Minute 90 4. True to its nature, ElDerbi de Barcelona closed matchday 24 with an agonizing ending. The fans of RCD Espanyol in Barcelona saw how Luuk de Jong’s header took victory away from them. It could not be.
FC Barcelona tied and closed the match with an unexpected ending, but it was not the only twist. Five minutes before the end of the first half, Sergi Darder tied the game for the blue and white. It was a decisive goal to temper the match and shake the RCDE Stadium. And particularly valuable: especially if we take into account that the probability of that goal scoring was 1.9%. Very few.
This data reached our screens just a few seconds after the goal was conceded. Yes, the paths to goal are inexorable, but they admit a certain predictive capacity. This is done by the new algorithm developed by LaLiga Tech with Microsoft technology, the result of which is incorporated into the goal replay.
This allows fans to know first-hand the probability that goal had of actually being scored. “The probability that that goal ends up being a goal is defined according to the conditions in which that shot was taken and based on what technology has been learning from the history of similar shots made in the past,” explains Ander Doncel, scientist. LaLiga Tech data.
A record of more than 20,000 shots on goal obtained from 2,000 matches from which the variables involved in the goal play are extracted and which simultaneously collects this new metric developed by LaLiga’s Mediacoach football analysts together with the Business Intelligence
The result is a graph practically in real time – it appears on the screen with the second repetitions of the LaLiga broadcasts – that accompanies the player and that concludes his scoring effectiveness in the circumstances in which the shot was taken, something that does not had been achieved until now at the audiovisual level due to the complexity of this data.
To do this, they have had the technological capacity provided by the artificial intelligence and machine learning of Microsoft Azure within the joint project Beyond Stats, an initiative that arises from the reinforced alliance between LaLiga and Microsoft with a clear objective: to bring fans closer to the metrics of performance with which the clubs’ technical bodies work.
The new model adds a scoring efficiency variable depending on the player’s position, an innovation from LaLiga Tech. “A shot made by a proven scorer is not the same as a shot made in the same situation by another less effective player in front of goal. , because the effectiveness is different,” explains Doncel. Thus, the footballer’s scoring ability enters the equation, giving rise to a graph that is already a pioneer among European leagues, adding a quality value that the model takes into account when calculating the probability.
On the screen it is only a number, but behind it there is a huge amount of data that goes far beyond what happens seconds before the ball crosses the goal line. “Every time a game starts there is a flow of data, 25 times per second, that tells us the position of all the players and the ball. What we do is apply our model constantly throughout the entire match,” explains Fermín Martínez, data engineer at LaLiga Tech. The probability focuses on goals, but could look at any shot or pass. Any movement or action can be analyzed according to its probability of coming to fruition.
Let’s imagine that a player has shot at goal and missed. What would have happened if he had passed it to the footballer next to him? The algorithm allows us to know if that player would have been more likely to score. “Calculating the probability of a goal at all times and in any situation gives a lot of play,” points out Martínez. “The objective is to attract the fan by offering them a series of innovative metrics that until now have not been shown and that provide information,” says Doncel.
It is more than just an evolution regarding the percentage of possession or number of shots on goal, metrics already known to the viewer. In this case they are more advanced. “It is the difference between showing descriptive data of what has happened in the past, to giving more predictive data that contributes a little more,” Martínez resolves. Beyond probability, it is an algorithm that, as Doncel completes, allows them to “identify players who are very effective in front of goal, those who always manage to score, even on the least likely occasions.”
Minerva Santana is Director of Global Innovation and Development at LaLiga Tech and one of the people who best knows everything that is hidden behind, apparently, a simple number. “In total, in the 90 minutes of the match, more than 3 million data units are generated from the information collected from the players and the ball. This translates into 25 frames or data units per second.”
It is an advance in terms of quantity, but also quality, since this tool allows us to analyze everything that goes beyond statistics. “Adding goal probability to 30-second broadcasts creates a new source of conversation, debate and understanding for audiences,” observes Santana.
Behind every goal that was – but that could not have been – stories unintentionally emerge that, when shared with the viewer, provide knowledge and make their experience and viewing much more complete, adding greater critical and interpretive capacity. “There are many interesting metrics that are used for sports analysis and, by adding our version to the real-time environment, we generate greater knowledge of these tools, expanding their use,” completes Santana.
Intra-stories are also created within a match. Readings that “would otherwise have been lost, and generating conversations that extend far beyond the game itself.” Something possible thanks to the synergy generated between LaLiga and Microsoft and that lays the foundations for what is already more than just a trend.
A way forward: personalization, interactivity and immersion in the game, anywhere. “The future lies in the use of AI to generate solutions that improve the experience of all fans: those who attend the stadiums, those who watch us on TV and those who consume our content through digital channels,” Santana concludes.
