Selecting Elite Footballers into League- and Season-wise Teams of Best Players Utilising Machine Learning

End-to-End Data Project

The FIFA Team of the Season (FTOTS) often draws criticism for subjective or popularity-driven selections due to its human-based voting system. This project (the final year project of my degree programme) explores whether machine learning can provide a more objective and performance-aligned method for selecting season-best squads across Europe's top five leagues.

Player data was scraped from reputable sources and processed through extensive wrangling, feature engineering, and multilayered statistical analysis. Four specialised classification models were developed, one each for goalkeepers, defenders, midfielders, and forwards, trained using historical FTOTS inclusions as the baseline.

For each league-season, the models generated the top n players with the highest probability of selection, matching the real FTOTS squad sizes. Performance was evaluated using a custom scoring framework inspired by the UEFA Champions League fantasy scoring system, enabling quantitative comparison between model selections and the official squads.

The results showed that the machine learning approach outperformed the human-based FTOTS selections in a majority of seasons, consistently identifying players with stronger statistical profiles.



Major tools utilised:              

Click on this link to access the project page.




Book a Session