- Copy link
As part of Barça Sports Tech Week, on Wednesday the second edition of the Barça Sports Analytics Summit, an event that puts up for debate statistical models in which data that may help in the game’s tactical challenges and in the new areas available in the sports industry, such as the visualization of data. The event was led by Javier Fernández, head of Sports Analytics at FC Barcelona, who describes the Barça Sports Analytics Summit as “a unique event in the field of football analytics, where data scientists can meet and discuss the application of advanced data analysis in the real world.”
Speakers at the event included Sergio Lana, data scientist at FC Barcelona and other figures from the world of sport such as Dan Cervone, Director of Quantitative Research for the Los Angeles Dodgers and Suds Gopaladesikan, head of information systems and data at Benfica.
Gopaladesikan explained his work when it comes to optimizing sprints carried out by players out on the field. When a player carries out a sprint, it takes at least 15 minutes before they can repeat that effort. This period of time has implications in the game. Gopaladesikan has carried out an analysis that allows the coach to identify sprints that are carried out unnecessarily. With his world, he attempts to help his team produces high quality passes to optimize effort.
From the University of Victoria in Australia, scientist Bart Spencer analyses the movement of the ball within the team. To define the success of passes he takes into account the context in which ball is touched and the control players have of space. As such, he can collect data that indicates how many times during a game there are viable passing options, which in turn allows him to measure the risk of each movement of the ball and the frequency with which the ball moves between team mates.
Luca Pappalardo, scientist from ISTI (Institute of Information Science and Technologies) from Italy’s CNR, focuses on the study of injuries, one of the biggest problems in football at the moment and the reason for a loss of 188 million euros. Papallardo’s date work helps to calculate the probability of a player suffering an injury. Methods used up to now have a 96% fail rate when it comes to predicting injury. His approach, MATH (Memory Accurate Transparent Holistic), with Artificial Intelligence reduced that figure to 45%.
Analysis of the movement of the ball
Sergio Lana, Data Scientist at FC Barcelona, explained how the movement of the ball is measured for efficiency. In Johan Cruyff’s philosophy, in which the players who creates space to receive the ball is as important as the player with the ball, a tracking system has been implemented in which three phases of the game are identified (initiation, progress and finishing) so that counting ‘microobjectives’ can allow the identification of patterns that are repeated to overcome an opponent’s defence.
Jacob Mortensen from the Simon Fraser University explained that given the resources needed to track players via GPS or optics, he has researched how to measure the external load on player using TV footage of the game. For that, he has established various equations that allow him to calculate the distance covered and the speed of players who are off camera.
Adrià Arbues from the Pompeu Fabra University, put forward that one of the most important details in the monitoring of footballer with regards to the game, tactics and also for injuries and recovery is the orientation of the player. Having this information allows the measurement of reaction times, the training of tactics introducing the field of vision of the player as a factor and its use for younger players too. His work consists in analyzing the position of the shoulders and hips of the players using video footage.
Jan van Haarven from the University of Leuven in Belgium presented his project to calculate the possibility of predicting a goal. A project that shows up the problems of data; he used the example of an assist to expand on the subject. It is not the same to play a two meter pass to a team mate who then beats three men to score than to provide a 40 yard pass to a man who is unmarked yet both are categorized as assists. Stress levels are susceptible to variables such as position in the table, percentage of defeats, if they are consecutive or not so it has been calculated which type of players have less chance of playing well under certain circumstances.
Harvard specialist Laurie Shaw was selected as the best of all at the Barça Sports Analytics. His analysis of date refers to the whole team thanks to his study of models of how teams are set up and the tactics they employ as well as the changes in tactics during a game and the transition from defence into attack. With all this material, his data treatment opens up the chance to know how changes in tactics during a game affect the final result in a game or the risk involved with a change in tactics.
Don Cervone, Director of Quantitative Research for the US baseball team the Los Angeles Dodgers, explained that data has always formed part of baseball culture. Historically, the analysis of data has been used to compare the value of players. On many occasions it was able to acquire a cheaper player with statistics similar to that of a more expensive player. At the LA Dodgers at the moment data is used to identify the defensive level of each batter and at the same time the pitchers. Nevertheless, he admitted that it has been hard work to get the coaches to take notice of the statistical data that may help them with their decisions.
Difficulties in communication with coaches
Finally, Javier Fernández led a round table discussion in which the experts recognised the difficulty for data scientists in communication with coaches and other team staff. Van Haarven explained that the most important thing are the questions that can asked of the scientists as their work is based about giving answers or presenting information so that coaches can make a decision. Gopaladesikan made the point that the coaching staffs are used to working with videos so it is important when presenting new models of communication and language to not overwhelm coaches with specialist graphics.
Asked about whether analysis is used more in attack than defence, all agreed that the attack depended on the team and the defensive system of its rival, so it was unpredictable. Sergio Lana, on this point, added that attack is more attractive than defence but one had to take into account pressing models which are being constantly developed.
- Copy link