The research record, at desk scale.
EdgeDepth records the order book, best bid and offer, trades, liquidations, open interest and funding for Binance USDT-M, Hyperliquid and Bybit perpetuals: our own full-depth capture, not a reduced vendor snapshot. On top of it sits the deterministic engine behind EdgeDepth research, sized for a team or a fleet of agents through the REST API and MCP. Current research access spans 670 Binance USDT-M crypto perpetuals; Hyperliquid historical research and cross-exchange research are not yet publicly available.
The record is searchable from 15 July 2025; order-book replay covers windows from 15 May 2026; earlier days are searchable but hold trades and candles only, so they cannot be replayed on any plan. What is recorded, stream by stream → · Depth start dates per venue →
High-resolution source data and a research layer you can reproduce.
The source capture matters. EdgeDepth records native-rate per-symbol Binance bookTicker plus sub-second per-coin Hyperliquid BBO: every received best-bid and best-ask price and quantity change, preserved per symbol. It pairs that symbol-specific top of book with trades, depth, liquidations and market state instead of reducing BBO to an occasional exchange-wide sample.
The research layer turns that capture into something a desk can query and verify: a versioned feature store, a deterministic grammar, and results anyone can rerun byte for byte and cite. High-volume REST and MCP access let a team or agent fleet use both layers without rebuilding the data path.
Research conditions computed from the recorded best bid and offer and order-book depth are in development. They stay out of public results until their derived data and coverage pass validation; the reading library lists what is searchable today.
- +High-volume research API and MCP: the full grammar (scans, snapshots, cohorts, commonality), metered for a team or a fleet of agents
- +The computed feature store, versioned: 87 features across the current Binance research record, not raw ticks you re-derive
- +Native top-of-book capture: native-rate per-symbol Binance bookTicker plus sub-second per-coin Hyperliquid BBO, with every received best-bid and best-ask price and quantity change, preserved per symbol
- +Reproducible research at scale: pinned reports, reproducibility keys, CSV, byte-identical reruns your whole desk can cite
- +Standing definitions and alerts at volume, with priority scan execution
- +Deeper history and custom coverage, scoped to your desk
Tell us your pairs, volume, cadence, and whether you need the API, MCP, reports or raw capture. Typical first response within two business days.
Or try the research engine free first →Full-depth history, when you need the ticks.
The underlying capture includes trades, full-depth L2, high-resolution per-symbol BBO, liquidations, open interest and funding across 670 Binance USDT-M crypto perpetuals, plus eligible Hyperliquid crypto perpetuals. Binance data includes native-rate dedicated @bookTicker streams; Hyperliquid includes sub-second per-coin BBO and full 20-level books. We scope deliveries per desk by venue, symbols, window, datasets and cadence.
The public searchable feature store is currently Binance-scoped. Hyperliquid historical data has been backfilled and recorded Hyperliquid replay covers about six weeks so far. Hyperliquid historical research is not yet publicly available; derived data and coverage must be validated before release. Cross-exchange research is in the works, and results will always say which venue they came from before any Binance-to-Hyperliquid comparison is allowed.
Big moves, free to download.
For selected major events we publish the raw capture behind the replay (trades, full-depth book deltas, liquidations) as tick-timestamped parquet, free. Download it, backtest it, teach with it, verify it against the terminal tick for tick.
Each bundle links its archived event, so you can watch the same ticks you're holding. New drops are announced on X. For research or educational use.