Case study · 11
Prediction-market bot
A Python bot that turns outside signals into a probability, then compares it with the market price.
- Role
- Personal project
- Status
- Side project
- Read
- 1 min
- Stack
- PythonAPSchedulerSQLiteFlaskKalshi API
½ Kelly
position sizing, behind a kill switch
Problem
Prediction-market prices lag the information around them, but only a disciplined system can act on that safely.
Approach
- 01Matches contracts to outside signals: Polymarket, macro data, rate-expectation data and news-headline sentiment.
- 02Edge is model probability minus market price, sized with half-Kelly.
- 03A risk manager enforces limits, tracks daily P&L and holds a kill switch. API calls are signed, and a paper-trading mode runs first.
- 04Every decision is written to a SQLite trade log, with a Flask dashboard on top.
Only a disciplined system can act on market lag safely.
Results
- A complete scan, size and log loop with paper trading. No performance claims.
What I'd do next
- Backtest the edge model before any live capital.