First, an apology: another meta post. I think this is important as the ideas here have been stewing in my brain for over a year a now, but it also gives me time to keep reading. Future posts will include a more expansive discussion of current work, which will more closely mirror what I hope will be included in daily subscriber emails. Thanks!
This week (and seemingly every week), the issue of the importance and value of communication in football analytics came up again, notably in a recent article written by Gabriele Marcotti in the Times. In it, he talks about the work pursued by Chris Anderson, co-author of the popular football stats book The Numbers Game and vice-chairman of Coventry City.
Marcotti writes that analytics won’t achieve a real breakthrough in the football universe until managers and coaches get a better education in mathematics and logic. It’s paywalled, but here’s the relevant passage:
The problem is not a lack of intelligence among managers. It’s a lack of education. Draw coaches out of the ranks of ex-professionals and you will mostly be dealing with men who have had virtually no meaningful academic schooling past the age of 16. Which makes understanding the maths and logic behind some advanced analytics extremely difficult.
That matters, because to benefit fully from analytics you need football men to be part of it. This means more than just blindly following the conclusions, but understanding the mechanisms behind the analysis and helping to generate the processes behind them.
Though it sounds a little paternalistic, I think there is something to this; any on-the-job club analyst I’ve spoken to emphasises the need to explain their ideas in a highly accessible, commonsense way. They also talk about the importance of “buy in”; unless one or several “football men” want to buy what they’re selling, analysts will fail to have a meaningful impact over their time at the club.
This, however, got the ball rolling in my head, and what follows is probably too big for its britches, but here goes…
The Movement with No Name
Sports analytics as a discipline has branded itself all wrong, and it may be too late to turn things around.
Let me explain.
Let’s go back to Marcotti’s argument. “…to benefit fully from analytics you need football men to be part of it.” The most relevant model here in sports is obviously Billy Beane at the Oakland A’s. Here was a former pro turned general manager, a trusted guy who fit the standard GM mold, who got in Paul DePodesta from Cleveland to finally bring SABR in from the cold.
But the story of Moneyball wasn’t about how DePodesta successfully communicated the value of SABR-driven methodology to Beane (in the film this is all done in a matter of seconds in an underground parking lot). It was about how Beane—who, as a young man, was told confidently by scouts he would be a star player only to carve out a mediocre MLB career—came to understand that our subjective judgments about the world around us are often fundamentally flawed.
These flaws lead us to make the same mistakes over and over again. But by understanding them, we can gain a vital edge over those who do not. Some of this involved advanced stats, yes, but at the core was this idea that our view of the world is skewed, is fallible, and that we have ways—including data analysis—to see it more clearly.
When we think of sports analytics, we think of “number crunchers,” “mathematical modelling,” “metrics.” We envision analysts from Harvard or MIT sitting in stadium offices, building databases and then writing reports to send to coaches and scouts. The term “analytics” itself is situated firmly in the discipline of mathematics: “the discovery and communication of meaningful patterns in data” as the Wiki entry helpfully puts it.
But I believe sports analytics is one small component of an expansive movement in the early 21st century, one I think eludes us because we’ve yet to give it a name.
I’m thinking here of the 21st century fascination with behavioural psychology that started with the work of Amos Tversky and Daniel Kahneman (author of the behavioural psych bible Thinking: Fast and Slow), popularized by authors like Malcolm Gladwell and Dan Ariely (Predictably Irrational), was the driving influence on books like Cass Sunstein and Richard Thaler’s Nudge: Improving Decisions About Health, Wealth, and Happiness, and is part of the same universe of skepticism which includes the likes of Nassim Nicholas Taleb (Black Swan, Antifragile).
This nameless movement will come to define popular thinking in the early 21st century.
Though these ideas were most important in helping to break down the last holdout of Enlightenment Era rationalism in the 20th century—classical economics—we can see evidence of the influence and receptivity to these ideas nearly everywhere:
- The rise of a “life hack” self-help culture which emphasises constant awareness of our inborn cognitive biases and distorted ways of thinking and working, and the attempt to manipulate of our “Paleo brains” to make marginal gains in our personal lives.
- The increasing prevalence of Buddhist psychology in secular Western culture, including the importance of mindfulness and awareness of the self as an aggregation of conflicting impulses and false or distorted thinking.
- A growing fascination with neurology and in particular the ‘modularity of mind’: “…the notion that a mind may, at least in part, be composed of innate neural structures or modules which have distinct established evolutionarily developed functions.”
- The development of targeted marketing techniques which manipulate our heuristic biases in order to get us to pay more for things we don’t need.
- A shift toward use of shifting demographics and polling rather than assessment of individual actors to understand political theory.
- A pervasive, general distrust of traditional pundits or self-described ‘experts’ in the media, and, related, the “End of Expertise.”
Where we once believed we were fundamentally rational creatures daunted only by vice and self-interest, we’re coming to greater awareness that what we once considered to be “rational” behavior and judgment is itself often hopelessly biased or distorted.
To borrow Dan Ariely’s description of his book, Predictably Irrational, “…we are all pawns in a game whose forces we largely fail to comprehend.” Because of that we’re wrong about stuff a lot more than we realize.
This, and not any argle bargle about numbers vs heart or geeks inheriting the earth is the message of Moneyball, and I think it should be the message of the entire sports analytics movement. It’s not about math being better than gut, it’s about how human beings are at their most dangerous when they act irrationally under the full belief they are doing the opposite. Or, as Josh Billings (and not Mark Twain) put it, “It ain’t what you don’t know that gets you into trouble. It’s what you know for sure that just ain’t so.”
The problem in sports, I think, is that we’ve confused the tools we use to understand what we think we know but don’t with the reason we use them. For example, there are countless mental blind spots in football culture that don’t require any math at all to understand:
- Does it make sense to hand sole responsibility for something as expensive and long term as player transfers to someone whose tenure will likely last less than two years?
- Should teams put all their hopes in a “great leader”, be it manager, technical director, owner/investor who will come in and make everything better? Or should they instead develop and systematize a top down club philosophy, a way of doing things that does not depend on any one person to survive?
- Is it smart to pay top dollar for an older player who just scored a ton of goals in one season after several years of so-so averages?
- It is wise to invest millions of pounds/euros/dollars in your academy infrastructure when your manager refuses to start young players in fear that it will affect results, and, in turn, their future job prospects?
Addressing these issues doesn’t explicitly involve data analytics. In fact, these are the kinds of organizational problems that might ideally be addressed before a club even begins to dabble in statistical science. For example, a club with a brilliant technical director whose recruitment ideas run completely counter to their manager’s will run into some significant problems, no matter how “low risk, high return” the signings.
Yet analytics obviously fits within this overall rubric of healthy self-doubt. It shouldn’t require an advanced degree to understand why having more information on the likelihood a new signing will succeed is often better than having less. But analytical thinking works best within a culture that embraces it.
This is what I mean when I constantly repeat Simon Gleave’s truism that we must learn to walk before we can run.
Reframing the Challenges Faced by Sports Analysts
So let’s not beat around the bush anymore about the challenges facing soccer analytics in the next couple of decades. The reality is some clubs won’t accept what the analysts are selling no matter how it is packaged or explained. In my experience, this is not something to overcome; it is a simple dead end. The job of a stats analyst is not to be an evangelist; it is to think as clearly and as precisely as they can. No amount of “simple communication” will overcome someone whose mind is made up. While this may mean fewer career opportunities for analysts, it also means those clubs which do embrace positive changes will retain their competitive edge longer.
There is another challenge however that I think is vastly more important (and interesting). The New Yorker columnist Elizabeth Kolbert put it best in her 2009 review of Predictably Irrational:
“If people can’t be trusted to make the right choices for themselves how can they possibly be trusted to make the right decisions for the rest of us?”
What if our predictive models, including Expected Goals, are dangerously flawed in ways we haven’t fully understood? What if managers really should be solely in charge of player signings, it’s just that they should be kept on the job far longer?
If the analytics movement is constantly self-correcting and debating and hemming and hawing at every turn, at every new metric and model and method from counting stats to TSR to xG, why should a club be so quick to adopt their methods?
I don’t have a pithy answer to this very difficult question, but I think this, and not ego or stubbornness, or is the real fault line of sports’ understandable resistance to analytics. This skepticism in my view must be respected. While public analytics should be bold, brazen, experimental and daring, in private, analysts should focus more on helping teams avoid major mistakes. Many analysts hate the idea they’ll be just another voice in a crowded room, but if that voice is saying loudly and confidently that signing so and so will provide little or no long term value to the team, they have done their job.
I think Chris Anderson said it well in his interview with Adam Bate last week:
“The moment that you take some of the emotion out of it, the moment you start using all the tools at your disposal, that’s the moment you do better. It’s not about getting every decision right but it’s about avoiding a lot of bad decisions. Ours is a football club that’s made a lot of bad decisions in the past and we’re trying to avoid those.”
Finally, there is a third challenge, and that brings me back to what I said earlier: sports analytics has branded itself all wrong, and it may be too late. By not positioning itself within the wider, nameless movement driven by behavioural psychology I detailed above, sports analytics will be likely forever regarded as part of some moronic ‘geeks vs jocks’ trope, “here comes the numbers guys”. But this view of sports analytics crams far too much in a tiny frame. It’s not just about math; it’s also about cognitive biases, about inefficiency, about hubris, self-doubt, about empiricism vs emotion.
I think Marcotti is right about the need for adoption from within, but wrong about the reason. The analytics movement is too important to be left entirely to the quants. It needs journalists, footballers, managers, owners to get on board. We need to expand the story to include people like Les Reed at Southampton, or Jean Michel Aulas at Lyon. Hell, we probably need to kill off the term “analytics” altogether. If we can ever get past the numbers cliche, then maybe we’ll just see what’s going on in football for what it is, and truly help those teams who want to succeed in the long term.