Or: What if your Football team can't afford an astrophysicist?
The power and importance of analytics and data science in football is no longer questionable
When Laurie Shaw - former astrophysicist, policy advisor for the British government, and a lecturer and research scientist at Harvard university - tweeted about his next career move in January, not many would have guessed that the articles covering the story would appear in the Sports sections of news websites and papers.
Shaw did operate a well-known data science-oriented football analytics blog and was a founding member of "Friends of tracking" (a professional Youtube channel composed of data analysis experts). However, for a man of his rich resume and range of research interests (varying from astrophysics and cosmology to finance and economics) - Sports was not the immediate niche one would have placed him in. Yet, Shaw's next career destination was none other than Manchester City football club where he would join as a lead AI scientist.
The pursuit of a competitive edge isn’t an exclusive luxury of rich clubs anymore
But the real question is maybe this: why would a super successful football club in the midst of an incredible winning streak, and on its way to winning three championships in four years sign an astrophysicist? Firstly, City believes in the power of data science and the use of analytics. Upon Shaw’s arrival at the club, he was welcomed by a legion of experienced analysts dealing with millions of stats and numbers daily with one goal in mind: helping the team win more.
Secondly, there's context. The hiring or poaching of Shaw shouldn't be looked at as a ‘one off’ or in isolation but as an escalation in a tough battle between the European elite teams in the pursuit of a competitive edge. Liverpool - one of City's biggest rivals on the pitch - is widely regarded as a leader in the field of football data science and has William Spearmen, a Harvard graduate with a Ph.D. in particle physics, as their lead data scientist. In a sport of such fine margins, the Citizens felt the need to close the gap.
But perhaps there's one more explanation. A more practical one. Man City added Shaw to an already huge team of data scientists, operating dozens of cutting-edge analytical tools and platforms because it had the budget to do so. Like Liverpool and many other elite European clubs, the word "affordable" is almost a non-issue or cause for concern regarding spending on the potential improvement of the team's performance. Legacy, prestige, and major income are all factors of success on the pitch, and while the European giants can use unlimited resources to achieve it - sub-elite clubs cannot.
When it comes to the European giants, the word "affordable" is almost a non-issue
Sub-elite clubs can’t afford an astrophysicist. But do they really need to?
But that reality is changing fast. American Sports have been relying heavily on analytics for many years. Now there's a universal realization that data in football is crucial, just like it is in American football, pro Basketball, Baseball, and Hockey. "Moneyball" was the trigger for the rise in popularity of Sports analytics back in 2011, but today the discussions are more widespread: commentators integrate stats into broadcasting and use analysis tools in their tactical previews, Football media covers analytics-based success stories, and data scientists are operating blogs and sites dedicated to analytics.
Another evolution, the latest one, came when top players started to understand the benefits of data and used numbers and performance KPIs to show clubs their importance. In fact, Kevin De Bruyne, a key player for Manchester City, hired data analytics companies to showcase how important he was to the team in order to secure a new and improved contract.
So, if the importance of football data analytics is so well understood and its benefits are clear and concrete, the next question should be - why doesn't everyone use it? Why don’t all clubs (and academies) get access to solutions that are clearly the next evolution in competitive football? Well, because not everyone can afford to hire an astrophysicist, right?
The thing is they don't need to. Most clubs that aren't called Manchester United, Bayern Munich, Real Madrid, or Juventus may not yet be aware, but today every sub-elite club, every player, and any football academy can enjoy access to the same data and most of the insights the giants have in store. They can also obtain them without hiring a legion of data scientists to operate and maintain solutions, buying expensive hardware, or using multiple complex systems.
Limited sources shouldn't drive sub-elite teams away from using analytics solutions
Current solutions offer a very partial answer for sub-elite teams' needs
Let’s review the pieces that are needed to provide actionable football analytics. Video, tracking data, and event data are the three main building blocks of football analytics. Every club would ideally want to have all three, and at the highest quality possible. Sub-elite teams would naturally also put an emphasis on obtaining these capabilities at affordable prices. Many vendors are operating in the football analysis ecosystem, but their solutions don’t provide a complete answer to this three-headed need. Let’s take a quick glance at the common solutions the market offers today:
Video solutions (i.e. recording of matches/training sessions) provide users only with match video and sometimes can also provide automatic highlight clips at the team level (not of a specific player).
Tagging solutions present match video to a user (usually a video analyst ) and allow the user to manually tag events and associate them with video frames.
Analysis solutions: provide insights of the match and season, but don’t produce the actual data themselves.
Tracking sensors (such as the well-known GPSs) are a great source for Physical data (acceleration, deceleration, sprints, speeds, etc.) but ֿ don’t provide data on the opponent team, nor the ball or video for a full tactical analysis.
Solutions that sub-elite teams use today are only partial
So how do teams obtain a complete solution? Today elite teams use multiple, specific systems to obtain all the data needed for a full analysis, and a crew of analysts crunches this data to make it useful. For teams with an Xlarge budget - that can work. But what about smaller teams that aren’t able to bear the high costs? There’s a combination of two new technologies out there now that is providing the answer.
Optical tracking meets Deep Learning to change the game
The most advanced tracking system as of today is called Optical Tracking. At its basis, it’s a technology that determines the position of objects at a rate of at least 25 frames per second. Combined with advanced deep-learning techniques, optical tracking can also identify specific objects (players), in a fully automated process. Football analysis wise - it offers the most accurate method of tracking every player on the pitch (from both teams), the ball movement, and is inherently coupled with the match video. Due to its uniqueness of characteristics, optical tracking offers stakeholders in football clubs the widest range possible of insights.
Having said that, optical tracking isn’t yet a technology that is widespread enough in the market, mainly because first-generation solutions require complex hardware installation and matchday operations. Nevertheless, advanced solutions tackle this challenge too and offer a simple, single installation point, with a fully automated process.
Now, Combine optical tracking with AI event analysis, and you get the next step in the football analytics revolution
Significant as it is, single viewpoint optical tracking would only take you so far. At the end of the day, even the most accurate positional data won’t be enough by itself to generate high-level analytics. That’s where event data comes into play. Events are basically every action that happens during a football match - corners, goals, cards, passes, throw-ins, shots on target, etc. Each of these actions needs to be identified and tagged before any data is derived from it. As of today, this effort - which takes many hours even if you don't tag every event on the pitch - is done manually. For that reason, the majority of teams - who cannot afford to hire a legion of analysts (or pay for event tagging while also using other costly solutions) don't have analytics at their service.
One unique solution that’s new in the market can be considered the holy grail of analytics solutions. It can also potentially democratize football analytics: Towards the end of 2020, Track160 launched "Coach160", an AI-based solution supplying teams with data, video, and events in one single SaaS platform. As of now, Track's product is the only one that combines single viewpoint optical tracking methods with the real prize: fully automated event tagging and out-of-the-box smart analytics. Sub-elite teams and academies who could not afford analytics costs and extra manual work will find all their coaching and performance needs taken care of, and at an affordable price.
Listen to CEO, Eyal Ben-Ari, speaking about Track160's fully AI automated solution