FC Barcelona – ATM from the predictive models and the Artificial Intelligence in Sports Board of RNE.
Members of the Olocip team attended last Saturday the program ‘Tablero Deportivo’ of Radio Nacional de España in order to analyze the match between FC Barcelona and Atlético de Madrid.
The analysis, from Artificial Intelligence and Olocip’s own predictive models, allowed to know all the details about the form states of each team and its players, possible ideal substitutes based on performance predictions and probabilities of victory of each team, among others.
In this way, Ander Alcon and Asier Rodríguez, Olocip AI specialist, shared among others, the percentages of victory of both teams established from those of predictive models.
With Messi in the field: 48% victory of F.C. Barcelona – 23% draw – 29% victory of Atlético de Madrid.
Without Leo Messi: 38% victory of F.C. Barcelona – 30% draw – 32% victory of Atlético de Madrid.
“We establish predictions of each one of the variables and statistics of each player, for example, based on our own model developed with more than three million shots we can predict, for example, what is the shape of each player’s face to goal and by both teams,” explains Asier Rodriguez, IA specialist.
The final result was 2-0 in favor of FC Barcelona, so the predictions were aligned with what happened during the game.
“We have a continuous flow of large amounts of data from different sources that allow us to build AI brains adapted to each of these sources and with the aim of obtaining solutions and answers that allow for the reduction of uncertainty from the predictive and prescriptive dimensions,” argues Ander Alcon, AI specialist.
With respect to similar players, highlighted as ideal candidates to Griezzman, and considering some of its most significant variables such as assists, buld up game, vertical passes, dribbles, expected goal among others, players like Gareth Bale, with 30% similarity, Andrej Kramaric 30%, Luis Suarez 29%, Iago Aspas 29%, Edinson Cavani 29% or Ciro Immobile 29% among others.
Or for example, similar player to Lucas, taking into account some of its most prominent variables and that define the player, such as balls recovered, pass ratio, interceptions, provoked offside, build up game among others and considering both possible positions, Lateral left and central defense, highlighted Terence Kongolo (LI/DFC) (Huddersfield Town (28%), Tin Jedvaj (LI/DFC) – Bayer 04 Leverkusen (28%), Presnel Kimpembe (DFC) – PSG (28%) or Sebastiano Luperto – Napoli (28%) among others..
“In order to explain our similarity solutions it is necessary to qualify that the Artificial Intelligence allows to locate the database of the players in the future club and then, establish performance predictions of each one of the players. Thanks to this contextualization we can establish the degrees of similarity with the referent player based on performance predictions, and not only considering their similarity based on past statistics and their clubs of origin. The concern lies in knowing if the candidate player is similar to the referent and desired, in my club and in a future context, and not the considering their past statistics, which we understand is not an honest comparison, “added Mario Garrido, responsible for business development of Olocip.