LuciferForge/cross-signal-data

这是一个标注的Polymarket崩溃恢复数据集，包含308个已关闭交易记录，用于研究预测市场中的均值回归策略。每个交易记录包含市场ID和问题文本、信号特征(崩溃前高点、入场价格、下跌百分比)、结果特征(退出价格、退出原因、美元盈亏、是否盈利)以及时间特征(入场UTC小时、入场星期几、持有小时数)。数据集记录了实际交易结果，盈利交易占比80.2%，时间跨度为2026年3月至4月。该数据集可用于构建均值回归交易机器人、复制研究结果或验证策略有效性。

A labeled Polymarket crash-recovery dataset containing 308 closed trades for studying mean-reversion strategies in prediction markets. Each trade includes market ID and question text, signal features (pre-crash high, entry price, drop percentage), outcome features (exit price, exit reason, PnL in USD, profitability), and temporal features (entry UTC hour, day of week, holding hours). The dataset records actual trading outcomes with 80.2% profitable trades, spanning March 2026 to April 2026. It can be used to build mean-reversion bots, replicate results, or validate strategies.