VerticalIntelligence/aihedged-trading-datasets
收藏资源简介:
AIHedged交易数据集——Bybit逐笔数据,包含来自Bybit交易所的ETH、BTC、SOL、DOGE永续合约的逐笔交易数据,覆盖时间为2025年4月20日至2026年4月19日(一年)。数据集总计约17.2亿笔交易,具体包括ETH/USDT(784,357,803笔)、BTC/USDT(594,109,922笔)、SOL/USDT(209,092,277笔)和DOGE/USDT(128,124,272笔)。数据模式包括时间戳(微秒精度,UTC时区)、交易对、买卖方向(taker侧)、价格、数量、交易所标识和交易ID。数据按交易对和月份分区存储为Parquet格式,使用zstd压缩,行组大小为10万行。数据来源为Bybit公共WebSocket交易流,通过regime-trader研究栈捕获,交易ID由交易所分配且每个交易对全局唯一。
AIHedged Trading Datasets — Bybit Tick Data: Tick-level trade data from Bybit for ETH/BTC/SOL/DOGE perpetuals, captured from 2025-04-20 to 2026-04-19 (1 year). The dataset includes approximately 1.72 billion trades in total, with details for ETH/USDT (784,357,803 trades), BTC/USDT (594,109,922 trades), SOL/USDT (209,092,277 trades), and DOGE/USDT (128,124,272 trades). Schema comprises timestamp (microsecond precision, UTC), symbol, side (buy or sell, taker side), price, quantity, exchange identifier, and trade ID. Data is partitioned by symbol and month in Parquet format, compressed with zstd, and has a row group size of 100k rows. Source is captured directly from Bybit public WebSocket trade feed via the regime-trader research stack, with exchange-assigned unique trade IDs per symbol.



