Football analytics of the future
For the players, managers and coaching staff at football clubs the timescales in football are very short. They are interested in next week’s opposition and getting things right for the rest of the season. A manager’s job often has a very limited lifetime. If they don’t get results, they are gone. Whatever platitudes they might offer at press conferences; with the exception of long-term success like Arsene Wenger and Alex Ferguson there is little incentive for managers to invest in the long-term future of the club.
Analytical tools for football take time to develop. The development time of a useful tool is something around a year or two. A tool, for example to analyse passing networks or evaluate how players contribute to goals, needs to be built in collaboration between mathematicians and footballing staff. Many iterations are needed so that the statistical, technical and footballing components can be successfully combined.
Given the time required to incorporate mathematics and analytics in to a football club, it isn’t the manager who will drive this development. He doesn’t have the time, or the incentive, to run such a long-term project.
More and more clubs are adopting long-term analytics projects, though. In 2012, Arsenal bought performance analysis company StatDNA, and are now using its services throughout their football operations. During their league winning 2015–16 season, Leicester City’s first team analysts used Opta data to produce the post-match briefings given by the manager. Liverpool’s head of sports analytics, Ian Graham, has a PhD in theoretical physics from Cambridge. Bayern Munich have a massive database of all their players movements during every match and for many of their training session.
These are just a few of the teams investing heavily in analytics. When writing the new chapters for the paperback version of Soccermatics, I had a chance to talk to quite a few of the people working within clubs. What I found out was that there is still a massive amount of room for innovation. Although most of the clubs are aware of the power of data and do utilise basic statistics, very few clubs are using mathematics to look at tactical aspects of the game. They have hard drives full of data, they still are not fully able to exploit.
Most of the data I looked at in earlier articles in this series are based on on-the-ball data from Opta. Every time a player touches the ball, the position is noted by an Opta operator. The type of action (pass, interception, shot etc.) is also noted. This is the data that most websites and newspapers use, and I have used so far in this series to build passing networks, heat maps and defensive maps.
But this on-the-ball data is not the only data available from matches. Clubs have access to detailed tracking data of players’ positions. And developing tools to understand this data is the future of the game.
To give a taster of how this data can be used I the animation below shows a short segment of a match.
The animation is from work by Kotzbek Gilbert at the University of Vienna. He uses data collected by Prozone sports in order to analyse how the team moves. I asked him which match it was from, but he doesn’t know. He stores the data in anonymised form to prevent him from incorporating any bias in to his research.
From this data he can track positions of specific players. A forward and a goalkeeper during a whole match are shown below.
The position of the ball is also tracked. The plot below shows its trajectory in two dimensions during an entire match.
In the next few articles in this blog I’ll be looking at Kotzbek and other researchers attempts to understand this data and turn it in to football knowledge. It isn’t an easy task. And it will take large-scale investment on the parts of clubs to make it work. But analysing tracking data and developing tactics is the future of football. It is football analytics 2020.
KOTZBEK G. u. W. KAINZ, 2015: GIS-based Football Game Analysis — A Brief Introduction to the Applied Data Base and a Guideline on How to Utilise It. In: Proceedings of the 27th International Cartographic Conference, August 23–28, 2015, Rio de Janeiro, Brazil.
This story originally appeared on Nordic Bet’s Blog.