Soccer Data Analytics Hackathon - Getting Started

Event Dates: February 27–28, 2026
Location: Northeastern University / Network Science Institute
Supported by: PySport

Overview

This repository provides starter code and instructions for the Soccer Data Analytics Hackathon. You'll work with IMPECT Open Data containing 306 German Bundesliga matches from the 2023/24 season to tackle one of two challenge prompts.

Challenge Prompts (Choose One)

Option A: Starting Eleven Lineup Construction

Recommend an optimal starting eleven and/or substitution plan to maximize team cohesion and ball progression. Build a player-to-player pass network, analyze network structure, and compare alternative lineups with clear visualizations.

Option B: Transparent Player Valuation Metric

Define an interpretable attacking or defensive metric using event data. Create a metric definition, produce a leaderboard comparing players, present a case study, and discuss limitations.

Quick Start

1. Installation (within Jupyter notebook)

!pip install "kloppy>=3.18.0" polars pyarrow

2. Explore the Notebook

Open getting-started.ipynb to see examples of:

• Loading matches and squad data
• Filtering for specific event types (passes, shots)
• Transforming coordinate systems
• Exporting to Polars/Pandas DataFrames

Project Structure

SoccerImpectHackathon/
├── getting-started.ipynb        # Tutorial notebook
├── environment.yml              # Conda environment
├── requirements.txt             # Python dependencies
├── .gitignore                   # Git ignore rules
├── LICENSE                      # MIT License
├── README.md                    # This file
└── CONTRIBUTING.md              # Contribution guidelines

Deliverables (Due Friday, February 27, 2026)

1. Slide deck (PDF, max 8 slides) with clear visualizations
2. GitHub repository with:
   • Clean, reproducible code
   • README.md explaining your approach
   • environment.yml or requirements.txt
   • Open-source license (MIT or Apache 2.0)

Timeline

Milestone	Date
Release & Data Primer	Monday, November 3, 2025
Registration Deadline	Wednesday, December 31, 2025
Checkpoint (draft slides/repo)	Monday, February 2, 2026
Final Work Session & Judging	Friday, February 27, 2026
Industry Talks & Awards	Saturday, February 28, 2026

Suggested Tools

• Python: kloppy, polars, pandas, numpy, mplsoccer, databallpy, networkx
• R: tidyverse, igraph
• Any language is acceptable as long as your work is reproducible

Resources

• Kloppy Documentation
• IMPECT Open Data
• PySport

License & Ethics

• IMPECT Open Data is for non-commercial use only
• Cite all sources appropriately
• If using AI tools, document where and how they were used
  • For example: This .README was generated with the help of Claude Sonnet 4.5
• Be transparent about limitations in your methodology

Judging Criteria (100 points)

• Problem framing & soccer context (10 pts)
• Data engineering & correctness (15 pts)
• Methodology quality (15 pts)
• Validation & robustness (15 pts)
• Results & insight (15 pts)
• Communication & visualization (15 pts)
• Reproducibility & ethics (15 pts)

Contact

Questions? Email northeasternsportsanalytics@gmail.com

---

Good luck and happy hacking! ⚽📊