Introduction
What is BetIntel?
BetIntel is a complete machine learning system for football match predictions. The system uses an ensemble of advanced models to generate probabilistic predictions for various betting markets.
Core Principles
1. Data-Driven Approach
All predictions are based on historical data analysis, not intuition or "expert" opinions a priori.
2. Expected Value
We don't just look for "likely" predictions, but for value bets - situations where odds offered by bookmakers exceed the real probability.
3. Bankroll Management
The system implements Kelly Criterion for optimal capital allocation, minimizing ruin risk while maximizing growth.
Main Features
ML Predictions
- 1X2: Home/Draw/Away result
- Over/Under 2.5: Goals in match
- BTTS: Both teams to score
Value Bet Detection
Value Bet = When P(model) > 1/Odds
Example:
- Model: 45% home win probability
- Odds: 2.50 (implied 40%)
- EV = 0.45 × 2.50 - 1 = +12.5%Covered Markets
| Market | Model | Accuracy |
|---|---|---|
| 1X2 | PRODUCTION_V3_GBM | 62-65% |
| Over/Under 2.5 | Dedicated | 64% |
| BTTS | Dedicated | 52% |
Tech Stack
| Component | Technology |
|---|---|
| Backend | Python, FastAPI |
| Frontend | Next.js 16, TypeScript |
| Database | PostgreSQL |
| Cache | Redis |
| ML | LightGBM, CatBoost, scikit-learn |
| Deploy | Docker, Docker Compose |
Next Steps
- Installation - Environment setup
- Configuration - API keys and preferences
- ML Models - Understanding models