Professional football clubs collect incredible volumes of data in both training and matches. So much so that it is proving to be a conundrum for clubs to decide how best to collate and apply this data. The Beautiful Science spoke to Dr Robert Rein, of the German Sports University in Cologne, about how big data practices could aid football clubs.
The technology boom has left football with a need to act both innovatively and immediately. Two things football as a movement, especially in England, has not been known for over time. For a large number at the top end of the football pyramid the approach was to introduce this measuring of data, although with limited strategy or focus. Some clubs of course took more structured approaches, but many felt that simply beginning to collect this data was a good start. They could suss out what it meant for their players and how to apply it as they went along.
Certainly not a scientific approach to a scientific practice. Nevertheless, inevitable progress has been made with growth in expertise and number of sports science personnel at clubs. For example teams use data collected through GPS software to monitor working intensities, heart rates and other metrics to assist in seeing whether they are overworking players and risking avoidable injuries. Arsene Wenger’s talk of his players being in the redzone is such a case.
A long road to go
The game has progressed its use of data, but at nowhere near the same rate at which the variety of data that can be collected has grown. The majority of this being statistical/numbers based over video progressions. This is where Dr Rein’s research and approach towards big data could spell a change for how football analysts work. Robert explained his findings first: “What we have found is that clubs gather this huge amount of data from training and also from their games. The problem is they do not know how to really cope with this data.”
This seems quite a naive practice for clubs to be doing and it is not just a handful of teams who are in the same boat. Robert said: “In the Bundesliga, all the teams get all the positional data from all the games. Most of them do not even download the data to their computers because they would not know what to do with it.”
For Dr Rein, an expert in skill acquisition and dynamic system theories, the problem facing clubs is two-fold. He said:
“From the computer science side they (the clubs) don’t really know how to handle this sort of data, they are experts in the use of videos. Now with this huge volume of positional data and fitness data they don’t know how to cope. Secondly, their techniques are not up there in terms of guidance for the practitioners in how they can apply the data.”
It seems quite clear from what Robert is saying that football clubs are not maximising the potential from the data they receive and additionally there is a need to either up-skill in data management or bring in further experts to improve understanding.
Big data, big potential
In the sporting world big data is an unknown practice. Across business and investment it is a very different story. A computerised process that allows large quantities of data to be input together for analysis. The German academic can see a real potential use for this method in sport and explained the flaw in current approaches to analysis.
Big data defined
Extremely large data sets that may be analysed computationally to reveal patterns, trends, and associations, especially relating to human behaviour and interactions.
“At the moment teams look at key events of just one or two games when forming analysis. There is a really high chance component in the game. Matches are often decided by one mistake where statistically the better team lost. Goals are so seldom in comparison to a sport like basketball. This means chance is a bigger component.
“To get around this you need a lot of data to work with. Through big data you do not look at one or two games together. You look at 20 or 30 games simultaneously. This way you remove the chance component and can see the patterns that are actually happening.”
This relates directly to the definition of big data, so quite clearly there is scope to at least experiment with it in the sport. The key to the whole method is volume of data so that each outlier or anomaly of a match event or result plays less overall significance on the concluding analysis . Robert detailed where he could see the use of big data:
“You do not just look at a small sample. You add in all of one of your upcoming opponent’s games from one and half seasons, at least 20 games say, and you will get a much better idea of the tactics and strategies they might use.
“Clubs could do that now with 20 scouts and sit them in-front of the video, but that is just so time consuming. This is where big data can have a big impact.”
Case and point
Dr Rein has good working relationships with the Bundesliga and the DFB (the German FA). The Germans always seem five steps ahead of the English in the world of football, but not in the analysis department. A group of data scientists are funded to work out of Cologne University for the DFB. Robert identified their approach:
“Before the national team goes to a tournament they look at all the possible opponents and they look at so many games for each opponent. 25 odd people working for a couple of months, 24/7, looking at videos to summarise the game for Joachim Löw. This is clearly time consuming.
“A computer science based approach such as big data could make this more efficient and effective.”
Where such progressions can be made it seems foolish to not at least attempt to put them into practice.
Step forward with caution
The success of big data can also prove to be its achilles heel. The method removes the human element from the data recording and translating process, giving the collated results. The improvements in terms of time and limiting the subjectivity of human opinion or error rom the process are clearly positive. However, big data shortcomings have been found when the coding is not full proof meaning the analysis does not adequately cope with outliers. The human error is not entirely removed as the analyst still has to interpret the overall trend and decided upon what this means to the team in a sports scenario.
Such examples can frequently be found in current uses of big data in customer research and predicting trading patterns. This is also the reason why there are concerns over using this in health care where the consequences are huge.
Overall though the potential to increase the efficiency and success of analysis in football through big data are there to be seen. Too often football waits for the finished article, but by integrating this practice into their work the sport could speed up the rate at which big data becomes universally successful. Such collaboration is a sticking point that even Dr Rein, who is from a computer science background, could accept, but the possibilities should outweigh this scepticism.
Dr Robert Rein is a researcher at the University of Cologne and has published work on the uses of big data in football. Future articles on the Beautiful Science will include contributions from Dr Rein including a feature on why football may take its time over adopting big data.