The STS data feeds supply match information in even greater detail

28 November 2024 – The DFL began making use of digital data in German football at an early time, establishing Sportec Solutions GmbH (STS), a joint venture with Deltatre, as a centre of excellence for match data and sport technology in 2016. STS went on to establish its online information system DFL Data Hub to collect, process and distribute Bundesliga and Bundesliga 2 official match data.

From the Data Hub, clubs, their performance and media departments as well as DFL media partners receive a constant stream of data feeds on all Bundesliga and Bundesliga 2 matches and the Supercup. These feeds not only form the basis of data-based services offered by media, such as statistics overlays, live tickers and event evaluation graphics; they are also of great value to the clubs as sources of detailed information allowing them to draw conclusions regarding strategies and tactics, training programmes and individual player development, for example.

Furthermore, thanks to innovative technologies, matches can be analysed with increasing accuracy and speed, whether live or on video. The portfolio, which currently comprises more than 30 individual feeds, is therefore constantly optimised in terms of quality and quantity, and expanded further. Since the beginning of the 2024-25 season, STS has been supplying a new, combined event and position data feed that provides the clubs with an entire range of additional match information.

Giving structure to the data deluge

Feeds are simple text strings in XML format. To convert them into a format that is intelligible to humans, such as text, graphics or statistics, users need suitable software. The most straightforward feed is the master data feed that contains largely static information such as clubs, players and matchday schedules. The “Match Information” feed summarises all the data for one particular matchday, such as team lineups and referees.

The picture gets much more dynamic in match event and position data feeds, which capture the action on the pitch using specific parameters. They cover events such as shots on goal, corner kicks, free kicks, passes or throw-ins as well as running speeds, distances covered, ball possession or player positions. A raw data feed comprises the full set of data captured, whereas the processed feeds are tailored to specific topics of interest or user groups.

Why so many individual data feeds?

“The sheer quantity of data captured during a single match is huge, typically 3.6 million data points per game, which ultimately record every single movement of each player and the ball,” explains Julian Schmitz, Strategic Account Manager at Sportec Solutions AG. “Event and position data is captured independently. To make it easier to utilise all this information for specific purposes, it is processed and edited as needed, then assigned to various, fit-for-purpose feeds.”

Events during a match are still captured manually today. During every match, a speaker located in the stadium describes the live match using a code language to the writer, who is based at the Sportec Solutions collection centre and enters the events into the system in real time. In contrast, position data is recorded in a fully automated process: “Up to 16 specialised cameras placed in strategic positions across the stadium maintain a constant focus on all players,” says Lukas Glöckner, Head of Technology and Innovation at DFL, describing the underlying technology. “All this information is routed to the centralised server which applies triangulation algorithms, to compute each player’s position through a perspective from several angles 25 times per second.”

New feed: “Extended Events”

A new feed called “Extended Events”, which was introduced recently, combines and synchronises the information from both, the event data and position data feeds into a single data stream. “For example, for each shot on goal this feed specifies the pressure on the player, the Expected Goals value, the angle to goal, distance to goal and many other details. All data relevant to any particular event is combined to draw a comprehensive, coherent picture that makes the information much more meaningful,” explains Schmitz. 

A machine learning algorithm is applied to synchronise the two individual feeds based on a set of neural networks that were trained in a Deep Learning approach. The feed is generated continuously throughout the game, and updated live every minute. It is supplied in chronological order.

Added value for the clubs

The high information density of this new feed delivers enormous added value to the clubs’ performance departments, says Schmitz: “Imagine this scenario: an expert evaluating a pass, for instance, gets the player coordinates, the direction and angle of the pass, the number of defenders outmanoeuvred, the pressure on the passing and receiving players, the success probability of the pass (xPass), and whether the pass was successful.”

In the past, the clubs had to extract all of these insights on their own. The new feed relieves them of this effort while providing a more detailed picture of every event on the pitch. “Each event is embedded in a greater context – it is this comprehensive contextual information that creates the added value for the clubs,” says Lukas Glöckner.