What's in the way of analytics solving football?

Sea shanties, Wordle, and wind tunnels


Mark Thompson

Jan 13 2022

7 mins read


Sometimes newsletters are timeless, sometimes they age faster than the popularity of sea shanties. The intro to this one will make it a time capsule.

The nerds are ruining Wordle. Not just the tech sharks trying to monetise a game designed to be free, the coders algorithming up a guaranteed solution. They've solved it. AI solved chess. One day, will analytics do the same for football?

Well, no. I don't think so, at least. But why not?

A lot of people, including this newsletter, spend a lot of time thinking about what football analytics can do. The possibilities it can reach to. But I think that there are limits. Thinking about what those limits are might help work out where the current ceiling can be pushed further.

The problem is the ball

Let's start with why I'm so confident analytics won't solve football.

The first reason is simply that if 'solved it' is the summit of Mount Everest, we're a long way off even getting to base camp. Football is, I think, at least half a decade behind basketball in the use of tracking data, if not moreso. And football has double the number of players on the pitch. And even basketball isn't solved yet. When basketball is as 'solved' as chess is, maybe I'll start worrying a little.

A second reason is that football pitches are just so darned big. Really big. It's what gives rise to the Rafael Benitez's 'short blanket' analogy - you can pull the blanket up high, but then your feet will be cold; you can push your defence up high, but then you'll leave space in behind. Players may get faster, and goalkeepers more prone to sweeping, but it seems unlikely that football will ever get around its 'short blanket' problem completely. And that leaves a variety of strategies on the table.

My third reason, and possibly the most crucial, is (with apologies to Americans) literally in the name of the sport.

1) 'Foot' 2) 'ball'.

The ball bounces around in a hard to control manner. The feet are the appendage permitted to control it; notably hands, the parts of the human body that evolved as expert manipulators of objects and tools, are outlawed.

Even expert players misplace fairly simple passes, or take slightly sloppy touches, in many games. Add to that the fairly permissive officiating, and football is just a sport that, by its nature, makes it very difficult to keep control in key areas.

This lack of control also contributes to the very short amount of time that players have to make decisions. All of these factors make it seem unlikely that there will ever be a definitive set of tactics that should be followed to win football matches. Cycles of evolution will, I think, always occur.

However, analytics will help these evolutions take place. Just like inventions have sped up the social and intellectual evolution of humans, it seems fair to assume the same will happen in sport. I think that the most likely future is one where high-quality data analysis helps teams to find the holes in the game's current 'meta' (most effective available tactic) and identify how to exploit them. This analysis may only take months, or weeks, or even days.

But after that, it's down to coaches and players to implement it.

The problem is people

If we accept that analytics won't solve football, but might speed up tactical evolutions, we now need to think about what limit there'll be on this process.

Let's take a look at how 'playing out from the back' developed. Assuming that more money going into the sport inevitably leads to technical quality rising, football was probably always going to get to a point where defenders looked comfortable on the ball. Barcelona and Spain's possession-heavy dominance of the late 2000s-to-mid 2010s probably gave it a shove. And then the high-pressing trend which immediately followed gave an added urgency.

But footballers don't just become highly technically-adept overnight, even at the top level. We're only just, more than a decade after 'tiki-taka' entered the lexicon, getting to a point where top-division defenders are more likely to be comfortable than uncomfortable on the ball.

This, while an extreme example, helps us frame our thinking. If crossing suddenly became the way to create chances, how soon would it take the wingers and full-backs of European football, starting from a point where barely any of them have been brought up to prioritise the skill, to become experts? Eighteen months, perhaps? Maybe even longer? It doesn't need to be all of them, just enough to make a difference in the majority of teams, but let's not forget that players go entire careers without reliably being able to put a good ball in the box.

And edging slightly away from technical ability, think about centre-forwards. There's a complete dearth in them at the top level at the moment, to the extent that the options last summer for Manchester City, who surely had the pick of the world's talent, appeared to be: 1) Harry Kane 2) an ageing Cristiano Ronaldo 3) no-one. Not able to get number 1, and understandably not wanting number 2, they went for number 3.

If football's market of talent was an efficient one there'd be a spate of players who could fill that focal point striker role - but there isn't, because the market for talent is seeded by the trends of several years ago. Things are liable to go even slower when the gap in the market that needs filling is not only a technical and/or intellectual one, but one requiring a specific physical skillset as well.

I think that there's a chance that coaching approaches could shorten the physical 'time lag' as well as the technical one, but not by much. If a tactical strategy requires an insightful passer, you can train one from the entire pool of players at your disposal: tall or short, freckled or bespectacled. If a tactical strategy requires a tall and weighty striker, you're stuck with what God gave you. (In reality, a way that coaching could shorten the physical 'time lag' is by creating strategies to get around physical 'shortcomings', thereby widening the pool of potential players).

A final metaphor

I'm big on metaphors and analogies so here's one final one.

I don't think football will solve football and provide one tactic that every team will follow. What I do think could be a more realistic (though ultimately false) fear is that analytics makes it more like Formula One.

In F1, the driver is alongside the engine as a major feature. In the olden days of motor racing, they (as well as reliability) were probably the only two key features. But now, aerodynamics is a big thing. Barely a season seems to go by without it being big news that one of the top F1 teams added some random bit of carbon fibre to make air go round the car a little better.

If, in this metaphor, the engine is the players, and the driver is the manager, analytics teams are the... the aerodynamicists(?, fitting given that some of them are former physicists). The analytics team of the future might identify, and possibly even design, new features for a tactical plan that could streamline build-up, help cool down opposition attacks, and generally make things run much smoother.

(If we want to stretch it further, these random pieces of carbon fibre are not always strictly within the rules. In football, analytics types appear to be responsible for the increase (or at least the advocacy of it) of basketball pick-style blocks at set pieces, which are also not necessarily strictly withing the rules. Working out where the rules can be pushed to breaking point and where officiating can be exploited is still, tiresome though it may be to watch its results, analysis)

'Best aerodynamics' is not the be-all and end-all of what wins F1 teams championships though. And football, currently, is probably only just discovering what a wind tunnel is.

Read more posts like this in your inbox

Subscribe to the newsletter

Nature of Football