A tactical intelligence and squad analysis platform for semi-professional football teams, combining match data, machine learning, player profiling, and rule-based tactical reasoning to generate actionable coaching insights.
gafferOSv2 is designed to help football coaches and analysts make smarter tactical and squad decisions without requiring elite-level infrastructure.
The system combines:
- Match analytics
- Machine learning predictions
- Player form tracking
- Physical assessment data
- Tactical trait profiling
- Explainable tactical recommendations
to simulate a lightweight “football operations system” for clubs.
Instead of asking:
“What is the best formation?”
gafferOS asks:
“What tactical adjustment best fits this squad, opponent, and match context?”
The engine evaluates:
- Team strengths and weaknesses
- Opposition style
- Squad form and fatigue
- Tactical player traits
- Physical capabilities
- Match context (home/away, pressing intensity, etc.)
to recommend tactical setups and player selections.
- Match outcome prediction using engineered StatsBomb features
- 30+ tactical and statistical metrics
- Rolling form analysis
- Team vs opponent differential modeling
- Home/away contextual adjustments
- Coach-assigned tactical trait system
- Position-specific trait banks
- Tactical tendency scoring
- Conflict validation for incompatible traits
Converts real-world testing data into Football Manager-style attributes:
- Pace
- Acceleration
- Stamina
- Strength
- Heading
- Jumping
based on:
- Sprint times
- Beep tests
- Vertical jump
- Height/weight
Hybrid ML + rule-based tactical decision system:
- Formation recommendations
- Press intensity logic
- Tactical focus selection
- Squad rotation considerations
- Trait-aware tactical reasoning
The system explains why recommendations are made instead of producing black-box predictions.
Example:
“A 4-3-3 is recommended due to strong progressive passing profiles and high squad pace.”
- PostgreSQL schema design
- FastAPI backend architecture
- Match ingestion pipeline
- Feature engineering pipeline
- XGBoost outcome model
- Tactical engine v1
- Player trait system
- Physical-to-attribute calculator
- Squad form tracking
- REST API endpoints
- Tactical explainer upgrade
- Rotation advisor
- Deployment
- React frontend
- Matchup-specific tactical adaptation
- Multi-club support
- Season management
- Advanced visual dashboards
- Python
- FastAPI
- PostgreSQL
- SQLAlchemy
- XGBoost
- Pandas
- NumPy
- Scikit-learn
- StatsBomb Open Data
Match Data
↓
Feature Engineering
↓
ML Inference + Tactical Metrics
↓
Trait & Attribute Analysis
↓
Tactical Reasoning Engine
↓
Explainable Recommendations
backend/
├── api/ # REST API routes
├── core/ # tactical and business logic
├── db/ # database models and setup
├── ml/ # ML training + feature engineering
└── main.py
To provide semi-professional clubs with accessible tactical intelligence tools that are normally only available to elite football organizations.
Active development — transitioning from a prediction-focused system into a full tactical decision-support platform.