Data Analytics in Football

Digital transformation is one of the great challenges of our time. On a smaller scale, this also affects sports and football. Advancing technology development and digitalization have led to a rapid increase in measuring devices, data collection, data volumes and possibilities. Meanwhile, a separate industry has been established dedicated to the collection, analysis, interpretation and marketing of performance and match data.

The world's major data companies, IBM, Intel, SAP and Microsoft, are competing for the best data analytics tools and also using sports as a demonstration domain for the power of their brand & products.

Analytics in football is not such a young discipline. Already after the Second World War, people started to collect data about football matches and to "play" with it. Of course, this was done with pencil and paper. It wasn't until the publication of "Moneyball" in 2003 : The Art of Winning an Unfair Game' that sports analytics in baseball was actually introduced to a wider audience. "Does it matter how he gets on base?" No.

Data Analytics in football Quantitative Analysis

A decade or so ago, the available data was limited to statistics on goals, hits, number of corners, possession, passes, etc. That's when you can talk about the era of Quantitative Analysis. This data has limited value to the coaching staff. While it may be concerning if a team concedes too many shots or has too little possession. However, knowing this fact does not provide unexpected insights to get better and win more games.


Performance Analysis

The second phase came in the form of performance data. This involves the use of various tracking technologies. Athletes wear a transponder that sends a specific signal to base stations around the field. The data received is then analyzed in real time. Among other things, running performance, sprints and speed are evaluated. This performance data is aimed at improving sports performance in competition. Training sessions are also analyzed, with continuous monitoring of players also used for performance management and injury prevention.


Advanced Metrics: Expected Goals, PPDA, VAEP an more

The Expected goals (xG) metric (origin is disputed) marked a quantum leap in data analysis. Football is and remains a sport with few goals. That is why the result often does not correspond to the actual course of the game or the chances played out. Expected Goals makes the performance of teams and players more comparable by measuring the number and quality of scoring chances per player. This was followed by many other metrics that we will discuss in later posts. These include, Expected Assists (xA), Expected Threat (xT), Expected Goal Chains(xGC), Valuing Actions by Estimating Probabilities (VAEP) and Passes Allowed per Defensive Action (PPDA) and a few more.


Match Events

Another quantum leap in data collection has been match event data. Specialized companies categorize and quantify thousands and thousands of events per match, with a great deal of manual work. For each game event, the associated (x,y) coordinates are also recorded. Several thousand events are thus recorded per game. This event data is valuable for the clubs, because in addition to the quantitative aspect for each action, space gain and importance of the action can be included in the analysis. The interpretation of the analysis is used in match analysis, opponent analysis, scouting and coach scouting. Thus, with data analytics, one is able to analyze thousands and thousands of games simultaneously at the push of a button.



Resistance in football

Until a few years ago, football was thought to be immune to this trend. Those in charge initially remained very skeptical, even dismissive. Most thought that football was different from American sports and that it would be pointless to analyze match data. But a few dreamers disagreed and set out on their journey despite doubts and rebukes.


Meanwhile, data analytics "early adopters" are reaping the competitive advantage that investment in people and data analytics is beginning to give them: Liverpool, Midtjylland, AZ Alkmaar or Brentford are just a few in the fast growing list. In my opinion, clubs that don't jump on the bandwagon run the risk of falling behind.


Systematic data analysis allows clubs to make better decisions in scouting and match analysis based on data. As for many companies, digital transformation is a major challenge for many soccer clubs. Before you can realize the great potential, you need time and investment in technology, processes and culture.


Data does not include results

Whether you are a football club or a company from another industry, the same principles apply. Data alone is not enough to discover new things. Data does not automatically lead to insights. Analyzing and interpreting data is hard work.


The challenge is to interpret the data correctly. And this is not about technology, processes and methods, but primarily about football know-how, openness and curiosity. To gain real and valuable insights, it takes numerous iterations between soccer knowhow and data knowhow.


Copernicus had the same data as everyone else, he saw, heard and learned the same things as his contemporaries. But he came to completely different conclusions. Because he questioned assumptions that were not up for discussion for everyone else.


You need people, competent people, who can ask the right questions, develop hypotheses and validate them in the data. On the one hand as validation of one's own hypotheses, but also to gain new insights that have not yet been thought of, which in turn raise new exciting questions. Data helps to see things that remain hidden to our mind and eye.


Distortion of Perception

When interpreting data, we often have to dribble out our own perceptions as well. The Confirmation bias is a distortion of perception. The human mind tends to uncritically look for evidence that confirms its worldview. The few salient events tend to be overrated, but are less meaningful than the many many other events. We also often forget to look for evidence that disproves our hypothesis.


Future of Data Analytics in football

We are only at the beginning of the data analytics evolution. What is considered "state of the art" today will be the standard in a few years. But we are far away from algorithms deciding which game system is used, which player is signed, who is substituted out and in. And we should not strive to get there either. Football know-how is and will remain at the center.


Let your own data guide you.

But don't follow the statistics. Follow your strategy!



Blog from www.footballytics.ch About Data Analytics insights in football - improve the game - change the ǝɯɐƃ

Share this entry

0 Kommentare

Aktuelle Beiträge

Alle ansehen