Polymarket Signal Hunter

building

Finding edges in prediction markets through order flow, factor analysis, and a lot of backtesting.

What

A system that watches Polymarket prediction markets and looks for trading signals. It collects order flow, tracks whale activity, runs microstructure analysis, and scores markets by how likely they are to be mispriced.

Signal detection pulls from a few different angles. Microstructure indicators measure informed trading and order flow toxicity. Factor models adapted from equities get combined with prediction market-specific ones like time decay and boundary effects. Whale tracking and sentiment analysis add another layer. Everything gets backtested with walk-forward analysis so the numbers hold up out of sample.

Why

Prediction markets look like financial markets on the surface, but the underlying assets resolve to 0 or 1, which changes the math in interesting ways. Resolution dates, binary outcomes, and unusual liquidity patterns all need their own treatment. Most trading tools are built for equities or crypto. Wanted something purpose-built for this.

Where it stands

Core infrastructure is running. Data collection across multiple Polymarket APIs, signal generation with several active strategies, a real-time dashboard, and a backtest engine with 52 parameter combinations. Next up is cross-platform arbitrage and better factor ensembles.