In 2012, Arsenal sold the pair of Robin van Persie and Alex Song to Manchester United and Barcelona, respectively. The Gunners invested most of their resulting budget on three replacements—Santi Cazorla, Lukas Podolski and Olivier Giroud.
However, not all the money was used directly on the playing squad. Just as in 1999, when half the fee received for Nicolas Anelka was ploughed in to redeveloping the London Colney training ground, Arsenal were able to take a longer-term view.
In December 2012, Arsenal spent £2.165 million to acquire StatDNA: a tool to help them discover the Songs and Van Persies of the future.
It is all part of what could be called Arsenal’s "Moneyball" philosophy. Moneyball, of course, is the idea spawned from Michael Lewis’ 2003 book of the same title. The tome charted the statistics-based work of Billy Beane with the Oakland Athletics baseball team, and the word has since gone on to become a catch-all term for his progressive philosophy.
The central theme is that the collective wisdom of contemporaneous baseball coaches and scouts was inherently flawed, subject to prejudice and subjectivity. The statistics used in analysis were also decried as too simplistic.
Instead, Beane worked with mathematicians to develop new measures and analytics to produce a more accurate assessment of a player’s performance.
In his own way, Arsene Wenger was doing Moneyball before it had a name. Song and Van Persie are good examples.
Conventional wisdom told him that Van Persie was an erratic winger with an attitude problem. Wenger saw a ruthless goalscorer. Song, meanwhile, was a gangly teenager who seemed to be lacking in technique. By the time he left London, he was deemed fit for Barcelona—the most technically adept club side in the world.
Like Beane, Wenger looked beyond the commonplace observations. He saw potential.
Beane himself is a confirmed fan of Wenger. The two are kindred spirits. Speaking to Jeremy Wilson of The Telegraph in 2011, Arsenal owner Stan Kroenke reported:
Billy Beane’s idol is Arsène Wenger. You know why? His ability to spend money and extract value. That is what it is all about to be successful in pro sports. If you can do that better than other people, you are always going to be pretty good.
And Beane himself has subsequently confirmed to the same newspaper:
I saw what Stan Kroenke said and it is true I have always followed Wenger. There is a desire to keep the team young and then keep it as competitive as it has been. Add to that the fact that it is a pretty well run business. Clubs can go out of business if they are not run properly so having someone who balances the business with player acquisition is a good thing in the long run. That is why I admire Wenger.
Wenger was once well ahead of his competitors. His eagle eye was enough to outwit most of the competition. Crucially, he knew what to look for: As in the case of Song, it wasn’t always the obvious contributions that defined a player’s quality. However, the game has progressed, and scouting has caught up with him.
Wenger has always been an advocate of statistics. The story goes that he signed Mathieu Flamini in part because of the startling fact that he once ran 14 kilometres in a single game.
However, that kind of knowledge is now relatively prosaic. All 20 of the Premier League clubs, and many in the lower divisions, employ data analysts to help them churn through the relevant numbers. Software like ProScout7 has become ubiquitous, allowing scouts to build a database of video footage annotated by statistics and observations.
In the analytics arms race, Arsenal needed an edge. That’s why they bought StatDNA.
We don’t know a huge amount about it. Arsenal are notoriously private about this area of their business. Indeed, chief executive Ivan Gazidis refuses even to mention it by name, instead referring to the firm as "AOH-USA, LLC"—the registry name used for the limited liability company in the U.S.
Asked about the affiliation at the club’s most recent AGM, Gazidis did admit (h/t Squawka):
The company is an expert in the field of sports data performance analysis, which is a rapidly developing area and one that I, and others, believe will be critical to Arsenal’s competitive position. The insights produced by the company are widely used across our football operations – in scouting and talent identification, in game preparation, in post-match analysis and in gaining tactical insights.
StatDNA was formed in 2009 and is currently led by CEO Jaeson Rosenfeld. Although registered and run from the U.S., a large proportion of the workforce is, in fact, in Asia.
The work is based around detailed video analysis. StatDNA’s own website discloses:
Each game we analyze takes between 10 and 20 hours to analyze. We not only capture every action that occurs in the game, but upgrades to advanced statistics like level of defensive pressure on each touch are available.
Many of our competitors analyze videos in real-time or faster than real-time. Because we want to provide the most accurate and detailed statistics on the market, our analysis is done over a period of time that can be 5-10x the actual length of the game. As part of this process we review our analysis with a QA process that includes statistical quality control.
The emphasis here is on depth. For those familiar with the work of Sarah Rudd, that will come as no great surprise.
Rudd is one of the few StatDNA employees who has had any kind of public profile—as a regular writer for the OnFooty blog, she earned a reputation as one of the most insightful statisticians in world football. A computer science graduate from Columbia University, she is an expert in statistics, machine learning and economics.
What's more, she's an Arsenal fan—and, as of January 2012, vice president of software and analytics at StatDNA.
This OnFooty breakdown of a match between Real Salt Lake and the Seattle Sounders in October 2011 gives some insight as to Rudd’s way of working. In the piece, she admitted that she hated passing percentage. That’s pure Moneyball: eschewing a traditionally fundamental statistic in favour of deeper analysis.
Rudd is named as the agent on the official documentation registering AOH-USA, LLC. If she was producing such sophisticated analysis in 2011, one wonders where her experiments might be taking her now. Although it’s impossible to guess as to the progress StatDNA are making, one suspects they’ll be forging into the area of regression analysis—a statistical progress that can be used to forecast future outcomes.
According to Marcus du Sautoy, professor of mathematics at Oxford University, this is entirely possible. He told The Guardian:
Football is much more of a game of chess than people realise. It isn't random what each team does from one game to the next. There are patterns. And the strength of mathematics is to change an activity into numbers and to spot patterns and predict things into the future.
Being able to make informed predictions about future performance would be hugely important to Arsenal’s tactical preparation. However, its most exciting function surely comes on the talent-spotting side. Currently, statistics only enable us to look back at a player’s past performance.
The next step is for them to be able to tell us how a youngster is due to develop or how an experienced player is set to wane. Those findings would have a huge influence upon the club's transfer policy.
That doesn't mean the budget will be spent by the stats guys. As long as Arsene Wenger remains at Arsenal, his say will be final. Take the aforementioned case of Flamini—Wenger still made sure to scout him in person before approving a deal.
However, perhaps there is a sense in which Arsenal are preparing for Wenger's inevitable departure here. His foresight and instincts are so valuable that an entire company has been purchased to attempt to fill that gap.
Billy Beane has been asked several times about the prospect of making his Moneyball system work in football. On each occasion, he has suggested it would be difficult: Fundamentally, football is a more fluid game than baseball. There are more moments of interdependency, which makes the game difficult to break down and analyse.
However, that won’t stop people from spending millions trying. Unlocking football’s statistical secrets is an expensive alchemic experiment. The rewards are potentially enormous. If anyone is primed to crack it, it’s Arsenal and StatDNA.