To give shivers to a fan of the Liverpool it is enough to say four words: “Corner beaten quickly… Origi!!!”.
They are the ones with which the commentator Steve Hunterhistorical voice of the Reds, accompanied the goal that allowed the English club to access the Champions League finalthen won, of 2019. On the developments of a corner kicked quickly taken by Trent Alexander-Arnoldthe Belgian striker Divock Origi he caught Barcelona’s defense by surprise, scoring in the 78th minute to make it 4-0 which allowed Liverpool to sensationally overturn the defeat suffered in the first leg (3-0).
It is no coincidence, in short, if Google DeepMindone of the research laboratories onartificial intelligence most advanced in the world, chose the Reds to make corners a perfect science. The result of this collaboration, which began three years ago, is a tool called TacticAI who is able to study corner kicks taken in the past and, based on what has been learned, suggest which positions the players should take in the future to have a better chance of scoring or preventing a goal.
In recent years, Google DeepMind has amazed the world with algorithms capable of beating champions of the age-old game Go (AlphaGo) and to predict in a revolutionary way the three-dimensional structure of proteins (AlphaFold). To predict the outcome of corners, Google researchers asked AI to analyze around 7,000 corners – played in the Premier League between 2020 and 2021 – and to examine, in particular, where the player who received the ball was and the possibility he had of directing the ball towards the goal. Thanks to this training, the artificial intelligence has learned to provide accurate advice on the schemes to adopt when attacking and defending. To the point of surpassing human strategies.
Based on the positioning of the players, just before a corner kick, TacticaAI was able to give advice that Liverpool’s real experts deemed better than existing schemes in 90 percent of cases.
Sport Artificial intelligence that predicts where a penalty will be taken by Massimo Canducci 04 January 2023
“Several artificial intelligence models have recently been developed to serve coaches and players’ decisions,” wrote Google DeepMind researchers. But envisaging collective action that includes numerous variables – such as the players’ fatigue and even the state of the pitch – is currently an imperfect science. “Corners, on the other hand, are a perfect case study, because the game is temporarily interrupted and the athletes have the opportunity to position themselves according to established patterns.”
TacticAI, in short, aims to increase a type of marking that is not very frequent: in the 22/23 Premier League season only 13.9% of the goals scored overall came from a flag.
For some time now, football – and sport more generally – has no longer been a question of talent alone. Statistics on footballers’ performances, collected thanks to increasingly sophisticated sensors, have an enormous impact from the moment of their purchase. A concrete example is Liverpool itself Kloppwhich in recent years has also triumphed in physical work Ian Graham, author of a database containing the progress of one hundred thousand players around the world. And on the data he built his success, recently, the Brentfordwho relied on sabermetrics – a statistical analysis used in baseball made famous by the film Money Ball starring Brad Pitt – to move from the lower leagues to the Premier League.
Thanks to data, the English believe they have understood why they lost Euro 2020 by Pier Luigi Pisa 14 July 2021
But who knows if a machine will really be able to replicate, in the future, the corner that Trent Alexander-Arnold took against Barça. “It was pure instinct” said the English footballer. Something that AI it cannot be taught.