Football Intelligence Hub

Tactical football analysis platform powered by data and LLM narrative. Given a team, competition and time range, the app generates strengths & weaknesses reports, positional heatmaps, pressing maps, passing networks, and player-role fit analysis — with an LLM coach that turns numbers into concrete decisions.

Features

• Team analysis: S&W report, heatmaps, pressing intensity, passing structure
• Player-role fit: radar vs. ideal profile per position
• Role comparisons: overlay players across teams for the same position
• Match report: event-level viz when StatsBomb data is available (Voronoi, set pieces, transitions)
• Exportable reports: Italia post-mortem, scouting reports, match reports

Data sources

Source	Coverage	Used for
FBref	All major leagues, current seasons	Primary aggregate stats
Understat	Top 5 leagues, current seasons	xG and shot maps
StatsBomb Open Data	Historical competitions (Euro 2020, etc.)	Event-level viz

Stack

Python · Streamlit · Pandas · Plotly · mplsoccer · soccerdata · Anthropic API · Ollama

Setup

git clone https://github.com/leoturambar/football-intelligence-hub.git
cd football-intelligence-hub

conda create -n football python=3.11
conda activate football
pip install -r requirements.txt

streamlit run app.py

Anthropic API key — set once as a system environment variable on Windows:

setx ANTHROPIC_API_KEY "sk-ant-..."

Reopen the terminal after running setx.

Ollama — install from https://ollama.com, then:

ollama pull qwen2.5:14b

Launch from desktop (Windows)

Double-click launch.bat.

Project structure

football-intelligence-hub/
├── app.py
├── config.py
├── data_loader.py
├── metrics.py
├── viz.py
├── analysis.py
├── llm.py
├── data/
├── CLAUDE.md
├── README.md
├── requirements.txt
├── launch.bat
└── LICENSE