Methodology

How It Works

PredZone uses gradient-boosted decision trees (XGBoost) trained on historical match data to predict three markets: match result (1X2), total goals (over/under 2.5), and total cards (over/under 3.5).

Six Prediction Factors

1. Team Strength — historical league position and points performance
2. League Strength — UEFA coefficients for cross-league comparison
3. Coach Stability — manager tenure and recent changes
4. Ex-Club Factor — players facing former teams (motivation signal)
5. Referee Tendencies — average cards and fouls per official
6. Recent Form — results, goals, and discipline over the last 5 matches

Data Source

All data is sourced from API-Football, covering 1500+ competitions worldwide. Models are retrained weekly and predictions are updated daily.

Transparency

We believe in open methodology. Every prediction shows the factors that influenced it, so you can understand why the model made its forecast. No black boxes.