Moneyball And The Rise Of Sports Analytics

In the late 1990s and early 2000s, the Oakland Athletics’ ability to remain competitive in Major League Baseball (MLB) despite its comparatively meagre resources, mystified many observers. The explanation turned out to be the birth of sports analytics as a major science.

In the 2002 season, the team won 103 games in the regular season, the same amount as the New York Yankees, whose wage bill was three times that of Oakland.

What commentators didn’t know was that Oakland General Manager Billy Beane was using sophisticated statistical models to analyse professional players’ performance, that allowed him to spot good players that others overlooked. Conventional baseball wisdom followed certain stats; Beane included others, and found recruits who were undervalued by the market.

The impact of sabermetrics

These methods were highlighted in Moneyball, a 2003 book by Michael Lewis that was later turned into a 2011 film starring Brad Pitt. The story is probably the first time that many people had heard of ‘sabermetrics’, the analysis of baseball through empirical evidence such as statistics.

The book documents the growth of sabermetrics from the ideas of Bill James, widely regarded as the father of the movement, to a multi-million dollar industry and their eventual application by Oakland in the 2002 MLB Draft.

The impact has been significant and most MLB teams now employ a sabermetrics analyst. Moneyball has also attracted the attention of other sports, keen to make the most of the data available to them to gain an advantage.

James proposed the use of analytics back in the 1970s; since then a vast amount of data has been created, and computing has only recently advanced to a stage where it can take full advantage of it. HP, IBM and Microsoft are among those trying to help sports organisations to use analystics, to improve performance or improve engagement with fans.

The influence of analytics is growing in many sports, but researchers have to overcome many challenges to get it established in other sports as firmly as it is to baseball.

Baseball’s suitability

Baseball is perhaps uniquely suited to the application of sports analytics. There has always been a plethora of statistical data available and each pitch can be classified as a distinct event. Every hit, in whatever direction, has probably been made thousands of times previously, meaning each play can be analysed against a historical database.

The problem Bill James found in the 1970s was that certain stats were not being measured correctly and fielding stats were virtually non-existent, meaning that certain figures were valued far more highly than they should be.

His Baseball Abstract reached out to a previously dormant community of baseball statisticians who wanted to analyse the sport more efficiently. In the succeeding decades, more data was collected, allowing analysts to more accurately predict the impact of a player’s performance on the team.

This approach redefined which stats were crucial to ascertaining a player’s worth and led to long-held assumptions being challenged, much to the suspicion of baseball teams and coaching staff. It is suggested by some key figures in Moneyball that these statistics were a better judge of effectiveness than a traditional scouting network.

Sports as a testing ground

“It’s an inevitable trend and it’s going to happen to every major sport,” said David Rothschild, an economist at Microsoft Research, who accurately predicted the outcome of this year’s Eurovision Song Contest.  “Sports are an awesome testing ground because they happen every night, there are tonnes of statistics on them and there are loads of people following them.”

He is convinced that the methods employed in sports can be adapted for use in any business, resulting in better decision making.

“I don’t view decisions that people make on a football pitch or basketball court as different to decisions made in any other walk of life,” adding that in effect, the Oakland athletics was an SMB using data to make more efficient decisions. “Efficiency is the goal of a lot of these companies.”

The biggest challenge for him is that many of the statistics that some sports rely on have only recently become available. Rothschild says this makes today’s opportunity very similar to the one which Bill James found in the 1970s.

Rothschild has worked with US sports broadcaster ESPN to help predict the outcome of the NBA play-offs and says one of the advantages of basketball is that there are just five players.

This means that cameras can capture positions of the players, resulting in an “absurd” amount of data, presenting another challenge for analysts.

F1 leading the way

“We have more data than we know what to do with and a lot of data won’t turn out that useful,” he said, although he said that it could be useful in answering future questions.

As the most technologically advanced sport in the world, Formula One accumulates huge amounts of data, which provides information about the car and helps teams both trackside and at headquarters make vital decisions on race strategy, tyre selection and pit stops.

One of Lotus F1’s cars has a complex network of more than 200 sensors which generate 25 MB of data every lap. In a sport which can be decided by one hundredth of a second, this data must be analysed efficiently if it is going to be advantageous.

Lotus F1 has agreed a deal with EMC to help make the most of this data, but Formula One was always likely to be ahead of the curve because using the latest technology in the sport is not only desirable, it is essential for success.

Strawberries and analytics

Tennis has also had a long-standing relationship with analytics. IBM has been a partner of all four grand slam tournaments for at least 20 years and its association with Wimbledon and the French Open extends even further.

These are predominantly technology and marketing relationships to ensure the running of a successful tournament and the operation of official websites, but IBM has accumulated a huge amount of data.

Matin Jouzdani, associate partner of sports analytics at IBM, told TechWeekEurope that tennis lends itself very well to such analysis because it is a series of discreet events. Every point can be related to an outcome, with turning points such as breaks of serve clearly identifiable, making it easy to model.

Its predictive analytics technology makes sense of more than 41 million grand slam points and identifies different patterns in play and style. This is presented to the public in the form of SlamTracker, which provides three ‘keys to the game’ that give each player the best chance of winning.

The primary aim of this is to encourage fan engagement and help broadcasters rather than assist players (the information is only made available once players are on court), but competitors are given a USB stick with video analysis categorised by elements such as backhand returns or second serves.

Touch, pause, analyse

Jouzdani is also convinced that analytics will have a greater role to play in other sports, both in terms of fan engagement but for teams and athletes.

“We are convinced that analytics can give a competitive edge in sport,” he declared, saying that rugby union and football were two it is targeting.

IBM adapted its SlamTracker technology for the England rugby team’s matches during this year’s Six Nations. TryTracker uses stats provided by Opta to show the ebb and flow of the match, identify influential players and provide three ‘keys to the match’

“We believe that the appeal of real-time analytics as a companion to the live event has broad appeal across many sports,” adds Jouzdani “We are seeing fans from all walks of life becoming far more literate with stats.”

The company has also agreed a partnership with Premiership Rugby club Leicester Tigers to use its analytics technology to understand and reduce player injury rates. The SPSS Modeller aggregates and processes thousands of data points that are collected during each game and can even help create individual player training regimes.

Rugby is also attracting the attention of Microsoft, which in its capacity as Official Technology Provider to the British and Irish Lions this summer, will monitor player performance, fitness and sleep patterns to be compiled and analysed by medical teams

The awkward football question

Rugby is a much more fluid sport than baseball or tennis, which provide a direct correlation between points and outcome. However, Jouzdani says that because many parts of rugby are stop-start, it can still be broken down into discreet events to be analysed, while there is a finite number of tactical approaches.

The same cannot be said about football, which makes IBM’s interest in the sport intriguing. It has created a demo application, which also uses Opta stats, but with football arguably the most difficult sport to apply big data principles to, it seems debatable whether it can have an obvious effect.

“Football is sometimes called ‘the beautiful game’ because it is unpredictable,” he said. “Moments of brilliance make it one of the most popular sports in the world.”

He concedes that there are some sports where analytics is “inherently less suited”, such as ones with small databases or data collection is difficult, but there is not one area where he doesn’t see a use for the technology.

He says there is more data about football than ever before and that this could be analysed more efficiently. Microsoft’s Rothschild agrees, saying that with the advent of goal line technology, this is set to increase even further. There must be some truth to this given the amount of money the FA has spent on analysis equipment at the new National Football Centre at St George’s Park.

However, unlike US sports organisations, there is not the same data culture among European sports leagues, which is why Manchester City’s decision to release the data from its 2011-12 season to help researchers, sports scientists and bloggers was so refreshing.

Perhaps recognising the difficulty in handling football data , the club hoped that third parties would find new ways to measure performance and value a player.

Not the Holy Grail

It is difficult to argue against the increasing influence of big data in sport, but there remain a number of challenges before analytics can impact other sports as strongly as it has baseball, if that is even possible.

Football highlights many of these problems, such as the availability of the data and the difficulty in creating meaningful conclusions. Some stasiticans maintain that their methods are better at judging baseball players than the naked eye, but analytics are not the Holy Grail for all sports.

Superior coaching, management and tactical nous are likely to remain just as, if not more important, and anything with a human element attached is going to retain a degree of unpredictability. After all, it is unpredictability that makes sport so entertaining.

“What we’re not trying to do is break football down into a mathematical formula that can be predicted because we know this is against the spirit of what makes football such a fantastic game,” explained Jouzdani. “Analytics is not is going to be a single silver bullet that revolutionises sports performance, but it’s going to be an important tool to create competitive advantage.”

Growing influence

But as more and more money pours into elite level sport, some teams and athletes will do anything they can to improve performance. The philosophy of Team Sky in professional cycling is built around the concept of marginal gains in performance which, when combined, amount to a significant edge.

Its huge budget allows for the latest advancements in sports science, logistics and equipment, so it will naturally turn towards analytics even if it can provide just a one percent improvement in performance.

This will not provide the answers for every sport, but it is clear that demand from sports organisations and fans is growing for this information, which could give them even the tiniest advantage.

Moneyball’s impact on popular culture was such that it was even satirised in an episode of The Simpsons in which Bill James himself made a guest appearance discussing his role in the development of sabermetrics.

“I made baseball as much fun as doing your taxes,” he said, tongue firmly in cheek. Let’s hope not.

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