Coaching assessment and AI tools: A look at the 8th Scientific Conference keynotes

How do we accomplish those goals? It begins with cutting-edge research and keeping up with the latest developments in sport.

Topics covered at the conference will include grassroots and women’s handball, coaching, injury prevention and rehabilitation, nutrition, biomechanics, mental training and sustainability, among others.

Five keynote speakers have been confirmed: Dr Julio Alberto López Gómez of the University of Castilla la Mancha, Dr Liam McCarthy of the Carnegie School of Sport in Leeds, Dr Jesper Bencke of the Copenhagen University Hospital and Dr Renata Baric and Dr Zrinka Greblo Jurakić, both of the University of Zagreb.

Baric’s presentation is titled “Mental health matters: The foundation of effective coaching in sport,” while Jurakić will speak on “Rethinking excellence in sport through the lens of safeguarding.” Bencke's keynote is titled EHF Injury Prevention Manual: From science to practice in handball.

Gómez, who will outline a project undertaken together with Eusebio Angulo Sánchez-Herrera and Francisco Pascual Romero Chicharro, will present “Position-aware player evaluation in handball: a bio-inspired and explainable AI approach.” McCarthy’s lecture is entitled “Re-imagining coach education and development: Case studies from across the world.”

Development of coach education assessment methods

McCarthy’s reimagining of coach education is concentrated mainly on assessment and ways to develop that area.

“From community and grassroots education all the way up to performance, sport coaching is a really difficult role that's not always recognised for being such,” says McCarthy.

“Support is available for coaches through professional development and coach education and development opportunities. Obviously, we have the RINCK Convention in handball, which kind of governs and guides the types of learning opportunities that coaches have in the sport. These learning opportunities have become better over time.

“One thing which we'd like to highlight through the keynote is the way in which assessment, which is a really common feature of those learning and development programs for coaches — it's just not really moved with the times, to keep it simple. It's often done in a way that it's always been done, so what we want to do is highlight some really positive examples of how assessment might be done better to give the coaches more positive experiences, but also to support their learning and development to a greater extent.”

McCarthy advocates for using assessment as part of the learning process, rather than only as a measurement of that learning.

“If a programme in its entirety is designed to help coaches learn, then why wouldn't we use assessment to do that? Typically, in my experience and through my research, what I've noticed is that it's just used as a measurement and judgement mechanism, so we want to offer something around assessment that's quite practical and that will allow people who head to the keynote to hopefully have some new ways of doing things.”

McCarthy will present actionable examples of alternatives in coach education assessment, with one of the standout cases being that of the English Premier League.

“I work with the Premier League to help them support academy coaches, so anybody working in the professional clubs in their academies, in their youth academies,” explains McCarty. “My role really was, six or seven years ago, to develop a programme — a support education programme — to help coaches who are working in that context.

“I was given pretty much free reign to experiment with assessment-related ideas. My PhD prior to that was in soccer, so my PhD was with the English Football Association. It was an evaluation of the level three coaching program, effectively, and through that evaluation came some really interesting insights around assessment, so the Premier League invited me to use those insights to do something quite different for academy soccer coaches.”

That work was published in 2023 and has recently been awarded an outstanding article award by the International Sport Coaching Journal. “It has been recognised as a really impactful piece of work,” adds McCarthy.

McCarthy has some specific background in handball, having been head of coaching for the England Handball Association in the lead up to the London 2012 Olympic Games. It was during that period that his interest in coaching assessment began. Aside from the Premier League, McCarthy works with the national governing body for swimming, US tennis, US soccer and FIFA, all focused on improving coaching education and assessment.

Using AI to understand positional and team profiles

Gómez’s topic comes from the other side of the game and addresses two different areas of player performance: Using AI to define elements of player performance and then extrapolating that to understanding what qualities are most important in success from each position or for a given team.

“We started working with applying artificial intelligence techniques to score and to elaborate indexes for measuring the performance of the players, both women and men, in different championships. Our aim is to weight the different features of the players during the matches over a championship and then establish an index which makes us an idea about the performance of the players. So, at the end of a championship, we can prepare, for example, the best line-up of the championship according to this performance. This is one part,” says Gómez, adding that the second part is the AI learning from expert opinions, for example, All-star Teams chosen for different competitions.

“It is also, we think, a very important feature because we can also contrast and explain why the experts have decided this player or this one for this position. This is more or less what we do.

Gómez explains that the AI algorithm considers differences by gender. “The features of the players are different, so we have different descriptions for the players in different positions, depending if they are women or men.”

The algorithm works based on the data available for different competitions, which means the output will vary based on whether it is, for example, an EHF EURO being assessed versus the Spanish league. The concept is still in somewhat early days, as the ultimate goal will be comprehensive descriptions of individual players and profiles of positions that will help both coaches and players in areas such as preparation for different opponents and areas they might like to improve on in their own game.

“We would like to make an automatic tool which can explain in natural language how is a player — according to these features, according to the data of the championship and according to our index, an automatic tool that can generate, in natural language, an explanation of the player. This player is characterised by a very powerful shot, is very quick, has a possession of 90% and is a very strategic player,” says Gómez, adding that as computer scientists, it is important for the trio that the programme, while complicated to build, is usable for anyone.

“We would like to generate this kind of explanation in natural language, because we think that it has many values for trainers and also for specialists.

“We know that is important to develop automatic tools closer to the teams, closer to the players and closer to the action.”

AI has become a prominent topic outside the handball context. Regarding the future of AI in sport in general, Gómez says: “I think that is something that we cannot avoid.

“This exists, it has and plenty of teams will use it, so if you don't use it, you on the one hand, first, you are lost, because you have other teams, other national teams, other leagues that are using it. They are improving their performance, and you are a romantic, but you don't have this kind of advantage — competitive advantage. So, I think that AI in sports and in handball is a competitive advantage for teams, for trainers and for players — not only for the team; also for the complete player.

“Teams have to evolve in this new era of artificial intelligence in sports, because it can be a very, very useful and helpful tool for all the structure of the clubs and it could improve a lot the dynamics of the matches, the dynamics of the competition and even the dynamics of the personal careers of players and trainers. Maybe there is a player that could be a very good player, but if it doesn’t have this information, maybe he or she is losing a way or alternative to be a better player.”

The Scientific Conference will take place on 13–14 November in Zagreb. The deadline for registration, the process for which can be found here, is on 3 October.

Photos: kolektiff images/Nebojsa Tejic/Axel Heimken/Uros Hocevar/Jozo Cabraja; Storhamar Handball Elite/Fredrik Olastuen/Stian Bye Høgsveen