Skip to main content

SoBigData Articles

Intensity vs Accuracy: Technical-tactical differences between male and female football teams

Women's football took its first steps since the early twentieth-century. Unfortunately, the ostracism from the English Football Association drastically slowed down its development, which experienced a long period of stagnation. Women's football resurfaced in the 1960s in the Nordic countries of Europe, and it is now spreading all over the world. From 2012 the number of women academies has doubled, with around 40 million girls and women playing football worldwide nowadays.

The gain of the popularity of women's football has stimulating exciting question: What are the differences between women’s and men's football? In principle, the rules and requirements of the two games are the same. In practice, as in other sports, there are natural differences between men and women in terms of physical skills, while technical-tactical differences between male and female players' are not deeply investigate yet.

To this aim, we analysed an extensive data set of soccer-logs describing all the spatio-temporal events that occur during the last men's and women's World Cups: 64 and 44 matches, respectively, and 32 men's and 24 women's teams with 736 male players and 546 female players. We quantified the performance of players and teams in several ways, from the number of events generated during a match to the proportion of accurate passes, the velocity and fluidity of the game, the quality of individual performance, and the collective behaviour of teams.

Men's matches have, on average, more events than women's ones. Specifically, women's matches have, on average, more free kicks, duels, accelerations, clearances, and ball touches but fewer passes and fouls than men's matches. Furthermore, men's passes are on average more accurate than women's ones. Moreover, on average, men kick the ball from a greater distance than women, where we measure shooting distance from the shot's origin to the opponent's goal center. Furthermore, pass velocity is lower for a female team than a male one, while woman regain the ball possession faster than men. In contrast, there is no difference in the time elapsed between two shots, between male and female teams, and men's passes are on average longer than women's ones.

We then ask the following question: Can a machine distinguish a male team from a female one, based on their technical performance? Our answer, based on the use of a supervised classifier, reveals that men’s and women’s football do have apparent differences in terms of technical features, which we investigate through the inspection of the classifier. In particular, the most important features that permits to discriminate between male and females football teams are:

●      the percentage of accurate passes in the match (AccP);

●      the average ball possession recovery time;

●      how long a team waits after a game stop before restarting the game with a free-kick, a corner kick or a throw-in;

●      the average time elapsed between two passes.

Interestingly, we find that the number of passes and shots, generally recognized as important metrics for a team’s performance, are considered less important in discriminating between a male and a female team.

This inspection allows us to highlight the characteristics of the female teams that are misclassified as male ones and vice versa, revealing interesting information about the national teams’ playing styles. For example, the figure below shows the predictions of the Decision Tree where in one case out of 21, the model misclassified a female team as a male one, while on five cases out of 31, a male team is misclassified as a female one. For example, in match Sweden vs. Mexico of the men's World Cup, Mexico is correctly classified as a male team (pass accuracy = 85%, recovery time = 30 s), while Sweden is misclassified as a female team (pass  accuracy = 72%, and recovery time = 42 s). Note that the accuracy of passes of Sweden in that game is lower than the average pass accuracy of male teams (84%), making Sweden more similar to a female team than to a male one (average pass accuracy for females is 75%).

Figure. Scatter plots displaying pass accuracy as a function of recovery time, among male national teams (left) and female national teams (right). The green circle indicates a team correctly classified, in a game; the red cross indicates a mistake. The dashed lines are at the median values for the two variables over the entire data set.

 

In conclusion, the way female teams play is more intense but less accurate and more fragmented: in a women's football match, the time elapsed between two consecutive passes is lower, and so is pass accuracy, and female teams tend to regain ball possession faster than male ones.

The paper describing this research has been just submitted to the workshop “MLSA 2020: Machine Learning and Data Mining for Sports Analytics”.

 

Written by:

Alessio Rossi (alessio.rossi@di.unipi.it)

Paolo Cintia (paolo.cintia@di.unipi.it)

Luca Pappalardo (luca.pappalardo@isti.cnr.it)