# PlayeRank: data-driven performance evaluation and player ranking in soccer via a machine learning approach

**Authors:** Luca Pappalardo, Paolo Cintia, Paolo Ferragina, Emanuele Massucco, Dino Pedreschi, Fosca Giannotti  
**arXiv:** [1802.04987](https://arxiv.org/abs/1802.04987)  
**Date:** 2018-02-14  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/1802.04987v3/

## Key points

- Evaluation of soccer player performance is a topic of interest for scientific community and companies due to availability of massive amounts of data capturing all events during a match.
- There is no widely accepted metric for measuring performance quality in all its facets.
- PlayeRank is a data-driven framework that offers principled multi-dimensional and role-aware evaluation of soccer player performance.
- The framework was built using a massive dataset consisting of millions of match events from four seasons of 18 prominent soccer competitions.
- PlayeRank significantly outperforms competitors, as shown by comparing it to known algorithms for performance evaluation in soccer and utilizing evaluations made by professional scouts.
- Interesting patterns about excellent performances and what distinguishes top players from others were discovered through the ratings produced by PlayeRank.
- PlayeRank can be used for searching players and assessing player versatility, making it suitable for use in designing scalable platforms for soccer analytics.
- This innovative approach to evaluating soccer player performance could have significant implications for the sport's future development.

## Summary

The evaluation of soccer player performance has become a topic of interest for both the scientific community and companies due to the availability of massive amounts of data capturing all events during a match. However, there is no widely accepted metric for measuring performance quality in all its facets. In response, the authors designed and implemented PlayeRank, a data-driven framework that offers a principled multi-dimensional and role-aware evaluation of soccer player performance. The framework was built using a massive dataset consisting of millions of match events from four seasons of 18 prominent soccer competitions. By comparing PlayeRank to known algorithms for performance evaluation in soccer and utilizing evaluations made by professional scouts, the authors showed that PlayeRank significantly outperforms competitors. Additionally, they explored the ratings produced by PlayeRank and discovered interesting patterns about excellent performances and what distinguishes top players from others. The authors also demonstrated some applications of PlayeRank such as searching for players and assessing player versatility; showing its flexibility and efficiency making it suitable for use in designing scalable platforms for soccer analytics. Overall, this paper presents an innovative approach to evaluating soccer player performance that could have significant implications for the sport's future development.