Skip to main content

Prediction Analytics

How the model spreads its calls, and how confidently it makes them.

217,281 Predictions
Model in production
64.1%
High Confidence (70%+)
47.3%
Main Outcome Accuracy
43.9%
Always backing the home side

+3.4 edge over the baseline · 153,923 graded predictions

What the model predicts, against what happens

Calibration

Left: how the model spreads its calls. Right: how the matches that have since been played actually finished. The closer the two, the better calibrated the model is. A well-calibrated model spreads its calls close to how matches actually finish.

Model predicts

Home Wins 54% 117,433
Draws 17.8% 38,626
Away Wins 28.2% 61,222

Actually happened

Home Wins 43.9% 67,647
Draws 23.5% 36,155
Away Wins 32.6% 50,121

Confidence spread

How many calls the model makes at each confidence level. Whether those levels are actually right is on the track record.

High Confidence > 80% 1.9% 4,077
Medium Confidence 50–80% 98.1% 212,762
Low Confidence < 50% 0% 0

Risk profile

Low risk 3.5% 5,145
Medium risk 96.5% 140,997

Every figure on this page describes the model running in production today. Predictions from retired models are excluded.

This page describes how the model behaves, not how often it is right. For that, see the verified track record.

View our track record

Quick Stats

Total Predictions 3,190
High Confidence 111
Today 122

Disclaimer

Images and logos are used under fair use for editorial purposes only.

We are not affiliated with any leagues, clubs, or players featured on this website.