NL

We attempted to predict player market values using Machine Learning — here’s what we learned! 🤖💰 The challenge was to build a market valuation model focusing on basic player information: Age, Team, League, Position, Contract length, and Preferred foot. The only performance-related metrics considered were Minutes played and the DataMB Performance Index for this season. After allowing the gradient-boosting model to run and learn independently, we fine-tuned the results through recursive model enhancement. Beyond the final predictions, the process provided valuable insights into the features that most significantly influence player valuations. Feature importance refers to the degree to which different variables impact a machine learning model's prediction. The top 25 features (variables) are shown here by descending importance. Minutes Played emerged as the most impactful factor along with the Performance Index, followed by Age and Contract Length. The league in which a player competes also plays a crucial role, with the Premier League and La Liga leading the pack followed by the other top five leagues. At a similar level of importance is playing at the winger/attacking midfielder position. Unsurprisingly, the positions that follow in influence are striker, midfielder, center-back, and full-back, with goalkeepers trailing behind. While somewhat expected, one of the most interesting findings was the significant impact of a player's dominant foot on valuation. A player favoring his left-foot appears to have an influence comparable to that of major leagues and key positions. (cc @ftblnl) Also of note: Saudi Pro League's importance in the model ranks above established leagues like Eredivisie and Portugal's Primeira Liga, driven by the influx of high-profile, high-value transfers in recent times. The player Market Value prediction model will be added to DataMB Pro in the future once properly refined and implemented. Thank you for reading! Should we post the Top 25 players in valuation next?