The World Cup Quarter-Finals, on FIFA 2018
“Life! Still a poor substitute for video games!”
The 2018 FIFA World Cup is reaching its final stages, with only eight teams left in the most-watched sports tournament in the world. We have seen a few unexpected outcomes, from the elimination of Germany in the group stage to the Russian victory over Spain in penalty kicks, showing us the beauty of life in a probabilistic world, where anything can happen. The big question, nevertheless, remains in our heads: who will take the cup home? Or at least, who will advance to the semi-finals? Following my obsession with predictions, and instead of drawing betting-market probabilities like the ones for Torneo, I decided to visualize what the quarter-finals would look like if reality was behind a controller, using the FIFA 2018 data (yes, the video-game).
This analysis was possible thanks to this dataset. It is also partially inspired by some great work done with the same data (such as this post by James Le). Even though a few players were not featured in the game for the 2018 edition (Giorgian De Arrascaeta and Steven Nzonzi, Pedro Geromel, to name a few) there is enough info for the majority of them.
I was impressed by how basic insights were very telling in terms of what are seeing in the World Cup, from the Belgian great offensive line to the stamina of the French team. It remains an unpredictable tournament, but there are insights about players, their styles and attributes that are very closely related to their performances. To make things more interesting, in this exploratory analysis I grouped the insights into the quarter-final matches, to see what it could tell us about the games ahead.
Ratings and their distributions: Sweden vs. England
When we look at the average ratings for the players in the WC under each team there is little surprise. Brazil, France, Belgium and England lead the pack with the most well-ranked players. It would be an easy choice to pick France over Uruguay or England over Sweden if we only looked at the team mean score, at least from the aggregated perspective.The interesting revelation, however, is not in the average scores, but rather in the distribution of the ratings: Uruguay and England have more polarized teams (6.0 and 4.6 standard deviation), with some players with very high ratings and others very low, while Sweden and Brazil brought very homogeneous squads (3.0 and 3.6 standard deviation, respectively). This can have two opposing implications: either (1) a few excellent players are enough to carry a team forward or (2) players with similar skill-sets play better together. It will be interesting to see the extremes of English players vs. the consistency of the Swedish squad on the field. History has already shown us this head-to-head is far from predictable.
The older the [better / worse] — France vs. Uruguay
Does age translate into experience and ability to handle pressure or does it simply mean loss in stamina and physical condition? It turns out FIFA also has a rank for Potential, or the expected ratings for the players somewhere in the future. Needless to say, it mildly, inversely correlates with age. Players like Mbappé (19) and Rashford (20) are playing well in their early days, and therefore we should expect that experience would only add to their talents. Potential is usually interpreted as a career potential, but what if it also indicates the potential to step up in key moments? To have the capacity to improve when needed the most, just like Rashford when offered to take a penalty in the shootout against Colombia? In other words, experienced players have already had the opportunity to show all they can do, but young ones are doing it all for the first time and bring the unexpected inherited on their lack of experience.
As we can see in the picture below, the biggest disparity in age comparison comes in the confrontation between France and Uruguay. A very young French team facing an experienced Uruguayan line-up. This is very evident as we watched a great defensive squad by Uruguay (with the key offensive moves of Suarez and Cavani) and an energetic French team led by the speed of Pogba, Griezmann and Mbappé (who reached over 40 kilometres per hour more than once this year). It is a dispute between experience vs. stamina, between smart plays vs disposition. If none wins, we still have Potential, something benefiting more the European team than the South American counterpart.
Player Value and Salary — Russia vs. Croatia
In terms of average player value and player wage, the results are fairly predictable as well. Brazil, Belgium and France are ahead, with England following close. The surprise in this attribute is Croatia, with wage levels similar to those of France and England. Not that Russia would get scared by the capitalistic impact of financial disparities for the game Saturday, but they should be careful with players like Modrić and Mandžukić in a match with the biggest gap in average player value in the quarter finals. Croatian players have shown their worth in the group stage and the key Russian factor at this point is the home-team advantage. An early goal will be fundamental for the Russian team.
Clash of Titans — Brazil vs Belgium
Perhaps the most interesting match to play (watch, I mean) is Belgium vs. Brazil. Both teams have similarly-ranked players, similar wages, player values and age-range. Both bring fantastic squads (including substitutes), good records in past matches, and a great combination of experience and dynamism, athleticism and skill-set. The best way to compare them is by the average of each attributes available for the players, trying to find marginal insights into what could become the key advantage during the game.
As we can see, Brazil seems to have a slight advantage overall, with Belgium winning in strength, penalties, heading accuracy, aggression, shot power, finishing, positioning and jumping. In other words, Brazil has to play quick and Belgium has to play strong. We might see a goal for the Red Devils from a corner kick or a foul, while the Canarinhos will probably score in a counter-attack or in rehearsed plays. Basically, Neymar, Coutinho and Willian vs. Lukaku, Hazard and De Bruyne. Whatever happens it will be an entertaining match to watch.
Who is advancing?
After all, who will advance to the next round? What is the final prediction? The temptation is to say it depends as the good old statistician would. If life was as simple as video-games, I would pick Brazil, France, Croatia and Sweden. But a lot of things can happen in 90 minutes of a World Cup quarter-final, after all we live in a probabilistic world and certainty only exists post-mortem. The FIFA dataset might not be enough to dig deeper into player’s attributes and performances, but an exploratory analysis with a few highlights is more than enough to show us with data what we are watching this summer on TV.
— — — — — — — — — — — —
Code for this post can be found on GitHub.
