The influence of performance parameters on market value

Research question: We uncover the influence of performance parameters on market value of football players in German Bundesliga. Hereby, the analyses should in some places reach beyond the current state of research by, for example, using relative operating times or running kilometres in the analysis. Research methods: To this end, we analysed all players in the season 2015/16 who had at least one significant participation in a game. Because of the unclear functional form of the links of the market value of players to the influencing variables, we carried out the analysis using Boosted Regression Trees in order to be able to map and interpret both different scale levels and nonlinearities. Results and Findings: We found the highest relative influence in the ranking of TV money from the preseason (>50%) and goals and high pass rate (each >10%). Partial-dependence plots recover the nonlinear influence of variable on the market value of players. Implications: It seems that the market value depends significantly less on sports performance than assumed. On the one hand, the preselection of the players in the respective clubs causes a clear difference in the market values of the individual players. In addition, the often-assumed linearity for some variables is just as little as the quadratic correlation, which is often assumed for old age, which is why classical OLS estimates cannot be sufficiently argued.