point-in-time data and gaps

Two rules decide whether a historical series can be trusted for research or a backtest.

Point-in-time. A point-in-time value is the one that was known at that moment, not a corrected version written later. Our daily archive files freeze every UTC day once it is complete; a later correction is never silent, it is versioned in the manifest of the day with checksums. A backtest built on them sees what a trader could have seen on that day.

Gaps stay gaps. When a stream was silent or our collector did not run, the stretch is left empty and listed with its exact times and the documented reason. Nothing is filled in or estimated in its place, and heatmap columns are never interpolated between snapshots. A chart with a visible hole is more honest than a smooth line that was partly invented.

Three ways to get the data

1. CSV per coin. The History tab of every coin shows one row per UTC day with the close, open interest, funding and long and short liquidations, for example BTC historical data. Next to it are files for single liquidations, open interest, long/short and the heatmap. The free plan shows recent rows on the page; CSV and JSON downloads at full resolution come with the paid plans, within each plan's history.

2. The API. JSON over REST with an API key, as far back as the plan allows. Heatmap columns state whether they are a live snapshot or the daily snapshot of all positions, and liquidation history can be paged through with a cursor. The guides Crypto liquidation data API and Funding rate and open interest API show the requests; the developer overview has the quickstart.

3. Daily archive files. Every UTC day is frozen once into a set of datasets as gzip-compressed CSV and JSON lines (NDJSON) files, with a manifest.json and a SHA256SUMS file to check each download. A free sample day for the free heatmap coins, at least a week old and without per-wallet data, is linked from the historical data page. The complete archive, including per-wallet positions and stops, is available online as a dataset licence on Data pricing: one payment for 12, 24 or 36 months of access, with the full history to date on day one and every new day during the term, for internal use within one organisation (businesses only). Next to the point-in-time files, the licence includes reconstructed history: recorded values compiled afterwards, never modelled, labelled as reconstructed and not point-in-time. Redistribution and custom delivery are agreed in an Enterprise contract.

How deep each plan reaches is a cap on top of what is stored: a plan with years of history still starts where the data starts. The historical data page has a table of history per plan and the depth of each heatmap interval.

What to check before using historical data

  • The start of each venue. A series summed over venues jumps when a venue joins. Compare periods only where the same venues are covered.
  • The unit. Funding is stored as an hourly rate; venues that settle every few hours are converted so they can be compared. Open interest is in USD; long/short ratios count accounts, top traders or positions depending on the venue.
  • The origin. Reconstructed history and our own collection can differ in resolution. Hourly sums from an archive and single events from a live stream answer different questions.
  • Exchange data as collected. Values per exchange (Binance, Bybit) are in the API and downloads as collected, without warranty. We may withhold a venue at any time; sums over venues stay available.

Limits of the data

  • Live-only data starts late. Positions, whale moves and the intraday heatmap begin with our own collection and cannot be extended backwards.
  • Tracked wallets. Intraday heatmap and Hyperliquid long/short from positions cover the wallets we poll, which includes the large ones far better than the small ones. The daily snapshot covers every position, but only once a day.
  • Lower bounds. Exchange liquidation streams publish only part of the liquidations, so exchange totals undercount.
  • Retention. One-minute heatmap columns are kept for a limited number of days, and older heatmap history is stored at coarser intervals. The depth of each interval is listed on the historical data page, and the methodology page describes how the data is checked.

The historical data page is the one place that lists what is stored and since when, with a timeline of every dataset and its gaps.

Sources

  1. Blockmetric, "Historical data" (accessed 2026-11-11)Back to text

How we collect and check the data: Data & methodology.

See today’s numbersThe live pages update continuously.

See what history is stored
LiveHyperliquid markets 7sOperationalHyperliquid positions & whales 5sOperationalLiquidation heatmap & levels 12sOperationalLighter markets & liquidations 24sOperationalLighter positions & heatmap 6sOperationalLighter RH positions & heatmap 10sOperationalBinance markets 59sOperationalBybit markets 14sOperationalBinance liquidations 22sOperationalBybit liquidations 22sOperationaldYdX positions & heatmap 18sOperationalGMX positions & heatmap 19sOperationalLong/short & open interest history 54sOperationalOrder books & trade flow 5sOperationalOn-chain transfers 14sOperationalSecurity monitor 6sOperationalBlockmetric Sentiment 58sOperationalAgency releases 6mOperationalAlert delivery 11sOperationalDataset files 7mOperational16:14:17 UTCData: Blockmetric Research