beachboyx10 - stock.adobe.com
Bundesliga delivering insight to fans via AWS
Through its partnership with AWS, Germany's top soccer league is delivering advanced analytic insights to fans and broadcasters in real time during matches.
Germany's Bundesliga is using analytics to enhance fans' viewing experience, delivering advanced insights instantaneously to viewers at home powered by AWS.
The Bundesliga is Germany's premier soccer league, and in early 2020 it formed a partnership with AWS to use the tech giant as its technology provider.
Using capabilities from AWS, the Bundesliga has since developed a system to capture millions of data points per match, and using those data points combined with machine learning models developed features that enable fans to better understand the action they're viewing.
The Bundesliga introduced Match Facts in February 2021. The platform includes such features as Most Pressed Player to show how often any given player is being closely defended, Attacking Zones to reveal where on the pitch teams tend to focus their offense to create scoring opportunities, Average Positions to show teams' intended playing styles, xGoals to display the probability of any shot resulting in a goal, and Speed Alert to track the running speed of players during a match.
All the information feeding the Match Facts is captured by 24 cameras placed strategically around the field that create a data stream of 3.6 million data points in a 90-minute match. That data is converted to metadata, and with AWS instantly blended with information from past Bundesliga matches and used to produce insights for fans.
"Sports has always been a sandbox for innovation, and in sports, technology is an enabler for new opportunities," Andreas Heyden, executive vice president of digital innovation for the Bundesliga, said on June 2 in a presentation during AWS' virtual Machine Learning Summit. "For us, technology is a way to drive customer value and customer satisfaction."
AWS, he added, enables the Bundesliga to add that value and satisfaction in near real time.
"In the 15 months we've been working [with AWS], we've been able to accelerate our innovation pipeline and push the boundaries of what was possible before we began using the cloud infrastructure and machine learning capabilities," Heyden said.
The xGoals tool, in particular, demonstrates what the Bundesliga is able to do with AWS.
Using past data that feeds a machine learning model combined with data from a live feed, the likelihood of any shot taken throughout a match resulting in a goal can be determined in real time.
Andreas HeydenEVP of digital innovation, Bundesliga
Among the data points factoring into the probability are the number of defenders in close proximity to the shooter, angle of the shot to the goal, goalkeeper's distance from the goal line, distance of the shot and whether the shooter is running as he shoots or shooting from a static position.
In a demonstration, Heyden showed Joshua Kimmich, a player for Bayern Munich, score an improbable goal.
Kimmich was running from his left to right as he shot. He lofted his shot from just outside the 18-yard box -- 21.47 meters, according to data captured by AWS -- over the goalkeeper's outstretched hand for a goal as the keeper retreated toward the goal line from 3.35 meters out at the instant the shot was taken. Two defenders were between Kimmich and the keeper, and Kimmich had to shoot back to his left at an 18.73 degree angle to get the ball on goal.
Within seconds, the system determined that Kimmich's shot had a 6.41% probability of resulting in a goal.
By contrast, a penalty kick, which is taken from 12 yards with the goalkeeper forced to remain on the goal line until the shot is taken, leads to a goal 75% of the time.
The process for determining that 6.41% probability began with data capture, which was done by the cameras positioned throughout the stadium. The data was then sent to an AWS Fargate container where it was combined with other data about the match as well as machine learning models developed in Amazon SageMaker.
Next, it traveled to AWS Lambda, a serverless compute service in which the Match Facts data was prepared for visualization and delivered to fans in the Bundesliga app and broadcasters on both national broadcasts for viewers in Germany and international broadcasts for viewers in over 200 countries.
"It's all about speed," Heyden said. "For the whole process to take 500 milliseconds, you need a very fast [system]. You need a serverless infrastructure which is scalable. Hundreds of thousands of people might want to see the data in near real time, so it has to go to the desired outputs.
"And then a feedback loop is created to retrain the model [with added data] and the model is going to get better and better."
Like the Bundesliga, the National Football League is an AWS partner and provides similar insights to fans throughout broadcasts. The National Hockey League also uses AWS.
Other sports organizations, meanwhile, user other analytics providers to deliver insights.
Major League Baseball, for example, uses Statcast -- powered by Google Cloud -- and is instantly able to provide such information as the launch angle of a home run, the speed of the baseball off the bat, the distance of the home run and whether it would have been a home run in another ballpark.
The National Basketball Association, meanwhile, has a partnership with Microsoft.
With Match Facts now part of every Bundesliga broadcast and available to fans via the Bundesliga app, Germany's premier soccer league is working with AWS to develop more features to enhance the viewing experience of fans.
Currently, the Bundesliga is focused on adding augmented reality, virtual reality and 5G.
"When we think about media consumption in the future, we want to have an enriched broadcasting product -- are there new ways of telling the story? -- and we truly believe there is much more to be told about our game with data than we do right now," Heyden said. "We believe with our cooperation with AWS we have a very bright future and we're just scratching the surface."