Project / 04 / Mayur Patil
World Cup Forecasting & Simulation
Model international football results with Dixon-Coles, simulate tournament progression with Monte Carlo, and explore Bayesian uncertainty. A Streamlit dashboard presents forecasts, backtests, and the final tournament story.
The problem
Express uncertainty in international football results and simulate tournament progression.
How it works
Use Elo as a reference, Dixon-Coles match probabilities, Monte Carlo tournament simulation, and a separate Bayesian uncertainty experiment. Backtests use training cutoffs before the 2018 and 2022 tournaments.
Evidence and evaluation
Repository-reported Brier scores: 0.566 for the 2018 World Cup and 0.614 for 2022, with 64 matches per tournament. Lower is better. These saved reports have not been independently rerun for this portfolio.
Limitations
The June 12, 2026 forecast came after opening day on June 11. It is an opening-tournament forecast. Models omit injuries and lineups and simplify shootouts. The Bayesian comparison uses a different training window.
About the author
Mayur Patil is a data scientist and ML engineer pursuing an M.S. in Data Science at The George Washington University, based in Falls Church, Virginia. His work spans data engineering, applied ML, language models, and sports forecasting.
More about Mayur Patil ↗