Probalytics

Traders, researchers, and quants who want to analyze or backtest strategies on prediction markets like Polymarket and Kalshi need full-depth historical order book and trade data, but reconstructing this from raw exchange feeds requires capturing and storing every book change at millisecond granularity over long time spans. Doing this independently means building and maintaining ingestion pipelines across multiple platforms, storing massive volumes of snapshot data, and normalizing schemas so Polymarket and Kalshi data can be queried consistently. Without this, anyone wanting to study market microstructure, backtest a trading strategy, or analyze buy/sell pressure on these markets has to assemble the data themselves from scratch.

Probalytics stores full-depth order book snapshots on every change for Polymarket (since November 2025) and Kalshi (since May 2026), timestamped to the millisecond, with market data going back to October 2020 and fills back to 2022. Users can query the data directly in ClickHouse, pull it over a REST API, or export it as Parquet files. The product exposes typed tables for order book snapshots, fills, and markets with an identical schema across both platforms, including fields like bids/asks arrays, outcome details, hashes, and continuity state. The site provides sample queries (e.g., reconstructing top-of-book from the full ladder) and lets prospective users download a sample dataset and view the full schema before writing code.