Data science and the role of data interpretation at FIFA

Throughout FIFA World Cup Qatar 2022™, you may have noticed new graphics appearing in the bottom right-hand corner of your TV screen. The metrics within these graphics are known as Enhanced Football Intelligence (EFI), and are different from the conventional statistics viewers have used to understand matches in the past. In this FIFA Insight interview, FIFA's Group Leader Football Performance Analysis & Insights Chris Loxston, football data scientist Juan Busso, and senior football advisor Arron Ackermann explain why these new metrics have been created, and show how they are useful to practitioners and fans alike.

00:48
The initiative behind the EFIs
Traditional metrics used to help understand the way a football match is unfolding include the number of shots, fouls or corners. These figures can be useful, but they don’t paint a full picture. For the World Cup in Qatar, FIFA wanted to provide extra datasets that would add a new dimension to the way fans perceived the game. With this initiative in mind, Busso's team asked experts at the pinnacle of the game what information they found most insightful. It was from these discussions that the EFIs sprang to life.

03:51
How FIFA creates EFIs 
To start the EFI process, two sources of data are required. The first is the live "events" in a match, or what goes on around the ball, such as a shot, foul or pass. The second is the live tracking data from each player on the pitch. These two sources of data are then amalgamated so that whenever an "event" takes place, we know exactly where all the players are in relation to that event. The combined data is then fed into carefully designed algorithms, which run in parallel to produce the EFIs.

08:28
Line breaks 
One of the most insightful EFIs is the number of line breaks in a game, or the number of times a pass cuts through a whole unit of the defending team. This is an important statistic for managers, because the data shows that the more line breaks a team concedes in midfield, the more games they lose. Moreover, our line detection algorithm allows us to see whether a line was broken by playing through, around or over the defenders. Thus, this EFI allows us to see how attacking tactics evolve throughout matches. 

14:30
"In contest"
Possession percentages are one of the most common metrics we see on our TV screens. But when we watch matches, we see that there are prolonged periods where neither team is in clear possession of the ball. The "In contest" EFI sheds light on this grey area, adding an insightful third dimension to possession stats. In an evenly-matched game where both teams are seeking control of the ball, the ball will be ‘in contest’ a relatively high percentage of the time, whereas this percentage will be much lower if one team is dominating the ball.