AI Probability Estimates

Predictions✧
AI predictions
AI Models✧
Ensemble
Independent multi-model
Reconciliation✧
Two-pass
α=0.7 consistency check
Calibration✧
Platt
Logistic calibration
AI vs Market Estimates
Monitoring
AI Architecture
Multi-model ensemble AI with independent reconciliation
AI Probability EngineIdle
Multiple independent sources · multi-pass reconciliation
Platt calibration active.
Model Agreement
Model agreement score — stdev across ensemble model estimates. Lower divergence = higher agreement.
Model Routing
Per-category model weights and empirical Brier scores — methodology transparency only.