Simple and reliable protocol for identifying talented junior players in team sports using small-sided games.

We designed and tested a protocol for measuring the performance of individuals in small-sided soccer games. We tested our protocol on three different groups of youth players from elite Brazilian football academies. Players in each group played a series of 3v3 games, in which individuals were randomly assigned into new teams and against new opponents for each game. We calculated each individual's average individual goals scored, goals scored by teammates, goals conceded, and net team goals per game. Our protocol was consistent across days and repeatable across groups, with intraclass correlation coefficients (ICCs) of 0.57-0.69 for average net goals per game across testing days. Players could achieve high success by scoring goals or ensuring their team concede few goals. We also calculated the first and second dimension of a principal component analysis based on each player's number of goals scored, goals scored by teammates, and number of goals conceded per game. Players that were overall high performers had higher PC scores, while PC scores represented the type of contribution made by a player to overall performance. Positive PC values were indicative of high number of individual goals while negative values were associated with more goals from teammates and fewer conceded goals. Our design allows coaches and scouts to easily collect a robust metric of individual performance using randomly designed, small-sided games. We also provide simulations that allow one to apply our methodology for individual talent identification to other small-sided games in any team sport.