How Arsenal and Arsène Wenger bought into analytics

Article tiré du journal The New York Times.

Arsène Wenger and Arsenal

Neither Marouane Chamakh nor Park Chu-Young occupies a particularly prominent place in Arsenal’s history. The former, a dilettante Moroccan striker, spent three years at the club after joining in 2010. He scored only 14 goals, was packed off on loan to West Ham United, and then released to Crystal Palace.

Park, a South Korean forward who arrived in London a year later, fared even worse. He had joined on transfer deadline day 2011, a surprise, last-minute capture from Monaco. But he never quite lived up to the drama of his arrival: In two years, Park played just seven times, and scored only once. He, too, went out on loan, and was then cut loose.

Both players left Arsenal unwept and unsung; if fans recall their names at all, it is only as they reel through the list of their club’s missteps in the transfer market, that cathartic process of reciting regrets at all those players signed and all that money spent for barely any reward at all.

Yet for all that Chamakh and Park failed to do at Arsenal, their effect since their departure has been considerably more lasting. In one light, in fact, it is possible to see these two most forgettable players from Arsenal’s past as leading characters in the club’s attempts to chart its future. More significant, they played a key role in teaching Arsène Wenger, that oldest of managerial dogs, the most cutting edge of tricks.

Toward the end of Park’s first season at Arsenal, the 2011-12 campaign, two of the club’s executives approached Wenger, the team’s seemingly immovable manager, with a proposal.

Throughout that year, Arsenal had engaged the services of StatDNA, a sports analytics company based in Chicago. The arrangement was based on exclusivity; Arsenal had paid around $250,000 to ensure that StatDNA would not provide insights for any of its rivals.

There had been some skepticism among Wenger’s back-room staff about the value of that deal. Now Ivan Gazidis, Arsenal’s chief executive, and its head of business development, Hendrik Almstadt, wanted to do something even more radical.

With the use of and belief in analytics growing at many Premier League clubs, a number of Arsenal’s peers had approached StatDNA, looking to employ its services. Gazidis and Almstadt were determined not to let that happen. To thwart their rivals, they proposed buying StatDNA outright, for a fee around $4 million.

Wenger had been open to the idea of using data from the start, but to persuade him to go further, Almstadt and Gazidis prepared a presentation on the benefits of bringing StatDNA, its founder Jaeson Rosenfeld and his team in-house.

Rather than making sweeping promises, they focused instead on what analytics might have helped Arsenal avoid. Almstadt, the driving force behind the plan, picked out Chamakh and Park as high-risk signings that a more empirical approach to recruitment would have averted.

The presentation won over Wenger. The deal went through.

For several years, Europe’s elite clubs have viewed analytics as soccer’s next frontier. Something of an arms race has developed to see who can find a successful formula first. Most major teams employ a team of analysts; in an increasing number of cases, their influence is growing exponentially.

They hold sway, particularly, in recruitment, something of a legacy of “Moneyball,” the best-seller by Michael Lewis that was interpreted in soccer as a guide to how to crack the transfer market.

In November, Liverpool promoted its onetime head of analytics, Michael Edwards, to sporting director. Swansea City last year brought Daniel Altman, the founder of North Yard Analytics, to the club as a transfer consultant. Data also is central to much of the work done at Manchester City, as well as its cadre of sister teams across the world, including New York City F.C.

StatDNA provides a similar service to Arsenal: It was a data-led approach that led to the signing of defender Gabriel Paulista in 2015, and that encouraged Wenger to try to land Gonzalo Higuain before his move to Napoli, despite the reservations of the club’s scouting department.

Kevin de Bruyne, now of Manchester City, was also flagged as a potential signing, only to be discounted because of doubts (incorrect, it turns out) about his ability to cope amid the tumult of the Premier League. The approach is not flawless: When Wenger mentioned a talented young wing at Spain’s Real Sociedad, he was told that his metrics were not overly impressive. Wenger smiled, and remarked that he would be keen to see how the player, Antoine Griezmann, now one of the most coveted strikers in Europe, had developed.

StatDNA’s work with Arsenal, however, runs much deeper than advice on transfers. The majority of clubs jealously guard the specific data gathered and methods used by their analysts — StatDNA was invited to contribute to this article, but declined — for fear of eroding whatever advantage they have accrued. Those familiar with Arsenal’s approach, though, believe it is among the most advanced in the field.

In part, that is because the data StatDNA produces is tailored for Arsenal. Many clubs still rely on external companies, such as Opta, to provide the raw figures from which their own teams of analysts work.

Arsenal, by contrast, has developed not just different metrics, but more thorough ones. Where it takes a commercial provider a couple of hours to code a single match, StatDNA requires around 14. The data Arsenal works from, in other words, is far cleaner.

It is also more in depth. StatDNA focuses not only on individual offensive output, but on defensive metrics, too, factors that are difficult to code. One measure at Arsenal, for example, assesses how frequently defenders make errors: failing to spot an opponent running past them, or losing a one-on-one duel.

Taken in conjunction with more traditional metrics like Expected Goals, which is the value of the danger of conceding in any given situation, Arsenal has a way of establishing not only how many errors a player makes, but the seriousness of them.

There is a focus, too, on trying to quantify the value of specific partnerships, or certain combinations, on the field, something that remains faintly quixotic elsewhere. In conjunction with Shad Forsyth, the American fitness expert recommended by Almstadt and appointed by Wenger, there are attempts to use the physical data the club gathers to help in injury prevention as well. The detail here is remarkable, too: gauging a player’s tiredness by measuring how long his foot is planted on the ground as he runs.

Crucial to all of it, however, is that it has Wenger’s full support. Rosenfeld is regarded as one of Wenger’s most trusted advisers, and what Wenger has described as a “core” of StatDNA staff members are regularly on hand at Arsenal’s training base. Wenger sees it as his job to pick the “four or five” key pieces of information that he requires for each game, but he values highly the many hundreds that come his way each week, knowing that at some point, any one of them could be useful.

Wenger once was regarded as a pioneer. His arrival in England two decades ago kick-started a revolution in nutrition, in conditioning, in tactics; he was one of the great modernizing forces in the Premier League. Recently, he has come to be regarded — not least by a substantial portion of Arsenal’s fans — as something of an anachronism, a man who has lost his edge.

His willingness to embrace the new, though, has never left him. He has continued to try to find the future, thanks in no small part to two forgotten players from his past.


Rory Smith