Inspiration
The FIFA World Cup 2026 marks a historic expansion from 32 to 48 teams, creating an unprecedented opportunity to predict outcomes using data analytics. With the tournament being hosted across North America (USA, Canada, Mexico) for the first time, we wanted to leverage Hex's powerful analytics platform to build comprehensive predictions.
What it does
This project analyzes 50+ years of international football data to:
- Predict the FIFA World Cup 2026 winner using historical trends and current form
- Visualize team performance patterns across World Cup history
- Factor in the unique 48-team format and North American hosting advantage
- Provide data-driven insights on favorites like Spain, France, Brazil, Argentina, and Germany
How we built it
Using Hex's integrated environment:
- Connected to Kaggle's international football results dataset (964 World Cup matches)
- Leveraged Hex's Notebook Agent for automated EDA and visualization
- Built predictive models considering historical win rates, FIFA rankings, and continental performance
- Created corrected predictions factoring in current squad quality and 2026 tournament structure
- Published interactive analysis for public exploration
Challenges we ran into
Historical bias was a major challenge - models heavily weighted past performance (1930-2022), which overrated traditional powerhouses like Germany and Italy while underrating current favorites like Spain (#1 FIFA ranking). We addressed this by adding corrected predictions that balance historical trends with 2026 realities.
What we learned
Data storytelling requires balancing historical patterns with current context. The 48-team format and North American venue create new dynamics that pure historical models can't capture.
What's next
Build a hybrid ensemble model incorporating:
- Squad market value (30%)
- Current ELO ratings (25%)
- North American travel/climate factors (20%)
- Expected goals (xG) metrics (15%)
- Tournament path difficulty (10%)
Log in or sign up for Devpost to join the conversation.