At GWI we watched with great interest as data scientists across the world tried to predict the outcome of the 2018 FIFA World Cup using data and machine learning.
One of the questions we wondered about was the way various team dynamics, playing styles and national pride could affect the outcome.
During the 2018 FIFA World Cup, 736 players from leagues across the world came together to play for their country. With representation from 62 different leagues across the world, we wondered whether the various playing styles had any impact on a country’s ability to win a World Cup.
On average, teams in the World Cup were made up of players from 9.44 different football leagues. Those teams that made it to the Quarter Finals or above had a lower average number of 7.75 teams to combine. This figure went down slightly as we reached the top 4, with only 7.25 leagues per team on average.
The average number of players per league is approximately 10.4 – but when you look at the median (the middle of all the player counts), the picture changes – it’s only 3, meaning there are some leagues with many players represented (such as the English Premier League, La Liga and Bundesliga) and lots with only a few.
Tournament winner France only had to integrate the playing styles of players from 5 different leagues, compared to the second placed Croatia, which had to bring together players from 11 different football leagues. On the other hand, in the playoff for 3rd place Belgium successfully integrated players from 11 leagues whereas England failed to capitalise on only 2 leagues representation.
In the championship winning French team, around half of the players came from the world’s biggest football leagues (English Premier League and La Liga). Another 40% of the team came from France’s Ligue 1 – a big portion of players who would be used to playing in the same style, and arguably easier to bring together as a cohesive unit.
Teams that failed to progress beyond the group stages included players from an average of 10.5 leagues. Notably, Saudi Arabia and Germany included players from 3 and 6 leagues respectively, proving that sometimes consistency doesn’t trump skill or luck.
Note, we haven’t taken into account player time on the pitch in this analysis, or how many players came from each league. We also haven’t looked at historic trends. That might be fun for another blog post!
So, if you were a national coach, what does this tell you? All other considerations being equal, it’s worth considering which league your players participate in. Representation from fewer leagues seems to correlate with increased chances of success, but it’s not a guarantee.
Dr. Vanessa Douglas-Savage
Consulting Director and CIO
- Boice, J, Wolfe J and Silver, N, 2018, Global Club Soccer Ranking, electronic dataset, Five Thirty Eight, viewed 13 August 2018, <https://projects.fivethirtyeight.com/global-club-soccer-rankings/>.
- Clayford, C, 2018, 2018 FIFA World Cup Squads, electronic dataset, Kaggle, viewed 13 August 2018, < https://www.kaggle.com/cclayford/2018-fifa-world-cup-squads>.