What sparked your interest in this topic? Are you a football fan?
I live close to the stadium, but it was only while working on this project that I became more interested in football. I looked at this sport from the perspective of data analysis—using neural networks and genetic algorithms. This topic forms the main thread of my research. I am still impressed by how much information can be derived from statistics. How historical match data can predict the outcomes of upcoming games and which players will be key in forthcoming competitions. I am studying these algorithms and verifying their performance in Fantasy Premier League (FPL) competitions. In this virtual league, players—FPL managers—have a finite virtual budget and must choose a team while adhering to numerous rules—of course, based on statistics from real matches.
In your project, you use AI—what do you hope to achieve with this technology?
Match statistics, which serve as the starting point for my algorithms, are understandable to humans but can be rather tedious when analyzed manually. Artificial intelligence methods allow us to focus on the essence of the problem. They help find not always clear patterns, filter out outlier results, all with the help of a computer that can check many potential team lineups in a limited time.
The neural networks, genetic algorithms, and classifiers learned during this process have broad applications in modern technologies, and the experiences gathered while planning a football team lineup are also useful in many other fields
How do you view artificial intelligence—is it an opportunity or a threat?
I believe that AI is primarily a tool. Much depends on the intentions of its users. If those intentions are good, it will yield positive results; if they are bad, the effects could be disastrous.
However, I think that particularly in my project’s case, artificial intelligence represents a tremendous opportunity. By using it, we gain the potential to multiply our data processing capabilities through increasingly faster computers. This creates new perspectives, but responsibility is also crucial.
Therefore, in the next stage of my work, I would like to focus on explainable AI—that is, obtaining convincing explanations for decisions made by machines and understanding why certain players were selected.
Who will benefit from the developed technology?
My project has more of an academic potential related to research on algorithmic support for managers in Fantasy Premier League gaming. However, the mechanisms used in the project, especially those concerning player selection optimization, have broad applications in other areas of life. They can be used wherever it is crucial to select appropriate resources while maintaining a limited budget and responding to dynamically changing conditions. This universality makes FPL strategies intuitive and understandable for many people, and their application can easily transfer from life experiences to solving problems we encounter.
How will your designed manager differ from existing tools supporting the football industry?
According to estimates by Grant Thornton, the total value of all European football clubs currently stands at €4.7 billion. Many clubs thus use advanced dedicated AI tools to support decisions regarding player selection, coaching staff, or match strategies. Examples include Liverpool, which uses TacticAI to enhance its match strategies; Manchester United; and Bayern Munich. This trend is also visible in Poland—according to press reports, artificial intelligence played a key role in selecting a coach for Wisła Kraków.
This project is smaller in scale. My goal is to study automatic team lineup optimization in FPL and predict match outcomes. Unlike existing solutions, my project considers statistics from all Premier League clubs without favouring any particular one. This makes the analysis more objective and can provide valuable insights for FPL fans and sports analysts.
What is the database on which you are building your tool?
The data I use pertains to Premier League competitions and is publicly available. However, for long-term historical analysis purposes, maintaining a local version of this database is necessary. It turned out that archiving data is essential since it gets removed every year, complicating access to official historical statistics. These statistics are crucial not only for analyzing past matches or player performances but also for assessing algorithm effectiveness—historical data about other managers is needed to easily calculate one’s ranking position.
How does your collaboration with your mentor proceed?
I had never participated in a mentoring programme before; therefore, I did not know what to expect. However, participating in this project turned out to be a great opportunity. My mentor, Dr. Piotr Duch, not only motivates me to continue working but also takes an individual approach. Meetings are regular—we discuss progress and plan subsequent stages. However, my mentor always first explains concepts and ensures that I understand what needs to be done. If I have any issues, he always finds time to answer my questions and set new goals. Thanks to this support, I quickly learn new things. With his help, I am also optimizing my study programme by expanding it with topics relevant to my research and participating in online courses outside of Lodz University of Technology.
Interviewed by: Agnieszka Garcarek-Sikorska