AI for Football Player Market Value Prediction: Bridging Positional Performance Data

Abstract

The use of artificial intelligence by football clubs has grown considerably, especially in the transfer market, which requires clubs to analyse several variables. To accurately estimate a player’s value, multiple factors, including their traditional box score statistics along with their on-ball creative capabilities and marketability off the field, are combined. This paper evaluates the use of position-based performance metrics, On-Ball Value (OBV) and Expected Threat (xT), in the accuracy of predictive AI-based learning models in comparison to conventional measures (goals, assists, appearances). In addition, this study also investigates the effect of career-stage factors (age, health, proximity to peak performance) and the online presence of players to manage the relationship between AI-predicted performance and observed market values. Guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol, a systematic search and review process was conducted, whereby articles reviewed in Scopus databases were selected, along with the inclusion of a qualitative study using transfer-market data to supplement patterns found through the quantitative analysis. Position- specific on-ball value (OBV), and eXpected Threat (xT) statistics often outperform conventional data metrics in terms of their predictive ability. Career stage factors provide a non-linear relationship between player performance and their estimated value, being most sensitive in mid-career while becoming insensitive if there is any injury or age risk associated with it. A player's digital presence can influence their valuation positively but provides diminishing returns when scaled. The study gives a reproducible, literature reviewed and grounded framework that bridges AI-predicted performance and real value.