Lately there’s been a lot written about the importance of analytics at Liverpool Football Club. The negative articles claim the analytics department there was responsible for flops like Aspas and Suso. The positive ones give the analysts credit for Coutinho, Sturridge, and Can. And pretty much all of them say there was tension between the club’s former manager, Brendan Rodgers, and the numbers folks.
Believe it or not, this isn’t how analytics works at every club around the world, or even in the Premier League. Usually I recommend analytics as a first cut for recruiting, to narrow down the pool of hundreds of potential signings to a few dozen for more intense scouting. This process saves the manager and scouts a slew of tedious hours watching matches on video and in person, allowing them to focus on the most promising prospects. It also preserves the importance of their judgment. Analytics works best when it serves the needs of decision-makers, including the manager.
Moreover, analytics can help with much more than recruiting. We don’t hear anything about how (or even if) analytics was used to enhance Liverpool’s tactics or its understanding of the strengths and weaknesses of opposing teams. Nor do we know if analytics have been used to forecast Liverpool’s finances, especially with regard to the relationship between revenue and performance on the pitch.
After all, why would we? Clubs that find an edge in any area using analytics have powerful reasons not to disclose what they’re doing. It’s only when some sort of acrimonious rift breaks out that the foggy rumors begin to circulate.
This is why “analytics” has become a sort of monolithic and uniform entity in the minds of the media and much of the public. Few people know how different analytics can be from club to club, so it ends up being thought of as the same thing everywhere. Either the club uses “analytics” — and we all think we know what that means — or it doesn’t.
Yet there is far more diversity in the practice of analytics than this perception implies. As I’ve written before, some so-called analytics purveyed by apparently successful firms is indeed devoid of coherent modeling and robust statistical methods. Worse, much of what passes for “analytics” isn’t even analytics at all. For example, saying that a center back has more tackles than any other player in Europe isn’t analytics. That’s just counting — and that’s what most scouting software does. Taking it a step further by saying he’s the best defender in Europe is just a subjective judgment; it has no basis in a model of results on the field. Real analytics, by contrast, is based on theory, rigor, and robustness, delivered through a process of calibration, testing, and refinement.
Is this how analytics is done at Liverpool? It may well be, but we still don’t know. In fact, it’s just possible that the analytics team at Liverpool simply isn’t very good.
Are they regularly dipping into labor market for analysts to make sure they have the best people? Are they monitoring all the new work being done in analytics to ensure they’re ahead of the curve? Are they challenging their models — if they have any — every day to guarantee that they’re capturing all the changes in the complex system known as European football? Or are they just counting tackles?
Of course, it’s better for me if they really are just counting tackles. And if other clubs reject analytics because of the rumors coming out of Liverpool, better still. When some clubs don’t use analytics — or when they use analytics poorly — it means the edge for other clubs will be that much bigger and last that much longer. It’ll be easier for me to find undervalued players and create tactical innovations for NYA’s clients. It’ll also be harder for new analysts to enter the market and compete with NYA. So thanks, Liverpool. You may be giving analytics a bad name from the public’s point of view, but you sure are doing me a favor.