Search Results (7)

Transfers in professional football (soccer) are risky investments because of the large transfer fees and high risks involved. Although data-driven models can be used to improve transfer decisions, existing models focus on describing players’ historical progress, leaving their future performance unknown. Moreover, recent developments have called for the use of explainable models combined with methods for uncertainty quantification of predictions to improve applicability for practitioners. This paper assesses explainable machine learning models in a practitioner-oriented way for the prediction of the future development in quality and transfer value of professional football players. To this end, the methods for uncertainty quantification are studied through the literature. The predictive accuracy is studied by training the models to predict the quality and value of players one year ahead, equivalent to one season. This is carried out by training them on two data sets containing data-driven indicators describing the player quality and player value in historical settings. In this paper, the random forest model is found to be the most suitable model because it provides accurate predictions as well as an uncertainty quantification method that naturally arises from the bagging procedure of the random forest model. Additionally, this research shows that the development of player performance contains nonlinear patterns and interactions between variables, and that time series information can provide useful information for the modeling of player performance metrics. The resulting models can help football clubs make more informed, data-driven transfer decisions by forecasting player quality and transfer value. Full article

The financial valuation of professional football players is influenced by multiple factors that evolve throughout a player’s career. This study examines these determinants using Gradient Boosting Machine Learning models, segmented by three age categories and three playing positions to capture the dynamic nature of player valuation. K-fold cross-validation is applied to measure accuracy, with results indicating that incorporating a player’s projected future potential improves model precision from an average of 74% to 84%. The findings reveal that the relevance of valuation factors diminishes with age, and the most influential features vary by position—shooting for attackers, passing for midfielders, and defensive skills for defenders. The study adopts a dynamic segmentation approach, providing financial insights relevant to club managers, investors, and stakeholders in sports finance. The results contribute to sports analytics and financial modeling in sports, with applications in contract negotiations, talent scouting, and transfer market decisions. Full article

This study aimed to examine the influence of the Video Assistant Referee (VAR) system on match performance indicators in professional football, specifically within the Turkish Super League. The objectives were two-fold: (i) to compare match variables such as yellow cards, red cards, goals, penalties, fouls, and offsides between seasons with and without VAR, and (ii) to analyze the evolution of these variables across four consecutive seasons following VAR implementation. A total of 2636 matches were analyzed, comprising 1224 matches played without VAR (2014–2018) and 1412 matches played with VAR (2018–2022). Match data were obtained from InStat Scout^® and included key indicators directly associated with referee decisions. Statistical analyses included the Independent Sample T-Test to assess differences between the pre- and post-VAR periods, One-Way ANOVA with Tukey post hoc tests to examine seasonal trends post-VAR, and generalized linear models to identify the effects of VAR implementation on each performance variable. The results revealed significant reductions in fouls, yellow cards, and offsides (p < 0.001), and a significant increase in penalties awarded (p < 0.001) following the introduction of VAR. No statistically significant differences were found for red cards or goals. Furthermore, the number of fouls committed showed a consistent decline across each season after VAR implementation, suggesting a long-term behavioral adaptation by players. These findings underscore the lasting impact of VAR on the dynamics of professional football matches and highlight the need for players, coaches, and referees to accordingly adapt their strategies. This study contributes to the growing body of evidence supporting VAR’s role in improving decision accuracy, though its broader implications for game flow and player performance warrant further investigation. Full article

Evaluating player actions in football is essential for understanding match dynamics and optimizing team strategies. Traditional models, such as the widely adopted Expected Threat (xT) model, assign static threat values to pitch zones without considering real-time player positioning, leading to a limited representation of the evolving tactical context. To address this limitation, we introduce the Dynamic Expected Threat (DxT) model, which adjusts threat values dynamically by integrating off-ball player positions. DxT refines the probability of shooting and ball movement using an Expected Goals (xG) model that incorporates off-ball player positioning. Built on event data from professional football matches, which encompasses over 335,000 actions, our results demonstrate that DxT significantly improves upon traditional xT models by offering a more accurate and dynamic evaluation of action threats. This framework enhances tactical analysis, providing valuable insights for coaches, analysts, and scouting professionals seeking a more realistic approach to performance evaluation. Full article

Soccer is a multifactorial sport, in which players are expected to possess well developed physical, psychological, technical, and tactical skills. Thus, the anthropometric and fitness measures play a determinant role and could vary according to the competitive level. Therefore, the present study aimed to verify differences in body composition and physical performance between two soccer team. 162 young soccer players (from the Under 12 to Under 15 age categories; age: 13.01 ± 1.15 years) of different competitive levels (elite—n = 98 and non-elite—n = 64) were recruited. Anthropometric characteristics (height, weight, lengths, widths, circumferences, and skinfold thicknesses (SK)), bioelectrical impedance, physical performance test as countermovement jump (CMJ), 15 m straight-line sprints, Yo-Yo Intermittent Recovery Test Level 1 (Yo-Yo), and 20 + 20 m repeated-sprint ability (RSA)) were carried out. In addition, Body mass index (BMI), body composition parameters (percentage of fat mass (%F), Fat mass (FM, kg), and Fat-free mass (FFM, kg)) and the areas of the upper arm, calf and thigh were calculated, and the Bioelectric Impedance Vector Analysis (BIVA) procedures were applied. In addition, a linear discriminant analysis was assessed to determine which factors better discriminate between an elite and non-elite football team. Many differences were observed in body composition between and within each football team’s category, especially in triceps SK (p < 0.05), %F (p < 0.05), and all performance tests (p < 0.01). The canonical correlation was 0.717 (F_(7,128) = 19.37, p < 0.0001), and the coefficients that better discriminated between two teams were 15 m sprint (−2.39), RSA (1−26), suprailiac SK (−0.5) and CMJ (−0.45). Elite soccer team players present a better body composition and greater physical efficiency. In addition, BIVA outcome could be a relevant selection criterion to scout among younger soccer players. Full article

Background: the machine learning (ML) techniques have been implemented in numerous applications, including health-care, security, entertainment, and sports. In this article, we present how the ML can be used for building a professional football team and planning player transfers. Methods: in this research, we defined numerous parameters for player assessment, and three definitions of a successful transfer. We used the Random Forest, Naive Bayes, and AdaBoost algorithms in order to predict the player transfer success. We used realistic, publicly available data in order to train and test the classifiers. Results: in the article, we present numerous experiments; they differ in the weights of parameters, the successful transfer definitions, and other factors. We report promising results (accuracy = 0.82, precision = 0.84, recall = 0.82, and F1-score = 0.83). Conclusion: the presented research proves that machine learning can be helpful in professional football team building. The proposed algorithm will be developed in the future and it may be implemented as a professional tool for football talent scouts. Full article

The identification of football talent is a critical issue both for clubs and the families of players. However, despite its importance in a sporting, economic and social sense, there appears to be a lack of instruments that can reliably measure talent performance. The aim of this study was to design and validate the Nomination Scale for Identifying Football Talent (NSIFT), with the aim of optimising the processes for identifying said talent. The scale was first validated through expert judgment, and then statistically, by means of an exploratory factor analysis (EFA), confirmatory factor analysis (CFA), internal reliability and convergent validity. The results reveal the presence of three factors in the scale’s factor matrix, with these results being confirmed by the CFA. The scale revealed suitable internal reliability and homogeneity indices. Convergent validity showed that it is teammates who are best able to identify football talent, followed by coaches and parents. It can be concluded that the NSIFT is suitable for use in the football world. Future studies should seek to confirm these results in different contexts by means of further CFAs. Full article