| Dataset | Files | Size | Coverage | Format | Granularity |
|---|---|---|---|---|---|
| Binance Spot Trades | 828 | 888.1 GB | Jan 2020 – May 2025 | CSV | Monthly + daily, 18 symbols |
| Bybit Perp Trades | 13,477 | 209.8 GB | Mar 2020 – May 2025 | GZ CSV | Daily, 11 symbols |
| HL L2 Book | 237,656 | 165.7 GB | Apr 2023 – May 2026 | LZ4 NDJSON | Hourly per coin, 10 coins |
| HL Fills by Block | 7,022 | 164.4 GB | Jul 2025 – May 2026 | LZ4 NDJSON | Hourly, batched by L1 block |
| HL Fills (node_fills) | 1,507 | 29.6 GB | May – Jul 2025 | LZ4 NDJSON | Hourly |
| HL Trades (node_trades) | 1,542 | 14.1 GB | Mar – Jun 2025 | LZ4 NDJSON | Hourly |
| HL Asset Contexts | 1,077 | 8.1 GB | May 2023 – May 2026 | LZ4 CSV | Daily |
| Coherence Research | 213 | 794 MB | — | Various | Anomaly detection datasets |
Per-wallet fill data from HL L1. Each record is one fill event attributed to a specific wallet address, with realized PnL, position tracking, and maker/taker classification.
Unique value: wallet addresses, per-fill PnL, maker/taker flag[wallet, {coin, px, sz, side, time, closedPnl, fee, dir, crossed, startPosition, hash, tid, oid, ...}]
Matched trades from HL L1. Unlike fills, each record shows both sides of the trade — the maker and taker are both visible in the side_info array.
Unique value: both parties visible in side_info array{coin, side, time, px, sz, hash, side_info: [{user, startPosition, dir, ...}, ...]}
Successor to node_fills. Fills grouped by L1 block number with block timestamps. Continuous live stream that is still actively recording.
Unique value: fills grouped by block, continuous live stream{local_time, block_time, block_number, events: [...fills...]}
Daily snapshots of all HL perp market context: funding rates, open interest, oracle prices, mark prices, and volume. Covers entire HL history.
Unique value: funding rates, open interest, oracle pricestime, coin, funding, open_interest, prev_day_px, day_ntl_vlm, premium, oracle_px, mark_px, mid_px
Full order book snapshots (bid/ask depth) for 10 coins. Hourly files per coin, enabling order flow analysis, liquidity measurement, and spread decomposition.
Coins: BTC, ETH, SOL, HYPE, XRP, DOGE, AVAX, SUI, AAVE, WIF{time, levels: [[{px, sz, n}, ...], [{px, sz, n}, ...]]} // [bids, asks] full depth
Every spot trade on Binance for 18 USDT pairs. Monthly bulk archives plus daily files for recent data. The largest dataset by far.
Symbols: BTCUSDT, ETHUSDT, SOLUSDT, XRPUSDT, DOGEUSDT, BNBUSDT, AVAXUSDT, SUIUSDT, AAVEUSDT, WLDUSDT, WIFUSDT, ENAUSDT, TRUMPUSDT, VIRTUALUSDT, POPCATUSDT, MOODENGUSDT, BERAUSDT, PENGUUSDTtrade_id, price, qty, quote_qty, time, is_buyer_maker, is_best_match
Daily perpetual futures trade data from Bybit. Compressed CSV files per symbol per day. Provides a third venue for cross-exchange comparison alongside Binance and HL.
Symbols: BTCUSDT, ETHUSDT, SOLUSDT, XRPUSDT, DOGEUSDT, BNBUSDT, AVAXUSDT, SUIUSDT, AAVEUSDT, WIFUSDT, HYPEUSDTtimestamp, symbol, side, size, price, tickDirection, trdMatchID, grossValue, homeNotional, foreignNotional
Benchmark datasets for anomaly detection model development and validation.
Contains: credit_card, NAB, NASA turbofan, PhysioNet ECG, SKAB, UCI powerSupplementary market data for basis/funding analysis. Currently being downloaded and will be uploaded to B2 on completion.
1m klines (spot + UM perps): BTC, ETH, ETHBTCMark price klines (UM): BTC, ETHPremium index klines (UM): BTC, ETHFunding rates (monthly): BTC, ETHL2 bookDepth (UM): BTC, ETH (from 2023)
| Concept | HL Fills | HL Trades | Binance Spot | Bybit Perps |
|---|---|---|---|---|
| Price | px |
px |
price |
price |
| Size | sz |
sz |
qty |
size |
| Timestamp | time (ms) |
time (ms) |
time (us) |
timestamp (ms) |
| Side / Aggressor | side (B/A) |
side |
is_buyer_maker |
side (Buy/Sell) |
| Symbol | coin |
coin |
Symbol (e.g. BTCUSDT) | symbol |
| Notional | px * sz |
px * sz |
quote_qty |
foreignNotional |
| Trade ID | tid |
hash |
trade_id |
trdMatchID |
| Wallet / User | wallet |
side_info[].user |
— | — |
| PnL | closedPnl |
— | — | — |
| Maker/Taker | crossed |
— | is_buyer_maker |
tickDirection |