This GitHub repository lets you backtest trading strategies on real Polymarket and Kalshi data
An event-driven engine that replays historical trades in chronological order - simulating order fills, portfolio tracking, and market lifecycle events
Built on top of Jon-Becker's dataset with 36GB of real trade history
What's inside:
> Three ready-to-use strategies: buying at low prices, calibration arbitrage, and martingale with mean-reversion
> Detailed charts: equity curve, P&L, drawdown, Sharpe, monthly returns
> Polymarket and Kalshi support out of the box
> Simple API for writing your own strategies - drop a file in the folder and it appears in the menu automatically
> Hooks for every event: market open, close, resolution, order fill
Most people test strategies on paper or go straight to live trading
This engine lets you run a strategy through millions of real trades before spending a single dollar
You see the drawdown, Sharpe, monthly returns - all on real data, not synthetics
Repo is in active development, full release planned in 1-2 months. Good time to get in early and star it while there's still barely anyone there
GitHub:
github.com/evan-kolberg/pred…