Guides / Education

Data Cleaning for Prediction Market Backtests

Before backtesting, check for: gaps in the snapshot timeline, duplicate timestamps, and markets that resolved early or were cancelled. ProbSights' orderbook and trade endpoints are timestamp-ordered per market, which makes gap detection straightforward — look for jumps larger than your expected snapshot interval.

Related

• Common backtesting mistakes
• Working with resolved prediction market data

Related guides

• Prediction Market Data for Academic Research
• Building a Polymarket Backtesting Framework in Python
• Common Backtesting Mistakes in Prediction Markets
• Execution Simulation With Prediction Market Data

Data from Kalshi and Polymarket. Search live · API docs · Pricing