What if the manager is asking the wrong questions?
A recent article by Alex Morgan for Football Everyday covered some of the football-related presentations at the recent Sports Analytics conference in London. The column featured some hitherto unknown titbits of the kind data leading club analysts have at their disposal, including a long and detailed look at the excellent work of Ian Graham at Liverpool FC. For that reason alone, it’s worth your time.
However, buried deep in the article was a passing mention of just how most clubs are employing this information:
Ultimately, Ian said his analysis wasn’t at all the deciding factor at Liverpool, only a “hook” to influence and back up claims with — a broad theme at the conference. Ian said he encourages the use of data, but one has to be careful with it. You can’t — yet — formulate a way to play football simply based on the data.
It should be said this observation is casually thrown in by the author and may not necessarily reflect the influence of Graham and others in similar roles. But this quote does get at the heart of one of the biggest challenges faced by data scientists with ambitions to test out assumptions about “ideal” ways to play the game—in the end, a club stats analyst may be nothing more than one voice among many in counsel with the all-powerful (and vulnerable) gaffer.
Perhaps this is as it should be—after all, when was the last time a data scientist won a European Cup? Ultimately the manager (or head coach) manages, and they need all the expert advice they can get. In theory, this sounds marvelous, like the footballing version of Doris Kearns Goodwin’s Abraham Lincoln memoir Team of Rivals, a book that emphasised the deeply varied and often mutually opposed figures the Civil War president assigned to his cabinet. And, depending on the manager, technical director, performance analysts, scouts and stats analyst, this kind of set-up may indeed be a successful approach.
What is perhaps less effective is when a data scientist is employed to be exclusively the Answerer of Manager Questions. “I want a fullback who is good at sending in early crosses and will cost me less than £10 million, can you help find me one?” “I’d like my team to defend better in the air on corners, what does the data say?”
This isn’t necessarily a bad thing per se; often crafting research projects based on the demands of coaches can push data scientists in directions they may not have previously considered. The problem is the case of a manager or technical director won’t engage in a dialogue about whether these are even the right questions to ask based on the team’s season long objectives.
Even worse is the scenario in which a club employs a brilliant-but-underpaid PhD level stats analyst who does nothing more than spin their wheels in an organization in which the players work almost entirely under first team coaches who think statistical analysis is all Grade A BS, and in which traditional scouts have complete final say in which players will be considered for the next transfer window. This is the equivalent of buying a Ferrari and sticking it in the garage.
A club shouldn’t hire a stats analyst as part of some paranoid league-wide arms race, or to stuff into a cubicle for the entire year to spin out interesting but competitively fruitless coach-directed assignments. They should do so in the understanding that, hopefully, they’ve acquired someone with deeply specialized knowledge who can offer far, far more than customized “advice” on a grabbag of different subjects near-and-dear to the manager. A stats analyst must be part of an on-going dialogue among equals at the decision-making level, and should be subject to the same level of accountability.
If clubs are unable to cede that kind of power to an outsider—and with the lack of public Moneyball-esque success stories in football for analysts to tout, this is understandable—there are alternatives. Why not let the reserve teams be a laboratory to test some of these ideas? Certainly the results will be skewed by the drop in competitive quality, but nevertheless, an analyst under less pressure to win individual matches could do some good work here. For all we know, some clubs may already be going down this avenue.
There will be success stories no doubt in the years to come, leading to more success stories. But history will remember the clubs that took the first major risks in how they used statistical science to influence decision-making.