TACTIX AI — Football Transition Intelligence

Football analytics dashboard exploring possession-loss indicators and subsequent opponent attacks using StatsBomb World Cup 2022 data, with interactive visualizations and statistical analysis.

Comment

Inspiration

In football, losing possession can create opportunities for the opposing team to attack. However, identifying these situations and understanding what happens afterward requires analyzing large amounts of match-event data.

I built TACTIX AI to help football analysts explore potential possession-loss events, investigate subsequent opponent attacks, and identify patterns that deserve further tactical analysis.

What it does

TACTIX AI is an interactive football analytics dashboard built using StatsBomb Open Data from the 2022 FIFA World Cup.

The application processes match-event data and identifies candidate possession-loss indicators, including dispossessions, miscontrols, unsuccessful passes, and unsuccessful dribbles. It then examines whether the opponent's next possession contains a shot.

The dashboard provides seven sections:

  • Overview of key metrics
  • Visualization of candidate possession-loss locations
  • Transition-outcome analysis
  • Match-specific analysis
  • Data-quality validation
  • Team comparison
  • Advanced statistical analytics

Users can explore results through interactive charts, filters, tables, and CSV exports.

How I built it

I developed the application using Python, Streamlit, Pandas, NumPy, Plotly, and StatsBomb Open Data.

The data pipeline retrieves and processes match events, identifies candidate indicators, examines subsequent opponent possessions, and generates reports for the dashboard.

I also implemented Wilson 95% confidence intervals and match-clustered bootstrap analysis to explore statistical uncertainty. Automated tests help validate the statistical analysis code.

Challenges I faced

One challenge was distinguishing candidate possession-loss events from confirmed turnovers. An event that appears to indicate a loss does not necessarily establish that a genuine turnover occurred.

Another challenge was interpreting subsequent attacking outcomes without assuming causation. A shot during the opponent's next possession does not prove that the candidate event caused the shot.

I addressed these issues by documenting the methodology, adding data-quality checks, incorporating statistical uncertainty estimates, and clearly communicating the limitations of the analysis.

Team-level statistical estimates require particular caution because the dataset contains only three to seven matches per team.

Accomplishments

I built and deployed a working football analytics dashboard with seven interactive sections. The project includes data processing, validation, visual analytics, team comparisons, statistical reliability analysis, and CSV exports.

The dashboard was manually checked across all seven sections, and the recorded statistical test suite passed 14 tests.

Actionable impact

TACTIX AI can help football analysts explore potential transition vulnerabilities, compare observed patterns, and prioritize matches or situations for deeper tactical investigation.

Rather than presenting its outputs as definitive judgments, the application provides an exploratory starting point for analysts to investigate candidate events and evaluate their tactical context.

Limitations and future work

The current analysis uses FIFA World Cup 2022 event data. Candidate events are not automatically confirmed turnovers, and observed associations do not establish causation.

Future improvements could include richer match context, additional validation, sensitivity analysis, and further investigation of transition patterns across larger datasets.

Links

Live Dashboard : https://tactix-ai-zqc29qk7pigitffodmsjfj.streamlit.app/

Source code: https://github.com/mrnikhilsinha07/TACTIX-AI

Built With

Share this project:

Updates

Submission history