If you’ve ever watched a grand slam tennis tournament, you’re probably familiar with the term ‘unforced errors’. But did you know that how they’re determined is still a bit subjective?
In a climate where ‘data analytics have become de rigueur in major sports’, tennis has been slow to embrace the trend. But now, a world-first tennis hackathon is set to change that. Learn more about the initiative, and the breakthrough Australian research that paved the way.
Giving tennis a better data ranking
In the wake of big name tennis stars dropping out of the 2018 Australian Open, big data could instead be taking over centre court. Tennis Australia’s Game Insight Group (GIG) has kicked off a Silicon Valley-inspired tennis hackathon – From AO to AI – to take the guesswork out of forced and unforced error analysis.
According to Tennis Australia’s Head of Innovation, Dr Machar Reid, automating the call of forced and unforced errors in professional tennis is vital. Without it, tennis lags behind other international sports in data and analytics.
For more than 40 years, tennis matches have been described using the terms first serve percentage, second serve percentage, unforced errors, forced errors, yet we are still unable to consistently define what some of them mean,” says Reid.
The hackathon draws on the biggest-ever single release of tennis tracking data by GIG and offers a total prize pool of USD$8,500. According to the contest’s partner, CrowdANALYTIX, a successful solution has the potential to radically transform the way tennis uses data science to collect match statistics.
Clearly, tennis is on the threshold of a statistical revolution. So how did it get to where it is today?
Pioneering Australian research
In 2016, Dr Reid and La Trobe’s Associate Professor Stuart Morgan (working at the Australian Institute of Sport at the time), along with several other industry experts, published their internationally award-winning research.
The aim of the investigation was to deep dive into three years’ worth of Australian Open HawkEye data – that’s 37,727 shots – to identify the crucial shots and movement patterns of the world’s best players.
Once identified, the team could then use their findings to accurately predict the probability of a player winning at any moment, taking into account individual player style and match context.
“Earlier work has often focused on the characteristics of ‘winners’, but we recognise the winning shot is often the least strategically-important shot in the rally. Rather it is the strategy over the preceding strokes to move an opponent out of contention in a rally that matters.
“Understanding how points are won requires a deeper understanding of the way players manipulate their opponents to establish dominance,” they wrote in their report.
The results ‘would give tennis fans, broadcasters and players an insight into what works and what doesn’t in each matchup,’ ESPN wrote of their research, which was awarded the Grand Prize at the world-leading 2016 MIT Sloan Sport Analytics Conference in Boston.
Dr Reid told Australian Open that the report and its recognition ‘showed we could compete with the world’s most popular sports in the analytics space’.
Beyond the numbers
Deep-diving into sophisticated analytics doesn’t just provide ‘more numbers’ – it also offers greater insight and context.
Dr Reid told The New York Times: ‘We intuitively all know Rafa Nadal’s forehand is one of the best in the sport, but why is that?’
“A level one stat might be that Rafa hits that ball at 140 k per hour. But now if we’re able to layer in or interpret that speed alongside his precision or weight of shot or capacity to hit winners or — on the flip side — produce errors, if we were able to amalgamate that and identify a shot scorecard that allowed us to rate a player’s forehand or stroke performance, that for me is a level two stat, and far more interesting.”
When Nick Kyrgios unexpectedly lost in the second-round of the 2017 Australian Open, commentators and former champions Jim Courier and Llyeton Hewitt were left speechless, the Australian Financial Review reported.
Reid, however, could look at the data for answers.
“In terms of quantifying it, with Nick in sets three and four, what you saw was a 30-40 per cent reduction in his work rate across those two sets. That is, the average intensity for each shot. In the fifth set the work rate elevated again [although it was good later],” he told the AFR.
“We saw Kyrgios’ intensity and movement dropped off on each shot, whereas Seppi sustained his intensity and movement. He did not lower his level, he maintained it throughout the course of the match.”
Future jobs in the elite sports industry
For sport science graduates entering the elite sports industry, Associate Professor Morgan says those with data analytics skills will have a huge advantage.
He says data analysis is ‘increasingly important’ in sport.
“The whole domain of sports science has changed over the last 10-15 years and the number of technical devices for collecting data has increased. The important thing about sports science now is the capacity to be able to handle and work with those kinds of data.
If you’re a graduate with data science and sport science qualifications, you’re going to be very, very competitive in the elite sport job market,” he says.
The sophisticated analytics opportunities offered by elite sport are not only invaluable for players and coaches, but also for the careers of talented data scientists, analysts and machine learners.
If you’re keen to apply intelligent systems to elite sport, hit a career ace with a Master of Sports Analytics at La Trobe.