manja316/polymarket-historical-prices

Polymarket历史价格数据集包含9,550个预测市场的950万多个价格快照，时间分辨率为15分钟，覆盖30天的连续数据。数据集包含三个主要表格：markets（市场元数据，如问题、类别、交易量、流动性和解决日期）、prices（YES/NO结果的15分钟价格快照）和orderbooks（买卖深度快照）。数据集还揭示了关键发现：在价格暴跌超过20%后，市场在15分钟内恢复6.6%的概率为78%，这一均值回归模式在280次实时交易中实现了80%的胜率。数据通过Polymarket Gamma API和CLOB API的自动化管道24/7收集，每15分钟刷新一次。

The Polymarket Historical Prices dataset contains 9.5M+ price snapshots across 9,550 prediction markets with 15-minute resolution, covering 30 days of continuous data. The dataset includes three main tables: markets (market metadata such as question, category, volume, liquidity, and resolution date), prices (15-min price snapshots for YES/NO outcomes), and orderbooks (bid/ask depth snapshots). A key finding from the dataset is that after a >20% price crash, markets recover +6.6% within 15 minutes, 78% of the time, with this mean-reversion pattern producing an 80% win rate across 280 live trades. Data is collected via an automated pipeline running 24/7 on Polymarket Gamma API + CLOB API, refreshing every 15 minutes.