With the XV of France, the AI ​​enters the fray

After a successful 2022 season, the French rugby team is preparing for the 2023 World Cup. The data and its analysis will be widely used, as will the IoT. Explanations.

2022 has been a great year for France’s XV with the Grand Slam at the 6 Nations and significant victories on the autumn tour. An important deadline now awaits the national rugby team: the World Cup in France from September 2023. And the collective selected by Fabien Galthié hopes to conquer its first world trophy.

To achieve this goal, the federation makes resources, including technological ones, available to its coach. Data, artificial intelligence, predictive analyzes are therefore part of the toolbox available to France’s staff, coaches and sports scientists.

Multiple but fragmented data sources

To beef up its game in this area, the FFR has also partnered with publisher SAS. To launch the match in the field of analytics, the federation can count on a large wealth of data, in particular GPS thanks to the sensors on board the players and now also the ball starting from the Autumn Nations Series.

Data is collected during training and competition, for each player. “This typology is extremely important, but it is also worth adding event data, i.e. game action data. They allow you to position the actions on the ground and their details,” says Charlotte Douette, Data Scientist for SAS France.

Read also: What unlocks the potential of AI in 2023?

This information is retrieved from a data provider, which encrypts the actions. The analysts of the XV of France also record the actions associated with a timeline, thanks in particular to the video. Three separate databases are then created.

As in business, the Fed has faced data inflation and data dispersion. The uses consisted mainly of episodic studies. “The main ambition was to centralize all information so that everyone can access the same database, history and have a working database.”

The second part of this technology collaboration: the analysis of data assets in order to bring out the performance factors that could affect the match results.

The key as an entry point for the data

Leveraging data in sport is not new, nor is technology. Football, for example, uses computer vision solutions to extract information. In rugby, automated image analysis is more complex, especially during scrum or ruck phases. It’s a future ambition. Meanwhile, exploitation focuses on structured data.

Furthermore, the analysis solution designed for the tricolor XV allows for high granularity and therefore for associating a game action (and its data) with its video. “Staff need to have this professional feedback through video,” explains the SAS expert. But first it was necessary to specify the needs of the athletes.

The first step in May 2021 was a design thinking workshop. Short-term and long-term goals emerged. First objective selected: the key, which has become a team asset of France. This choice can be explained by the involvement of Karim Ghezal, the XV touch coach, and by “his awareness of the data”.

“It was also the ideal action to allow us to acclimate to rugby data by focusing on a fairly precise and strategic game action”, also specifies the Data Scientist in charge of the project.

Based on the actions, starting from the key, the solution provides useful information such as the pitcher’s name, composition, position, ball release speed, etc. The analyzes are based on a history that dates back to 2019 and includes just over 1000 matches, including 37 for the XV of France and 450 touches.

“For example, the user can select filters, histories, teams, land areas, in order to view a dashboard that updates automatically and provides the information sought. This will be, for example, the percentage of shots that allowed a maul or a subsequent cross,” explains Charlotte Douette.

Data & AI, decision support tool

After the throw-in, these same data-returning functionalities in the form of dataviz were developed for the other phases of the game: scrums, kicks, combinations after a ruck, etc. Users were also trained in the use of the platform during several sessions to allow them to be autonomous in searching for information.

The data scientists also helped staff members convert a business question into a request. An example (which could be important during the World Cup): what keeps the All Blacks going back the most? The answer lies in one of the developed modules (ps: for New Zealanders, the solution lies in the game combinations module).

The purpose of the platform is therefore to help in decision making before matches and deciphering after them. Usage is not real-time, nor is data feedback. However, a match is played very early, identifying the opponent’s defects and calibrating training based on the data.

“Training should look as much like the next match as possible. For their preparation we have developed a dedicated module which will allow them, for example, to predict the facts of the matches (…)” thanks to the use of an algorithm. This will indicate, for example, a number of hits or scrums and the associated ground areas.

These predictions fuel decision-making, not replace it, insists the data scientist who converted to rugby during this project with FFR. Predictive functions and a training module will be used by September to better prepare for the big meeting of 2023.

At the end of the autumn tour an evaluation was carried out between the SAS teams and the analysts of the XV of France. This aims to identify the next evolutions of the platform. Among these, the use of data from connected balloons through their integration into dashboards.

Some modules will be pushed, including the one dedicated to passages and combinations. “We have a new axis to work on,” declares Charlotte Douette, without saying too much to preserve the secrets of the tricolor group. These tools will also be made available to other national teams, including the women’s XV.