Google's DeepMind team claims to have demonstrated the effectiveness of AI models in predicting upcoming outcomes. soccer It creates not only the set pieces for a football game, but also the on-field tactics.
Football – called stateside soccer – is the most popular sport on the planet. It is estimated that around 1.5 billion people tuned in to watch the 2022 World Cup in Qatar.
Using a graph machine learning model and data from 7,176 corner kicks, the team at Deep Mind was able to build a tool called TacticAI that accurately predicts the first receiver of the ball and the immediate result of the kick. Ta. According to a paper published in Nature Communications, it can also predict the likely outcomes of different set-up positions and generate tactical variations that can help improve the outcome of a game.
Zhe Wang and the DeepMind research team collaborated with Liverpool FC, six-time champions of Europe's most prestigious competition, the UEFA Champions League, to collect data and evaluate the model with five soccer experts. did.
This model was built to analyze corner kicks, which are set pieces where a team has the opportunity to launch a spherical ball in front of the goal they are attacking. The developers said it could also be adapted to other set pieces, including free kicks and throw-ins.
“Identifying the key patterns of tactics carried out by rival teams and developing effective responses is at the heart of modern football. But doing it algorithmically remains an open research question. ” states the paper.
TacticAI includes both predictive and generative components, allowing coaches to effectively sample and consider alternative player setups for each corner kick routine and select the player with the highest predicted chance of success. ,” DeepMind added.
A survey of soccer experts showed that the model's recommendations were favored over existing tactics 90% of the time. “TacticAI achieved these results despite the limited gold standard data available,” the research team said in their research paper.
“We have demonstrated an AI assistant for soccer tactics and provided statistical evidence of its effectiveness through a comprehensive case study with human expert evaluators at Liverpool FC,” the paper concludes. “The system's latent player representation is a powerful means to search for similar set-piece tactics, allowing coaches to analyze relevant tactics and counter tactics that have been successful in the past.”
As the United States looks forward to hosting the next FIFA World Cup soccer tournament in 2026, it seems reasonable to assume that this will be the first tournament where teams widely employ AI as tactical assistants.
Algorithm or not, England will need more than AI to win the tournament, especially when penalties are involved. ®
