In the not too distant past, Sven Goran Eriksson coached the England national soccer team through a number of mediocre tournament performances: quarter finals in the World Cup in 2002, the European Championships in 2004 and the World Cup in 2006 – losing to Luis Filipe Scolari coached sides each time (Brazil and Portugal). One of the distinguishing features of Eriksson’s management style was his use of substitutions, particularly in friendly games.
New research by Friederike Mengel suggests that indeed, substitutions are harmful – the more substititions a team makes in the earlier stages of a tournament, the less likely they are to succeed later in the tournament. Data is used from the World Cup, the Under-20 World Cup and the Olympics Soccer tournament to achieve this result.
It’s not something I’d previously thought about, whether substitutions early in a tournament might be costly, and to some extent goes against intuition: giving more players some experience on the field early in the tournament seems like a wise thing. The author runs a probit regression of reaching the quarter finals based on the number of substitutions in the third group match, finding that the number of substitutions has a negative effect.
My only concern is that no control appears to be entered to the model for whether the match mattered or not. The only controls appear to be whether the match was won, and whether the team won its group match. One imagines a form variable could be constructed, and some measure of the quality of each team, based on previous World Cup performances, or FIFA rankings (assuming they go back as far as 1980 when the data starts). But it’s a fascinating paper, well worth a read!