For his Master’s thesis in Computer Science at the University of Twente, Jelle Maas set out to answer exactly that question. In collaboration with FC Twente, he studied how substitute players influence the performance of professional football teams and how data and machine learning can support scouting and substitution strategies.
For Jelle, the topic felt natural from the start. Football has played an important role in his life for many years. “I have been playing football since I was six and have played in almost every position,” he says. “Being able to combine that passion with my studies felt like everything came together.”
What changes after a substitution?
A substitution is more than simply replacing one player with another. When a player comes on, it can affect positioning, coordination and tactical decisions within the team. The impact is therefore not only individual, but also collective. In his research, Jelle combined different types of data. This allowed him to look not only at individual actions, but also at the effect of substitutions on the team as a whole. How do team dynamics change? What happens tactically after a substitution?
Using machine learning, he analysed large amounts of data to identify patterns. Not to explain a single match, but to uncover broader structures in how teams function after a substitution. These techniques make it possible to compare player profiles and better understand how new or substitute players fit into a team.
His background as a footballer helped him in this process. Many patterns in the data felt familiar based on his own experience on the pitch. At the same time, he also discovered the limits of data. “Data can provide valuable insights,” says Jelle. “But if you do not understand what the data actually represents, it is difficult to interpret it correctly.”
Comparing players with data
In addition to substitutions, Jelle also focused on scouting. How can data help determine which players are a good fit for a team? By applying machine learning and statistical methods, he compared players based on their performance. This makes it possible to look beyond individual statistics and build a broader picture of a player’s profile. That can support the identification and integration of new or substitute players within a team.
His research shows that machine learning can be a valuable addition to scouting and decision making around substitutions. Data does not replace the judgment of coaches and scouts, but it can support and strengthen their decisions.
According to Jelle, there is still much more potential. “It strongly depends on what data is available,” he says. “But I think there is much more possible beyond my research. For example, you could think about monitoring a player’s fitness during a match, adjusting the formation based on the opponent or determining the ideal moments to make substitutions.”
Studying at the intersection of science and professional sport
Jelle’s research shows how data and science can offer new insights into a complex team sport such as football. For him, the project was also a special experience. Completing his thesis at a professional football club gave him a unique look behind the scenes of top level football.
“How often do you get the chance to do your thesis at a professional club?” he says. “The project taught me a lot and gave me many great experiences.” It also confirmed his interest in the field: “I would like to return to this sector in the future. There are many interesting opportunities to apply data.”
Within the Master’s programme in Computer Science at the University of Twente, students work on challenges in data science, artificial intelligence and algorithms. Theory and practice come together in real and complex environments. For students, this means they not only learn how models work, but also how to apply them in real world contexts. Whether it is football, healthcare or the high tech industry, data plays an increasingly important role. Data may not be on the pitch. But it is definitely part of the game!
