Estimating the market value of attacking football players using multiple linear regression

Abstract

The challenging problem of understanding the value of football players inspired countless analyses both in the academic and professional world. This paper has introduced a regression model to estimate attacking football players’ market value, building on the findings of prior academic research. The model included performance metrics, personal information metrics, and commercial potential metrics as well. The alternative model introduced was derived from a sample consisting of Premier League attackers, but the regression was later tested on attackers from the Bundesliga as well. Using cross-sectional data, including 105 attacking players from the English Premier League, 27 metrics were exported. All data were retrieved from freely accessible databases (Transfermarkt, FBREF, Instagram). This research paper successfully introduced two new variables not considered by previous academic research: (1) commercial potential and (2) nationality based on the homegrown rule. The proposed alternative regression has an R2of 0.65, eliminating multicollinearity and heteroskedasticity, and out-performs the regression models based on prior academic literature, both in-, and out-of-sample. Based on this study, the most important metrics to estimate a football player’s market value are (1) expected goals and assists, (2) pressures, (3) player’s age, (4) player’s nationality, (5) club’s prestige, and (6) player’s commercial potential.

Date of Award	18 Oct 2022
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Julien Fouquau (Supervisor)