Computers, not managers, will soon decide tactics - Germany head
Clemens (centre) trialled new match analysis technology during the Confederations Cup
Written by Simon Austin — March 7, 2018
TACTICAL decisions during matches will soon be made using positional data and machine learning rather than a manager's subjective judgement, according to the head of scouting and analysis for the German national team.
Christofer Clemens helped trial new match analysis technology for Fifa during Germany’s Confederations Cup final win over Chile last summer. Staff of both teams were given tablets providing live positional data along with video match footage with a 30-second delay.
The data was collected using an optical tracking system and provided metrics including speed, passing, pressing and so on. Each team was given three tablets - one for an analyst in the stand, one for an analyst on the bench and one for the medical staff.
Clemens said: “The use of real-time data in the technical area is a real benefit. Data adds another component to the relationship between the technical scouting and analysis department and the coaching team.
“Up until now, the processes for impacting the game were purely backed up by subjective observations – with all pros and cons. Reliable data now allows to validate these decision-making processes much more objectively.
"The common technical-tactical data is complemented with physical data and, more importantly, can be linked to footage from a number of different camera perspectives allowing the coaching staff to have a much better overview and more easily explain certain decisions to the players.”
However, he believes this technology will soon be cranked up another gear, with machine learning analysing data and giving “potential solutions almost automatically”.
He added: "Positional data – particularly taking the position of the ball into account – will provide us with all relevant constellations for the opponent and the ball.
"The future of data analysis will, in my opinion, lie in the use of real-time positional data and modelled machine learning during matches.
“By using clear patterns, it is possible to calculate behaviour and potential solutions almost automatically. Fast neuronal networks or pattern recognition on the basis of pre-defined parameters will help to discover clear tactical instructions and above all formulate them so they can be conveyed to players in an easy-to understand way.
“For this, the underlying technology for making sense of the data will be as important as the data itself. It will still exclusively be down to the players on the pitch to find the best solutions in constantly changing conditions but all of this could however make it easier for them.”
-1.png)
