Big DataData Storage

Machine Learning Is The Future Of Sports Data

Sam Pudwell joined Silicon UK as a reporter in December 2016. As well as being the resident Cloud aficionado, he covers areas such as cyber security, government IT and sports technology, with the aim of going to as many events as possible.

Machine learning and AI are playing bigger roles in sports data collection, as Patrick Lucey, Director of Data Science at STATS explains

The role of data in professional sport has grown significantly in recent years, helping teams, coaches and athletes analyse opponents and make better decisions in real-time.

For example, the Mercedes F1 team is trialing the use of 802.11ad Wi-Fi in an effort to give race engineers access to more detailed data more rapidly and Manchester City Football Club is relying on data analysis to provide insight in areas such as recruitment, marketing and the team’s on-field performance.

In rugby, Accenture is currently applying its big data tech to the Six Nations tournament in a range of ways, including analysing player performances, making match predictions and providing visuals for broadcasters.

One of the biggest sports for data collection is of course football and Patrick Lucey, director of Data Science at sports data and technology company STATS, believes a new era of data collection is approaching, driven by machine learning and artificial intelligence (AI).

STATS

Data + context

To provide some background, STATS has over 35 years of sports data intelligence, providing solutions for player tracking, athlete monitoring and video analysis to teams in sports such as football, basketball and ice hockey.

Player tracking is all well and good, but Lucey believes the key now is adding context to this data: “Speed and distance is kind of the easy stuff. The thing we do really well is finding context. So it doesn’t matter that someone ran that far, what does that actually mean? How many sprints did they do? Did they do them when they were pressing?

“There is so much context in football that we’re able to pick up. A lot of those metrics don’t really matter if you don’t have the context.”

In line with this, a lot of the focus now  for STATS is on helping people make data-driven decisions. The key for a data company such as STATS is to “help a coach quickly find what they’re looking for” and get “immediate answers which they can rely upon”.

By adding context to the data being collected, coaches and analysts are able to spend more time focusing on “higher level strategic things” such as the correlations between actions, or the quality of chances created rather than just the number.

“People think of tracking data as the be all and end all but it’s just the starting point,” Lucey said. “If you look at reporting in football, statistics don’t tell the full story. Context plays an important role.” 

Machine learning

One of these ways STATS is providing this context is through new technologies that “maximise the value of the data that we have”, vitally important in a sport as tactically subtle and complex as professional football.

The company says it has invested heavily in machine learning and AI, as its growth in sport has mirrored that which is currently happening in other industries.

“Football is such a complex, strategic game, so how do we get [the data] in a form or a language or a context where coaches and analysts can understand it? Machine learning enable us to do that because it allows us to come up with objective measures, it allows us to detect things which currently can’t be done by human hand,” said Lucey. 

STATS

One such example is “ghosting”, a process where certain events or scenarios can be simulated to predict what will happen. Ghosting can enable a coach to analyse how a certain player or even a whole team reacts to a specific play, allowing the coaching staff to prepare their team accordingly.

“If you can see how variable a team is or how volatile they are to defensive structure, that’s all very useful,” said Lucey, describing it as “core artificial intelligence and machine learning”.

And not only do machine learning and AI provide more insight, they also enable analysts to do their jobs quicker and “spend more time on the pitch”, a key consideration in a sport where matches often come thick and fast.

“If you want to ask a strategic question, particularly in football, such as how often does this team press, where do they win the ball etc, an analyst would have to go through a lot of video. But using machine learning we can go through historical data and get that answer straight away.”

“It’s about enabling the coaches and analysts to do their jobs quicker,” Lucey concluded. Something they will only be too happy to hear.

 Are you a tech Olympian? Find out with our sporting IT quiz!