Independent research. An accountable record.

What we publish

Reviewed probability estimates, sourced fighter information and matchup analysis. The original editorial collection remains available with its limitations clearly marked.

Evidence before narrative

Unknown inputs remain unknown. Model interpretation is separate from verified statistics, and a human reviewer checks analysis before publication.

Every outcome stays visible

Published prediction versions are retained. Results and corrections feed public accuracy, Brier score and calibration; losing forecasts are not removed.