Player databases covering 50+ leagues worldwide were screened using proprietary statistical models - Models evaluated players on output metrics (xG, xA, progressive actions) rather than traditional statistics (goals, assists) - Age, contract status, and estimated market value were used as additional