Winning by numbers: how performance analysis is transforming sport

By Joao Medeiros

26 June 12                http://www.wired.co.uk/magazine/archive/2012/07/features/winning-by-numbers?page=all

View Gallery

Magazine Features

This article was taken from the July 2012 issue of Wired magazine. Be the first to read Wired's articles in print before they're posted online, and get your hands on loads of additional content by subscribing online.

Soon after becoming a professional squash player at age 18, Nick Matthew left his coach. He brought in David Pearson, then England's national squash coach, and performance analyst Stafford Murray. Matthew had been playing since the age of eight and had consistently ranked among the top junior players in the UK. He was a strong athlete -- he had been a cross-country runner for many years -- and typically won by wearing down his opponents. But as a senior player his physicality was no longer an advantage against much stronger opponents. He realised that he wasn't good enough.

"His technique was not the best," says Murray. "He was very strong, and mentally one of the best. But you talk about naturally gifted players and he certainly wasn't one." Murray is the head of performance analysis and biomechanics at the English Institute of Sport (EIS) in Manchester, overseeing 25 staff who work across all national sports teams. In his mid-thirties, he has the no-nonsense manner of a sports coach and the academic enthusiasm of a scholar. Murray was mentored by Mike Hughes, who pioneered performance analysis in the early 80s, using PCs to record and process data in real time during squash matches. "I tell my students -- Stafford is the best performance analyst in the country," says Hughes.

When Murray began working with Matthew, they would meet four times a week. Murray would bring cameras to the court and analyse Matthew's technique using Dartfish, a video-based software program. Murray saw that Matthew was mostly using his wrist, rather than his whole body, when taking a backhand shot; as a result, he couldn't generate much power. Using Dartfish, Murray would break down the movement in strobe motion. He would also compare it with how Matthew was performing that same move previously by overlaying a ghost image on the screen. Matthew would hit a dozen balls and then look at the monitor. His coach would feed him the ball slowly, as if he were a beginner. He would spend hours on court each day, just practising his backhand, over and over.

"I had been playing since I was a kid and now I was learning how to hit a ball from scratch," Matthew says. "I felt a lot of self-doubt. I kept thinking, 'Do I really need to go through this? I'm already a good player.' It was the hardest thing I've ever had to do." When he began training with Murray in 2001, Matthew had been playing squash for 12 years and was ranked among the top 100 players in the world. It took him two years to relearn how to hit a ball but, by 2004, he had climbed to the top ten. Six years later, he became the world's number one player.

In December 2010, Matthew played against his compatriot James Willstrop in the final of the Saudi PSA World Open Squash Championship. Matthew had already beaten the four-time world champion Amr Shabana in the semifinals, and won the final against Willstrop by three sets to one, winning the last two games 11-2 and 11-3. "Destroying someone with my squash game as opposed to tiring him out was unique for me," says Matthew. "That's when it all came together."

***

In elite sports, being the most talented is no longer enough; top athletes also have to ensure they are the better prepared.They understand that their only sustainable advantage is to learn and improve faster than their opponents. The technology used by performance analysts allows them to measure every force, dissect every movement and time every action with absolute precision. That feedback allows athletes to find areas for improvement and aids the learning of new skills.

In the general domain of skill expertise, objective feedback rarely follows action. Experimental physicists have to wait for years to obtain validation of their theories; doctors seldom get immediate confirmation that their diagnoses are accurate. Elite athletes, on the other hand, can get immediate and precise feedback for every movement that they make, right down to the tiniest manoeuvre. They have become used to using scientific knowledge and data feedback to optimise the way they train, making it more efficient and effective. Coaches call this "accelerated learning". Elite athletes haven't merely mastered their sport. They have mastered the art and science of learning.

At the EIS, Murray's team of performance analysts and biomechanists are working with Olympic athletes as they prepare for the London 2012 Olympic Games. From tae kwon do to sailing, from cycling to boxing, Murray's team of performance analysts collects data and analyses it. "The fundamental methods are generic. The guys working with the sailing team will have a look at what the tae kwon do analysts are doing and the guys in boxing will come down to study the methods of the squash analysts," Murray says.

At the Manchester Velodrome, BAE Systems has developed a timing system based on military technology that uses lasers and bar codes on athletes to give exact identification, split times and velocity data. Bikes have instrumented cranks that capture force measurements, velocity and acceleration. The data is logged in real time via a system developed by McLaren and performance analysts stream the video of the cyclists' workouts directly to the coaches' iPads. It has paid off: the British cycling team brought home 14 medals at the 2008 Beijing Olympics. Performance analysts also make extensive use of Dartfish, created at the Swiss Federal Institute of Technology in Lausanne, which allows them to record performance, annotate video clips, overlay extra footage and add time codes.

Murray likes to cite a particular statistic: coaches can recall only about 30 percent of what they see during a competition. In other words: 70 percent of potentially vital information goes unnoticed. That's where his expertise as a performance analyst lies -- in finding objective data that will make his athletes accelerate their learning and outperform the competition. "I worked with a squash player who wasn't hitting the ball at the right point during the swing," Murray says. In squash, he explains, players should take the ball on the top of the bounce when it comes off the floor. If you take it on top of that trajectory, you're not giving your opponent any time to get their breath back. "This player was hitting it just off the top of bounce," he says. "I calculated his timing and estimated that he was losing about two minutes throughout the match. He was basically giving his opponents two extra minutes. This is the sort of data that we have to find. If we're not changing their behaviour and accelerating the way they learn new skills, we're failing."

In their 1967 book, Human Performance, American psychologists Paul Fitts and Michael Posner put forward a model which posited that people go through three distinct stages when learning a new skill. They called the first stage the cognitive stage. This is where we intellectualise the task at hand and search for strategies to accomplish it. The conscious part of the brain is fully engaged. Take learning a language as an example: at the cognitive stage, you are trying to understand the basic rules of grammar, you are learning the first words and are attempting to pronounce the different phonetic sounds correctly.

After some experience, learners reach the second stage of learning, which the authors called the associative stage. At this point, the learner has developed knowledge of what, how and when to do something. He or she starts to practise and receive feedback. The conscious mind is no longer grappling with the unknown, but being competent still requires full conscious concentration. As the learner practises, the new skill slowly migrates to the unconscious level.

At the third stage, the autonomous stage, the learner stops being a learner. The skill becomes second nature and can be performed almost unconsciously. Psychologists say that the learner has reached a level of unconscious competence. Speakers reach this level of competence with their native language, whereas adult learners of a second language typically struggle to reach the autonomous stage, even when fully immersed in the foreign culture. Reaching autonomous competence isn't easy.

In sport, elite athletes possess more than mere unconscious competence: they have unconscious mastery. This level of skill often produces what is known as a "flow experience", a mental state at the extreme of the unconscious competence spectrum that typically results in unbeatable performances. In 1988, at the qualifying stage for the Monaco Grand Prix, the late Brazilian Formula 1 driver Ayrton Senna described a record-breaking lap thus: "I was already on pole, then by half a second and then one second and I just kept going. Suddenly I was nearly two seconds faster than anybody else, including my teammate with the same car. And suddenly I realised that I was no longer driving the car consciously. I was driving it by a kind of instinct, only I was in a different dimension. It was like I was in a tunnel. Not only the tunnel under the hotel, but the whole circuit was a tunnel. I was just going and going, more and more and more and more. I was way over the limit, but still able to find even more."

When we think of the prowess of Olympic athletes, of course, we tend to consider its physical manifestation. For instance, Cristiano Ronaldo, who is considered one of the best football players of the last decade, is what we might call the complete physical athlete. He has the long legs of a sprinter, the lean physique of a middle--distance runner. He can jump about 80cm in the air, higher than the average professional basketball player. Added to his physical attributes, what makes Ronaldo a great player is his supreme coordination and masterly skill with a football.

But how is the physical ability of athletes related to neurological capacity? One example is anticipation. Mark Williams, a professor of motor behaviour at Liverpool John Moores University, has looked at how elite tennis players, for example, react to the movements of their opponents. He placed tennis players in front of a large screen on which he projected a life-sized opponent performing a serve. He had them anticipate their opponent's action by responding physically to the film. He then occluded parts of the clip sequence -- for instance, making the screen go black just before the ball was struck -- to find out whether the tennis players could anticipate the opponent's serve.

"What you find in fast-paced sports, such as tennis, is that the time taken for the ball to travel from one opponent to the other is often shorter than the combined sum of the athlete's reaction time and movement time," Williams says. Athletes need to initiate a response before an opponent actually strikes the ball. Typically, they have already anticipated where the ball is going 120 milliseconds before the ball made contact with the racquet.

By using an eye-movement sensor, Williams found that the players don't even look at the ball, but predominantly at the trunk, hips, shoulders and the arms of their opponent. Unconsciously, they extract this visual information to anticipate accurately what's going to happen.

Another cognitive skill used by elite athletes in team sports is the ability to recognise patterns during a match. Williams devised a study in which footballers of varied proficiency were shown a film simulation of an attack. "Say you have a centre-half on the edge of the penalty area and the ball is on the halfway line. What we find is that this player will be using lots of fixations of short duration, scanning, picking up position and movements of players," Williams said. "He's analysing the structure of the game. Subconsciously, he's doing maths. He's calculating event probabilities for each given situation at every single moment, based on his experience of the game. It's as though he has a database in his head, with a probability assigned for every play pattern that he might encounter in the game."

The factor that sets elite athletes apart, then, is not just what they can do but knowing precisely when to do it. In a La Liga game against Rayo Vallecano last February, Real Madrid's Ronaldo chased a ball that had bounced off a defender on a straight line and, with his back to the goal -- which was ten metres away -- backheeled the ball past four defenders and the goalkeeper and into the net.

"He didn't even need to check where the goal was," says Zoe Wilmhurst, a visual-skills sports expert who works with Olympic athletes. "Ronaldo's spatial awareness is off the charts. Players like him are like scholars. They simply tap their memory banks, with their thousands of different permutations of the game, and based on the data they retrieve, they are able to make decisions with very little information and without even thinking."

Even after he'd started working with Stafford Murray, Nick Matthew would go into tournaments thinking about his technique. "Mentally, it was a lot of baggage to take into the game," Matthew says. "You get analysis paralysis. I would feel really happy with my progress during a session and then go into a game, and as soon as I was put under a bit of pressure, my technique would just fall to pieces."

This is not unconscious competence. Matthew was still aware of how his body was moving, of when he was releasing his wrist. He was still at the cognitive stage. "People appreciate how hard you have to work physically, but they don't understand the attention that goes into detail from a scientific point of view," Matthew says. "The endless hours to change one percent of the angle of your swing. That might make the difference between playing a drop shot into the nick and just missing it."

What Matthew is describing is not a failure of willpower or of talent. Rather, it is one of the key components of accelerated learning. Unconscious competence doesn't just happen. You reach it by forcing yourself to remain in the cognitive stage in training. This means that you have to practise the skills in which you are not yet competent, making your practice mentally effortful. To use the sports colloquialism, you have to "overload" your training.

One of the ways to do this is to do things with more intensity. Take Sarah Storey, a British Paralympian cyclist, who won two gold medals in Beijing. "When I practise on the road, I can fly at 80kph, get round the corners at more than 50kph, overtake cars," Storey says. "But when I started I was like, 'No way I'm overtaking a car'." She had to engage in what she calls "neurological training". Every time she went on the track at the Manchester Velodrome, she would force herself to cycle as fast as she could and she now often trains at velocities above world-record pace. It took her a year before she was overtaking cars frequently. "It's no use if you're going at 60kph round a 250-metre track and you get dizzy and your front wheel is wobbling because it's too fast to handle," she says. "Your nervous system has to be able to handle the bike at speed and to send messages to the brain so that the legs turn faster."

In more complex sports the challenge is not just to reinforce existing skills but to create different ones. In a study, University of Alberta sports scientists Janice Deakin and Jean Côté found that highly skilled figure skaters spent more time practising the moves that they couldn't do so well, trying different solutions, and failing (and falling) much more often, whereas the novices preferred easier practices. At the University of Liverpool, Mark Williams has found similar results. "The best athletes always practise at the edge of success and failure," he says. Williams compares it to a Darwinist process, because expertise arises from adaptations to constraints. "Athletes have to find solutions to problems they will encounter in competition," he says. "They fail, repeatedly, and it's by trying that they will eventually solve the problem and register the solution. The result is adaptation."

Nick Matthew describes an exercise that he uses to improve his positional awareness in the court: the traffic light system. "Stafford installs two cameras in the court," he says. "One behind the court and focused on the front wall, which is red at the top, amber in the middle and green at the bottom. A second overhead camera looks down on to the floor, where there are red areas in the corners, green around the middle -- a big circle -- and then amber in the other areas." Players who usually control the green areas win the match, so he spends whole sessions trying to play in green. "You change your whole movement to get that change -- suddenly you use the full dimension of the court," he says.

Murray invited Williams to give a series of talks on the theory of skill acquisition to his team of performance analysts in October 2010. In late 2009 the British Olympic archery team recruited a new coach, an American called Lloyd Brown. When Brown later assessed his new team he noticed that some of the athletes were using an inefficient technique when drawing the bow. With three months left for competition, Brown was keen to apply Williams's theory to accelerate the learning of a better technique.

Sports scientists make an emphatic distinction between two types of feedback. Intrinsic feedback comes from the athlete itself and varies in degrees -- say, the awareness of the archer that he or she has just missed the target -- to a more attuned perception of movement, such as how much force he or she has produced in the bow. Extrinsic feedback on the other hand, originates from outside sources, from the coach shouting instructions on the sidelines to the post-match video analysis. In a sense, extrinsic feedback is what allows the athlete to compare how he or she has really performed with how he or she thinks they have performed.

"It's fundamental that athletes get feedback to learn, but what's the best method for coaches to provide feedback and promote accelerated learning?" Williams asks. Sports coaches usually assume that the best way to run a practice session is to provide lots of instructions and hands-on demonstrations. This, however, is not the best way to accelerate learning. "Athletes become dependent on that feedback," Williams says. "When it comes to the competition, they aren't able to replicate that skill. Athletes need to create
self--sufficient mechanisms of feedback."

A study by Nicola Hodges and Ian Franks, professors at the University of British Columbia School of Kinesiology, for instance, showed that novice football players would learn faster by following general verbal instructions, such as "Can you pass the ball into the near-post area?" rather than being taught specifically how this could be achieved. Coaches, then, need to be able to provide the least amount of extrinsic feedback required to progress and let the athletes engage in self-discovery, in trial and error. This fine balance between extrinsic and intrinsic feedback helps athletes calibrate their own perception of their performance, which accelerates their learning. Novices need more feedback. As athletes become more skilled and experienced, the feedback they require is more detailed and less frequent.

After the workshops with Mark Williams, Brown and EIS performance analyst Oliver Logan designed a programme to teach their archers to relearn how to draw the bow. They first asked the archers to try the new technique in the easiest possible way with an elastic band, rather than a bow. They would have cameras overhead and use live video feedback and practise while watching themselves from above. Later, they delayed the feedback by ten seconds, allowing the archers to cross-reference what they thought they did with the visual feedback from the video.

Sometimes, unless the trial had been particularly bad, they wouldn't get any feedback at all. After a couple of months, the archers were already practising with targets, with background noise to mimic real crowds and in direct competition with other archers. Brown would call their name out and commentate live. They would also give the archers small financial incentives depending on their score, or force them to drink a pint of water so that they would shoot while needing urgently to urinate.

"It's important to make training as replicable to match play as possible," Stafford Murray says. In six months, the archers had learned not only how to draw the bow differently, they had improved their scores by an average of ten points. Through accelerated learning they had reached unconscious competence. Of that group, three archers are expected to represent Great Britain in London 2012.

"In squash practice, sometimes we project a match against the front wall and have the player play a virtual match, so it's much like shadow boxing," Murray says. Players would be playing against the world number one, ghosting the movements in the match -- exactly the same kind of movements that they would be doing were they playing in the real match. Physically, they would cover the same distance. Their heart-rate would be almost the same. "Performance analysis shows you exactly where the demands for competition are," Murray says. "We overload the training based on that data. There's no guesswork. You can't argue with the data."

We commonly assume that athletic skill is a random combination of innate talent and mere accumulation of years of experience. Why? Because this analysis simplifies our understanding of skill. It allows us to think of skill of something more akin to a mystery than as an art or science that can be dissected and understood.

But that mindset wouldn't have helped the British archers or elite athletes like Nick Matthew. Instead, they methodically diagnosed and addressed their shortcomings using data and feedback. That's what accelerated learning is: a smarter approach to learning a skill. It provides a scientific way to look into skill, decompose it into its elementary components and address them.

Insofar as there are general principles that underpin expertise, it is potentially possible to take accelerated learning out of the sports arena and apply it to our skills in order to become better at whatever it is we do. With enough motivation, we can adopt the principles of accelerated learning and do what the great athletes do: shelve our perfect skills and employ a scientific approach to correct our imperfections.

In Beijing in 2008, Great Britain finished fourth, with 47 medals, their second-highest haul in Olympic history. One of the 19 golds was won by Rebecca Romero, who had recently switched sports from rowing to cycling. She had rowed since the age of 17 and had won silver at the Athens Olympics in 2004 in the quadruple sculls, and gold in the World Championships the following year. In January 2006, she retired after a back injury, and decided to try cycling. The British cycling team tested her on a bike in a lab to measure her power output. "They said I had one of their best results ever," Romero says. "It was insane. I didn't have the physiology or the bike skills. Very quickly I was being taught everything and I was going from a nothing to being a member of the team and aiming for an Olympic medal."

Romero and Dan Hunt, her coach, devised a training plan. They determined what would be world-class times and what power output and speed she needed to break those times. Every detail was part of the equation: size of the gear, drag, humidity and temperature of the velodrome, body shape, body position. Romero would take notes of all her training sessions, recording what she done, her thoughts and where there was room from improvement. "I call it data ammunition," she says. "When I was preparing, we tailored all the training so that everything I was doing would lead me to be at the right point in two-and-a-half years. We couldn't afford unnecessary training."

In 2008, she qualified for Beijing -- and won gold in the individual pursuit category final. Romero had been training for less than two years. "People talk about the ten years it takes to achieve something, but I believe you can accelerate the learning process if you are smart about the way you practise," says Romero. "It isn't just about the hours and hours of training, it's all the tiny things that add up."

João Medeiros wrote about medical microchips in 11.11

Next: How to accelerate your learning...

How to accelerate your learning

Overload your training and push yourself
If you want to get bigger muscles, you overload your training with heavier weights. To some degree, the brain is like a muscle. You can overload the brain by increasing the intensity of your training (by doing things faster, for instance) or by practising skills you're not so good at rather then the skills you've already mastered.

Get objective and precise feedback
Without feedback, there's no change in performance. External feedback from personal observation is usually biased. The more objective the feedback, the greater effect it has. As athletes improve, they need less, but more detailed, feedback. They also become better at evaluating their own performance.

Mimic the pressure of competition
To gain the greatest benefits in terms of improved performance, practice sessions should match the demands of competition. Try to re-create the randomness, variability and emotional pressure of contest. The closest to the real thing you can get, the more effective the preparation.