John W Henry made his millions by using hard data to play the futures markets and get the edge on those prepared to rely on instinct. When he bought the Florida Marlins in January 1999, he discovered that baseball was another industry prepared to base huge financial decisions on hunches. "Many people think they are smarter than others in baseball and that the game on the field is simply what they think it is, through their set of images and beliefs," Henry reflected. "Actual data from the market means more than individual perception. The same is [as] true in baseball [as the stock market.]"
But the seminal American book on the subject, Moneyball, demonstrates that the Marlins were a tougher nut to crack than Wall Street. The book, which maps the way the Oaklands A's were transformed by general manager Billy Beane's determination to instil the metrics system he developed from the ground-breaking work of Bill James, suggests that Henry struggled to get the Marlins coaches interested. "For a man who had never played professional baseball to impose upon... a major league baseball franchise an entirely new way of doing things meant alienating the baseball insiders he employed: the manager, the scouts, the players," author Michael Lewis wrote. "In the end he would be ostracised by the organisation. And what was the point of being in baseball if you weren't in baseball?"
The secret of Beane's success has not been so much using statistical data to help tactically and in the transfer market – the sport was already doing that – as using data that no one else was looking at. "What we tried to do was find value in areas where most people weren't necessarily applying the right values and that meant we could go for the players no one else was interested in," Beane told The Independent. "Every event that happens on any pitch has a value attached to it. Technology now allows us to gather any kind of information. The rubicon is applying it."
He is, at arm's length, attempting to do that in football. Just before joining a group of investors which took over the San Jose Earthquakes Major League Soccer franchise in 2007, Beane heard about a talk given by the Leeds Business School professor of sports management and finance, Dr Bill Gerrard. The subject was how metrics – or "statistical performance analysis" as Gerrard described it – might be translated from baseball to football. The two spent over two years attempting to do that for the Quakes.
Dr Gerrard says he produced over 80 reports for Beane. These included analyses of the common features of every game the Quakes won and loss. He also developed a player-rating system based on 30 actions each individual completed in a game, each of which was weighted according to its significance to the Quakes winning. Taken with the salaries for each player, published by the MLS players' union, Dr Gerrard was able to produce a value rating for every player. Beane was also interested in a statistical approach to the MLS draft system, under which each side is allowed to protect 11 players. "He wanted me to produce a roster, within the $2m salary cap, of players who had not been protected," said Dr Gerrard.
Beane says metrics enabled him to sign players, though he will not disclose names. However, the Earthquakes' owners, including Beane, had pledged not to force the ideas on coaching staff. After two years, coach Frank Yallop, the former long-serving Ipswich Town defender, wanted to pursue a different route.
These ideas are not alien to the British game. Bolton's innovative former performance director, Mike Forde, now at Chelsea, embraced them and asked Gerrard for a report after hearing him give a presentation. But Bolton ultimately went their own way without Gerrard, whose time working at Arsenal was also brief. "People see you as a threat and a meddler and can be very negative to you," Gerrard said. "At the football coaching end there is a great belief that you need to have played the game at the highest level to understand it."
Beane acknowledges that metrics forms only a part of the matrix of football success. He has not forgotten an encounter with Sir Alex Ferguson at a conference in the United States a few years back. "I'm physically a much bigger man than he is but there's something in him which makes you want to be a part of what he is doing. He doesn't waste words but when he says something it means something."
But Beane urges patience. "There is a misperception that with metrics you are not going to spend money but that's not true," he said. "Great players can cost a lot of money but become worth many times more than you pay. Metrics can help you see the potential in young players, who stay healthier and who you can pay less, leaving you more money to buy the great players." Beane believes Henry will make metrics work at Anfield because he is "one of the most innovative businessmen I have known". Recent history suggests powers of persuasion may also be required.
Metrics: How data can (and can't) help football
Defensive statistics are generally very useful. Minimising defensive errors is a key to success in all sports and work Gerrard carried out in rugby league has found these to be the best predictor of a win or loss. Missed tackles in the final third are particularly significant.
Successful entry into the final third
Since one goal is generally scored for every eight shots on goal, the completion rate of balls into the final third is a critical indicator when taken with the number of shots produced from those balls in.
Integral to the intelligent use of metrics is an understanding of how the team manager wants his team to play. "Tell me what your team would look like if they were playing as you want them to play," is what Gerrard says to managers he works with. If it is a team such as Stoke, who depend on aerial balls into the area, balls sent into the opponents' corner, perhaps winning a throw-in, are important.
Wages as a measure of a player's value for money
Gerrard's work for Beane at the San Jose Earthquakes included a player-by-player analysis based on 30 actions in a game which, when applied to the salaries of those players, enabled the side to calculate precisely which players represented good value for money.
And the data that does not help....
"Activity statistics", as they are known, are not generally helpful. Time in possession does not offer any prediction of how much a player is contributing, as it does not tell if a player is contributing to scoring and is not generally predictive of an outcome. A lot of teams that lose do have a lot of offensive play but are unable to do enough with it.
Can be equally unhelpful, since it is passes into the final third of the field which can affect the outcome of games. Arsenal's Cesc Fabregas tends to make a lot of misplaced passes, because he is trying to split defences. There is a risk-return calculation where he is concerned.Reuse content