Arsenal-Forecast

python, monte-carlo, EPL and UCL datasets

A probabilistic machine learning system that forecasts Arsenals 2025/26 season outcomes across Premier League and UEFA Champions League using Monte Carlo simulation.

view repository

Challenge

Transform colloquial notions of "winning vs bottling" into rigorous, data-driven probabilities while accounting for fixture difficulty, current form, and multi-competition dynamics.

Approach

Built a gradient boosting classifier trained on 2022-2024 EPL data with features including team strength, rolling form, and home advantage. Ran 10,000 Monte Carlo simulations per competition to estimate title probabilities, then combined EPL and UCL forecasts for multi-trophy scenario analysis.

Outcome

Delivered 65% EPL title probability for Arsenal vs 24% for Man City, 32% UCL win probability, and 21% chance of winning the double. The system reveals that Arsenal can afford one slip-up while City needs near-perfection.

Highlights

  • 10,000 Monte Carlo simulations per competition
  • Match-level probability predictions with opponent strength weighting
  • Multi-trophy scenario analysis (double, single trophy, neither)
  • 54% model accuracy on historical data validation
NextCrowd-Density