top of page
  • Karnav Popat

Football: In Loving Memory — Where Data has Failed The Beautiful Game

Football used to be simple. Four Four F*cking Two, hoof it long, big man-little man with two hard-tackling hard-running hard men in the middle and a ball boy who knew how to see the game out. We didn’t have any of those fancy inverted fullbacks, trequartistas, mezzalas or lasagnas in the good ol’ days.

Now it’s all changed. Coaches draw complicated passing patterns and transitions on the whiteboard instead of thinking about passion and clarity. Players have fitness coaches, throw-in coaches, nutrition coaches and life coaches. But most concerningly, teams don’t think about good simple passing and tackling anymore — they think about their spreadsheets and xG.

xG (Expected Goals) was the dawn of the apocalypse for the beautiful game. It heralded the modern era of football, where a man can’t make a living just kicking the ball, but is constantly judged on his xT and his xPp90 and his NxPP%p90. Decisions that used to be passed through coaches and scouts are now controlled entirely by greedy American billionaires in boardroom seats and teams of nerds who can’t stop talking about VLOOKUP. Fans have fallen victim to the same unfortunate change in both demographics and mindset: good ol’ Steve who loved footy and beer has been replaced by the Liberal Arts Eco-Fin majors who’d rather talk about Brighton underperforming their xG and the failure rate of Bundesliga transfers.

All of this would be fine (well, not fine, but understandable) if any of this data stuff actually worked. But it doesn’t. The twists and turns of modern football have repeatedly proven that what matters, in the end, is heart and passion, the wings of destiny at your back and a roaring crowd engulfing the stadium in deafening chants. The moments we treasure are not the steady grind of the xG league table. It is Vincent Kompany’s match-winning screamer, a statistically horrific decision, that pulls every fan out of his seat and into a three-day bender. It was Lionel Messi’s absurd talent for eye-catching, mind-defying passes, Neymar’s skilful ease, and Suarez’s toothy smiles that made Barcelona the greatest team in the world. Look at the neat little rows and columns as much as you want, they will never make you feel like the Stretford End does on derby day; they will never make you jump like the penalty kick at 90+6’.

For years, the proponents of football’s data revolution have been taking loss after loss. Not just because a spreadsheet fundamentally lacks the capacity to capture the full picture of a dynamic sport like soccer, but also because the modern commoditized industry is filled to the brim with visionaries who seem to severely lack vision, and with executives who appear to constantly change their opinion on a data-inspired approach.

Consider Chelsea Football Club. The fans hate it. The haters love it. When Todd Boehly’s consortium took the club over, the perception was that he would bring a modern approach to the sport, one stemming from across the pond where the use of data is central to every approach towards strategy and recruitment.

Arguably, he has tried. On the one hand, Boehly and his team have no obvious depth of traditional football expertise to back themselves with, so of course they’d turn to new-age statistics. On the other hand, Chelsea have demonstrably been better statistically than on the pitch, so it seems fair to suggest that management’s strategy is working, at least in part.

Except, it’s not, is it?

The failure of Boehly’s Chelsea experiment is the strongest argument possible against the new age of decision-making by stats merchants and MBA holders, and the best evidence that actually knowing things about football still counts for something. It is as close to a thought experiment as possible: what happens if a successful franchise owner from a different, over-analyzed, data-dominated sport, switches over to football, bringing with him his commercial business sense and eye for analytical moves? The answer seems to be that he crashes and burns: football is not that simple; football is not a game that can be solved at the desk; football success can’t be bought.

The fundamental understanding that escapes Boehly and his ilk is that data simply cannot capture the essence of this sport. Football is not cricket, with neat little rows of balls, overs, runs and wickets. A cricket game can be modelled very easily, because it can easily be reduced to repetitive events with statistical variations. A game of football is complex, with limitless statistical outcomes shaped by moments of individual brilliance. xG, xT, and xP are all attempts to trivialise a beautiful sport,to reduce it to mindless numbers. They are doomed to failure, simply because Football is not meant to be trivialised. It is far beyond our capabilities to capture the wider context, narrative, and glory of a football game in a few numbers.

There is a tendency among football fans to make comparisons between football’s data revolution and baseball’s Sabermetrics. When Fenway Sports Group (FSG) took over Liverpool, they were widely expected to bring a ‘Moneyball’ approach to the game, and Liverpool fans tend to play this up as the reason for the club’s success over the past few years. And yet, Liverpool’s success has come from entirely the opposite direction: innovative tactics, advanced nutrition, a positive mindset and a long-term project with room for mistakes and constant innovation.

The areas where the FSG project has innovated with analytics hav ebeen entirely in contrast to the wider industry — their focus has been on optimising aspects of the sport that go unnoticed, but have the most disproportionate impact on day-to-day outcomes. They’ve been in the news, for example, for their employment of Thomas Gronemark as a throw-in coach: a classic example of Klopp’s project using analytics to find an improvable edge in an ignored aspect of the game rather than trying and failing to convert a game into a series of numbers.

That is how data should be changing football — not in breaking down the sport itself, but in optimising the industry that has grown around it and the knick-knacks that support our favourite ball-kickers. Unfortunately, that is also the aspect that executives seem most disinterested in consulting their stats merchants about.

Take the transfer market. In the modern sport, billions are spent every year, sometimes even nine-digit figures on individual players. Of course, 100 million deals would warrant a lot of research, right? Surely, surely, an industry that can’t stop talking about data would use every tool at its disposal to make sure its eye-wateringly expensive transfers won’t flop.

Evidently not. To experience the life of a stats merchant, I decided to conduct an experiment of my own. Consider this — one of the clearest trends of the last half-decade has been the rise of the Bundesliga as an attacking spectacle. The German league has given us some of the best offensive arsenals of the sport, ranging from Lewandowski and Co. to the deified Frankfurt trio of Haller, Jovic, and Rebic. This has naturally resulted in a lot of interest and transfer activity for attackers in Germany, featuring high-profile targets like Jadon Sancho (€75 million) and Kai Havertz (€85 million).

In fact, from the 2017-18 season to 2023, Premier League clubs have spent a total of €738 million on 28 different attacking players from the Bundesliga. Less than two hours of research — the bare minimum I’d want elite, professional sporting institutions to put into their transfers — would tell you that all of them have flopped!. Every single one of them! The last attacker from the Bundesliga who didn’t flop in England was Pierre-Emerick Aubameyang, in January 2018.

This is particularly egregious because 2018-19 is when the exodus of German attackers to England really kicked into full gear, both in terms of the volume of players and the amount splashed out on them. Pulisic (64m), Haller (50m), Joelinton (44m), Werner (53m), Sancho (85m), Sargent (9.5m), Bailey (32m), Rutter (28m) — I look at that list and see a string of some of the most expensive failures in football.

So where is the data analysis when you need it? When Aston Villa were considering the purchase of Leon Bailey for a then-club-record transfer fee, did nobody think to look at the long tradition of Bundesliga tricky wingers who’d come to England and been completely dismal? This is not an inexplicable phenomenon, nor one that is difficult to spot; of course offensive players from one of the most offensive leagues in Europe will struggle to adapt to the much more defensive and physical Premier League.

But it’s more than that. The most casual fans of German football would tell you that the league is extremely reliant on counter-attacking. This isn’t just a fact that seems obvious from the eye test: it also has backing in foundational statistics. The Bundesliga sees twice as many goals from counters as the Premier League, as well as more goals in general. Why, then, would you buy counter-attacking players from Germany and try to make them adapt to English football? How much research did Frank Lampard put into his recruitment of Timo Werner, a striker almost entirely reliant on counters, with the expectation of breaking Chelsea’s striker curse? Or, for that matter, Sancho, or Pulisic, or Havertz, or Haller — what data fuelled these decisions?

The essential issue with the integration of data analytics into the football industry is that it has consistently eaten into the football part, rather than the industry part. Players today are increasingly forced to subsume themselves to nonsensical schedules built to squeeze revenue, force themselves into spreadsheet strategies that implode on the field, be judged on convoluted, flawed metrics that completely fail to capture a hundred different variables. Meanwhile, industry executives are given free rein to strategize based on vibes, without the slightest concern for either traditional football history or modern analytical insights.

Data is a lever — not an ideology, but a tool to be used according to its purpose by those who know where to use it. Its place is in the manager’s office, to study broader industry-wide trends and to make better decisions using informed context. It can be extremely useful, and it does have a place in the sport — in the manager’s office, supplying good, hard numbers to backstop the exponential spending and hair-trigger business decisions that have come to characterise the sport of late. Its place is not, however, on the pitch or anywhere near it. No metric will ever be able to represent the beauty of de Bruyne’s passes, no analysis should ever explain what made Cruyff’s tactics special.

The paradigm shift of football’s data revolution has allowed many outsiders, non-footballers and general stats merchants the ability to have opinions on football: but is that a good thing? Do we really want more Todd Boehlys? A sport where decisions are made by committee and spreadsheet is not a sport where Ronaldinho and Messi and Hazard would thrive. The word ‘magic’ is thrown around quite often when talking about football — “Keep the magic alive”. What part of the entertainment spectacle benefits from dribbles and rainbows being replaced by aggregates and returns to the mean?

Today’s football is not the game I fell in love with. Its narratives are not the ones that lit fires in my heart. Football’s modern owners have made the sport their plaything, reduced it to a speculative bubble stuffed with publicity stunts and excused by vague appeals to “Big Data” and “Solid Numbers”. In the process they have sucked the heart out of a sport that captured so many. For all that the world can’t stop talking about data, AI, analytics, statistics, and for all that I respect the value that xG and its equivalents have brought to the sport, I can’t help but hope — can we go back to the good ol’ fashioned centre forwards, sending the ball long, and Four Four F*cking Two?

117 views2 comments

Recent Posts

See All


Sep 20, 2023

I SINCERELY hope that this is satire.

Nov 07, 2023
Replying to

that's a stupid take

bottom of page