Bayesian-Poisson statistical reports for Polymarket prediction markets. Data-driven edges, delivered before market resolution.
The closer to resolution, the more data — and the sharper the prediction.
VIEW TRACK RECORD ↓Reports will be published as settlement approaches.
Reports will be published as settlement approaches.
Reports will be published as settlement approaches.
Reports will be published as settlement approaches.
Reports closer to settlement incorporate more observed data. The closer, the sharper.
Our model uses a three-component weighted Bayesian framework (λ_prior + λ_obs + λ_recent) anchored to xtracker timestamp data. λ_recent carries 50% weight — the single most predictive variable for short-horizon forecasting.
Why closer = better: At T-12h, the model relies primarily on the prior. By T-1h, λ_recent dominates. But the best EV window is often T-5h to T-3h, where model confidence is already high but the market price hasn't caught up yet.
Each report includes the full posterior predictive distribution, a bi-directional EV matrix, position sizing recommendations, and risk monitoring thresholds.