Performance Dashboard

Track model accuracy, analyze prediction performance, and monitor tier effectiveness across all positions and time periods.

Overall Accuracy
93%
↑ 2.4% vs last month
Avg MAE
4.99
↓ 0.3 vs baseline
Predictions Made
31,247
2019-2024 seasons
Tier Accuracy
91%
GMM clustering

Prediction Accuracy by Position

Percentage of predictions within 3 points of actual score

Detailed Metrics

QB544 predictions
MAE
6.17
±3 Points
89%
Tier Acc.
89%
RB1,088 predictions
MAE
4.92
±3 Points
91%
Tier Acc.
91%
WR1,632 predictions
MAE
4.99
±3 Points
90%
Tier Acc.
90%
TE544 predictions
MAE
3.88
±3 Points
94%
Tier Acc.
94%

Model Performance Insights

Strong TE Predictions
Tight ends show highest accuracy (94% within 3 points) due to more predictable usage patterns.
Tier System Validation
91% tier accuracy confirms GMM clustering effectively groups players by similar production levels.
Consistent RB Performance
Running back predictions maintain strong accuracy with lowest MAE variance week-to-week.
QB Volatility
Quarterbacks show higher MAE due to rushing upside variability and game script dependence.
Weather Impact
Outdoor games in poor weather conditions reduce prediction accuracy by ~7% for skill positions.
Continuous Improvement
Model accuracy improves throughout the season as more game data becomes available.

Data & Methodology

Training Data
  • • 31,247 player-game records
  • • 2019-2024 NFL seasons
  • • Pre-game features only
  • • No data leakage safeguards
Neural Network
  • • Ensemble architecture
  • • Position-specific models
  • • Feature engineering
  • • Cross-validation
GMM Clustering
  • • 16-component mixture model
  • • Tier confidence scores
  • • Value gap identification
  • • Draft strategy optimization