Transparency
Model Accuracy
Every prediction is checked against the real result. We score each fight using only what the model knew beforehand — a fair, out-of-sample test across 1,485 fights.
58%
Accuracy
picks the winner
0.241
Brier score
lower is better
42.8%
Confident calls
committed beyond 60/40
Calibration
A trustworthy model wins about as often as it says it will. When we call a fighter a 70% favourite, they should win ~70% of the time. Bars show the actual win rate in each confidence band:
50–60%
said 55% · won 53%
n=849
60–70%
said 64% · won 62%
n=557
70–80%
said 72% · won 72%
n=79