Crypto & Prediction-Market Historical Data — Download Tick Files, Order Books & OHLCV

Most "historical crypto data" is a summary: candles, sampled book snapshots, trades backfilled from a REST endpoint. This archive is the feed itself. A recorder co-located at each venue subscribes to the complete public market-data stream and writes every message it receives — every order-book snapshot and incremental update, every trade print, every top-of-book quote where the venue publishes one, every funding, mark-price, index and liquidation event — with the venue timestamp and the receive timestamp in nanoseconds. Nothing is sampled, thinned or aggregated on the way to disk.

Completeness is verifiable, not promised: every event carries a prevEventId/eventId chain, so a gap in the recording shows up as a chain break, and the free reader's audit command (stats FILE --deep) reports parse errors, chain gaps, crossed book states and the covered time span of any file before you build on it. The 1-minute OHLCV, VWAP, spread and depth statistics in the analytics are computed from these files — a summary of the archive, never its source.

That is the difference to a data vendor that resells exchange candles or polls an API: you can rebuild the order book at any instant, replay a day tick by tick in the browser viewer, measure queue position, cancellations and latency, and aggregate to any bar size yourself — because the messages the exchange sent are all there.

All venues ranked, month by month and by instrument class →

Historical order-book data is the part of a market that no candle feed keeps: the full Level-2 evolution of every instrument, as one snapshot followed by every incremental update, each with the venue timestamp and our co-located receive timestamp in integer nanoseconds. Replaying it gives the book at any instant — depth at every level, spread, queue position, imbalance — for Binance, Bybit, OKX, Deribit and 30+ other venues, and for every Kalshi and Polymarket event contract.

Top-of-book is included where the venue publishes a separate BBO stream and derived by replay where it does not (BitMEX, Coinbase, Kraken spot, Kalshi and Polymarket record depth only). Derivatives files add liquidation events, funding and mark/index prices, so market depth history, liquidation history and trade flow come from the same file — one schema, documented in the format reference and the market-data specification.

The free Python reader turns a day file into a fixed-grid order-book dataset in one command (book FILE --every 1s --depth 20 --out book_1s.parquet) — the input the order-book dataset guide builds features and labels on.

Read more: historical order book data, depth-only venues and the live spread/depth snapshot →

Prefer candlestick data over raw ticks? Every instrument has a 22-column 1-minute statistics CSV computed from the same capture: OHLC, VWAP, trade counts by aggressor side, liquidation count, USD turnover with buy/sell split, average spread in price, bps and ticks, and top-of-book and top-20 depth in USD. The rolling last 24 h are free on every instrument's analyze page (?format=csv, no sign-up); Premium extends the export to a rolling 7-day window and to any recorded day.

For longer daily history, every asset page carries a daily OHLC and turnover table of the most-recorded pair since January 2025, and the Data Shop shows daily candles per instrument-day. Minute bars for older days come out of the tick files themselves — the free reader aggregates trades to any bar size locally, so a purchased day is also its own OHLCV source at every resolution.

time,open,high,low,close,vwap,trades,trades_buy,trades_sell,trades_liquidation,turnover_usd,turnover_buy_usd,turnover_sell_usd,spread_avg,spread_bps,spread_ticks,top1_bid_usd,top1_ask_usd,top20_bid_usd,top20_ask_usd,top1_bid_qty,top1_ask_qty

Read more: OHLCV and candlestick data, daily Bitcoin history and bar aggregation →

1. Find the instrument: search the Data Shop by pair or venue, browse by exchange, by asset or by trading pair, or open a Kalshi/Polymarket series on the prediction-market hub and an option chain on the options hub.
2. Try a free day first: full-day samples for major venues, the busiest prediction contract of the day and a Deribit option contract are on the downloads page — same bytes as the shop files, no account needed.
3. Pick the UTC days in the calendar and check out — card with invoice, or Premium and pack credits (1 credit = €1 = one instrument-day). Many instruments at once: "Select to buy" in the instrument scanner adds a shared date window for up to hundreds of instruments in one go.
4. Download: single files from the order page, whole bundle days as one ZIP, whole date ranges of a bundle as one resumable tar, or everything you own as a batch ZIP from your account. Days in deep archive restore in about 12 hours.
5. Convert as needed: ?format=trades.csv.gz, ?format=bbo.csv.gz, ?format=liquidations.csv.gz or the Parquet variants for trades and liquidations on any download link — or the free Python reader for order-book grids, funding series and greeks. Agents can install the Agent Skill or use the MCP server.

Purchased data is for your own use — backtesting, research, analytics and training your own models, in-house or in your own products; the trained models and derived data you build from it are yours. The raw files and tick streams themselves may not be redistributed, resold or published. The full terms are on the data license page.

Building a training set? The training-data page covers dataset sizes, labels from settled event contracts and a Parquet pipeline.