• Got really into Premier League soccer recently
• Curious about which teams actually outperform expectations over time
• Used historical data from to 1992 - 2023
• Tried to build models that could predict match outcomes
• Used Kaggle for my data

epl_data <- read.csv("prepared_epl_data.csv")

• Structure: 12026 matches across 31 seasons
• Key variables:
  • Teams (Home/Away)
  • Goals scored
  • Match date and season
  • Final result (win/draw/loss)

• Home teams win: 45.9% of matches
• Draws: 25.8% of matches
• Away teams win: 28.4% of matches

• Random Forest: 73% accuracy vs. baseline of 46%

• Home advantage has declined significantly over three decades

  • From ~60% in the 1990s to ~45% in recent seasons
  • Suggests tactical evolution and professionalization of away approaches

• Modern Premier League is slightly more goal-rich than early seasons

• Elite clubs maintain their dominance despite more competitive league

  • Top teams win ~65% of home games (vs. league average of ~46%)

• Random Forest achieved 73% accuracy

  • Goal difference and team identity are strongest predictors
  • Significantly outperforms baseline prediction methods