×
📊 Model Validation Results (2024 Season)
Tested on 5 races with full 20-driver grids • 90 total predictions
60%
Winner Accuracy
3/5 races correct
60%
Podium Accuracy
1.8/3 drivers avg
63.3%
Within ±2 Positions
57/90 predictions
77.8%
Within ±3 Positions
70/90 predictions
45.6%
Within ±1 Position
41/90 predictions
16.7%
Exact Positions
15/90 predictions
Race-by-Race Performance
| Race |
Exact |
±1 |
±2 |
Winner |
Podium |
| Bahrain GP |
26.3% |
57.9% |
68.4% |
✅ |
2/3 |
| Monaco GP |
27.8% |
55.6% |
66.7% |
✅ |
2/3 |
| Silverstone GP |
6.2% |
37.5% |
50.0% |
✅ |
2/3 |
| Spa-Francorchamps |
16.7% |
33.3% |
72.2% |
❌ |
2/3 |
| Monza GP |
5.3% |
42.1% |
57.9% |
❌ |
1/3 |
✅ Model Strengths
- Winner prediction: 12x better than random (60% vs 5%)
- Strong overall order: 77.8% within ±3 positions
- Podium prediction: 60% accuracy across all races
- Best on processional circuits (Monaco: 27.8% exact)
- Mean Absolute Deviation: 2.24 positions
⚠️ Known Limitations
- Cannot predict mechanical failures or DNFs
- Weather effects not modeled (rain, temperature)
- Mid-pack volatility (P9-P15 most unpredictable)
- Strategic variations impact accuracy
- Drama races show weaker performance (Monza: 5.3%)
