v0.3.0 - Full Grid Race Prediction (Phase 1)

🏎️ BoxMachiBox v0.3.0 - Full Grid Race Prediction

Major feature release: Predict complete race results for all 20 drivers!

✨ What's New

🆕 Full Grid Prediction Endpoint

• POST /api/predict/full-grid - Predict finishing order P1-P20
• Input: All 20 drivers with starting grid positions
• Output: Complete race prediction with probabilities

📊 Enhanced Predictions

• Predicted finishing order (P1-P20)
• Top 3 podium finishers
• Points-scoring positions (P1-P10)
• Fastest lap prediction
• DNF (Did Not Finish) risk assessment
• Confidence scores for each position
• Expected championship points calculation

🔧 Phase 1 Implementation

Current Approach:

• Uses existing XGBoost model (93.89% accuracy)
• Ranks all 20 drivers by podium probability
• Adjusts for DNF risk based on grid position
• Circuit-specific difficulty factors (Monaco, Singapore, Hungary)
• Weather impact on predictions

Phase 2 (Coming Soon):

• Retrained model for exact position prediction (multi-class)
• Position-specific features
• Enhanced accuracy for mid-field positions

📝 API Example

Request:

{
  "grid": [
    {"position": 1, "driver": "Max Verstappen"},
    {"position": 2, "driver": "Lando Norris"},
    ...
    {"position": 20, "driver": "Franco Colapinto"}
  ],
  "circuit": "Monaco",
  "weather": "Dry"
}

Response:

{
  "predicted_results": [
    {
      "position": 1,
      "driver": "Lando Norris",
      "starting_grid": 2,
      "podium_probability": 0.892,
      "confidence": "High",
      "expected_points": 25
    },
    ...
  ],
  "podium": ["Lando Norris", "Max Verstappen", "Charles Leclerc"],
  "points_positions": [...],
  "fastest_lap_prediction": "Lando Norris",
  "dnf_predictions": ["Franco Colapinto"]
}

🔗 API Endpoints

Production: https://boxmachibox.onrender.com

New Endpoints:

• POST /api/predict/full-grid - Full race prediction

Existing Endpoints:

• POST /api/predict - Single driver podium prediction
• GET /api/drivers - List all drivers
• GET /api/drivers/enhanced - Drivers with team codes
• GET /api/circuits - List all circuits
• GET /api/circuits/enhanced - Circuits with track names
• GET /api/standings/2025 - Championship standings
• GET /api/model/info - Model metadata

📊 Technical Details

• Backend Version: 0.3.0
• Model: XGBoost (Phase 1 - probability-based ranking)
• Accuracy: 93.89% on podium predictions
• Training Data: 2022-2025 (R1-R20)
• Features: 47 engineered features

🎯 Use Cases

• Pre-race predictions - Predict full race results before lights out
• Strategy planning - Understand expected finishing positions
• Fantasy F1 - Make informed driver selections
• Championship scenarios - Calculate points impact
• Race analysis - Compare predicted vs actual results

🚀 Frontend Integration

Frontend developers can now:

1. Build full grid input UI (20 driver dropdowns)
2. Call /api/predict/full-grid endpoint
3. Display complete race predictions
4. Show DNF risks and confidence levels

Documentation: See API docs at /docs endpoint

🔮 What's Next (v0.4.0)

• Weather API integration
• Enhanced circuit-specific features
• Phase 2: Multi-class position prediction model

---

Live API: https://boxmachibox.onrender.com/docs
Frontend: https://boxmachibox.vercel.app/

👥 Team

@sarva-20 - Backend & ML

<img width="1409" height="938" alt="image" src="https://github.com/user-attachments/assets/91981682-8029-4569-8fb5-a2cc2003ca4b" />
<img width="1123" height="941" alt="image" src="https://github.com/user-attachments/assets/c72851ac-5bdc-440b-94e1-5efa7a5dfd1e" />

Enhanced driver and circuit display with team codes

v0.1.0 - Foundation & Documentation

🎉 BoxMachiBox v0.1.0 - Foundation Release

First official release! This version establishes the core infrastructure and documentation for BoxMachiBox.

✨ What's New

• 🏗️ Clean Repository Structure - Separated backend/ML from frontend
• 📄 Professional Documentation - Comprehensive README with setup guides
• 📜 MIT License - Open-source licensing for collaboration
• 🔧 Updated .gitignore - Proper file exclusions and model protection
• 🚀 Live Deployment - API on Render, Frontend on Vercel

🎯 Current Features

• 93.89% Prediction Accuracy - XGBoost model trained on 2022-2025 data
• FastAPI Backend - RESTful API with 6 endpoints
• Next.js Frontend - Modern UI with dark F1 aesthetic
• 47 ML Features - Qualifying intelligence, driver momentum, constructor strength

🔗 Live Links

• Frontend: https://boxmachibox.vercel.app/
• API: https://boxmachibox.onrender.com
• API Docs: https://boxmachibox.onrender.com/docs

📊 Technical Details

• Model: XGBoost Ensemble (multi-seed)
• Training Data: 1,558 samples (2022-2024 + 2025 R1-R20)
• Test Accuracy: 93.89% on 2025 R21-R24
• Backend: FastAPI 0.115.6, Python 3.10+
• Frontend: Next.js 15, TypeScript, Tailwind CSS

🙏 Credits

Backend & ML: @sarva-20
Frontend: @Prabakaran

---

Next Release (v0.2.0): Enhanced driver/circuit display with team codes and track names