What Will Soccer Look Like in the Future?

Lots of subs and set pieces

Magtdemonstration i topkamp: FC Midtjylland ydmygede FC København ...

Dan’s request is today’s topic: “I know that this has been a general theme in your writing, but it would be really cool to hear about some of the most aggressive strategy shifts today (e.g. RB Salzburg, RB Leipzig, others I haven't heard of), and some of the biggest changes in terms of team composition (e.g. Liverpool).  It would be really cool to hear about what the crazy shifts could be over time. For instance, think of the NBA and how many 3's are taken vs the past.  Or, think about football and how many teams have shifted to a lot of shotgun and a much higher passing vs running rate.  What could those shifts be in soccer?”

What was happening 10 years ago?

I was about to graduate from college, still years away from getting my first real haircut, but you’re not here for that. No, a decade ago, Inter Milan were being coached by Jose Mourinho, and they were grinding their way through the Champions League, en route to an eventual European title. At that time, Mourinho seemed like the only manager who could stop Pep Guardiola, and he still seemed like the only manager you’d want if you absolutely had to win a single game.

Inter were dominating Italy, while Pep’s Barcelona were scratching at triple-digit points in La Liga. Lionel Messi broke the 30-league-goal barrier for the first time -- something he’d do in seven of the next nine seasons. Bayern Munich were winning the Bundesliga, while Marseille ... Marseille ...

Marseille were winning Ligue 1. In England, Liverpool were dropping out of the top four for the first time since they won the 2005 Champions League, and Chelsea, under Carlo Ancelloti, were winning the Premier League with one of the great teams of the competition’s modern era: a plus-71 goal differential that’s only been eclipsed by the two historic Manchester City teams from the past two years.

Messi would go on to win the Ballon d’or, while his club teammates, Andres Iniesta and Xavi, would finish second and third in the voting. Check out the rest of the top 20/my recent Wikipedia history:

Some things have shifted since: Mourinho is currently just barely hanging on as a true influential force at the top level of the sport, Liverpool are back at the top, Juventus have re-surpassed Inter to become the dominant power in Italy, and PSG are now owned by the Qatari sovereign wealth fund. But Bayern and Barcelona winning isn’t anything new, and Lionel Messi remains the world’s most dominant individual player.

However, while the rules of the sport broadly remain the same, the way the game gets played within those constraints has begun to shift.

Per Stats Perform data, the average team took 13.3 shots per match in Europe’s Big Five leagues back in 2009-10, and that number has ticked slightly down to 12.8 in 2019-20. However, the quality of shots has gone in the other direction, and by a much greater magnitude. The average expected-goal value per shot in 09-10 was 0.075, and this season it’s up to 0.111. The number of possessions per match has declined since 09-10, but that’s because of more successful possession play. Pass completion is up to 80 percent this year from 75 percent in 09-10. The average sequence (an uninterrupted possession, i.e. when the ball doesn’t go out of bounds or a foul isn’t called) lasts for 8.2 seconds in today’s game, compared to 7 seconds 10 years ago, while the average sequence now contains 3.2 passes, compared to 2.6 in 09-10. The average speed of a sequence (the number of meters the ball advances up field per second) has declined from 2.06 to 1.58. And while it might seem strange that the pressing rate has declined (opponent passes allowed per defensive action, or “PPDA”) from 10.85 to 11.89, that has more to do with the increased passing proficiency than anything else. The average sequence begins farther from a team’s own goal line (46.2 meters in 19-20, 44.5 meters in 09-10), so the ball is being won higher up the field than it used to be. Lastly, while the number of passes completed into the penalty area is nearly identical (11.97 in 19-20, 11.77 in 09-10), teams have slowly weaned themselves off of a cross-heavy diet: 22.8 percent of final-third passes were crosses 10 years ago, and that’s declined to 17.5 percent this season.

I was completely unaware of any kind of soccer analytics back in 2010. Frankly, I didn’t even really pay attention to goals, assists, or even a team’s goal-differential. I was a creative-writing kid who also played soccer; I took one math class and one econ course, and I hated both. You, a subscriber to this newsletter, are aware that has changed. And while the influence of analytical thinking has grown rapidly since then, the actual implementation of that thinking still hasn’t had a massive effect on soccer; it’s nothing close to any of the major American sports leagues. Liverpool are the only big club that are really doing it -- in the sense that the guy at the top of the sports-related-decision-making arm of the club, Michael Edwards, is committed to using numbers to help make decisions. Plenty of other teams have analysts -- most of them, even -- but they have a harder time getting their ideas heard or acted upon. Plus, even Liverpool aren’t really a good example of this. They’re one of the richest clubs in the world. They don’t need to think too far outside the box; they can find their edges, but mainly they’re able to just identify players way more efficiently than anyone else at the top of the European pyramid.

Still, despite the lack of true inroads data-based thinking has made into the game, some of the shifts in game makeup do jive with analytics thinking. Winning the ball higher up the field has been shown to have numerical value, ever since Charles Reep was notating games with a miner’s helmet on back in the 60’s. On the whole, crossing is an incredibly inefficient way to score goals. And the core, modern analytical finding -- take better shots -- seems to have grabbed hold. I’m not sure if that’s specifically because of the influence of numbers; in fact, it probably isn’t. Rather, it’s that the coaches and teams who implemented these various approaches tended to be successful -- as the various studies have suggested they would -- so more teams and coaches started employing the same strategies. Yes, a long-range goal won them the Premier League last year, but I can’t imagine Pep Guardiola loves it when his players let rip from deep; that seems like a quick way to lose playing time and get chewed out in a video session. Sure enough, City lead all teams in the Big Five leagues with chances created from pull backs over the past two-plus seasons -- 52, no one else above 40. These passes go against the defensive grain, and they also create shots with your feet from the center of the box -- an obvious-seeming desirable thing that fits into the grand analytical tradition of “three is more than two.”

To really see where soccer is headed, I think you need to look at FC Midtjylland in Denmark. They’re owned by Brentford owner Matthew Benham, who made his money by outsmarting betting markets, and while Brentford have done an amazing job in the Championship -- they remain competitive every year despite a massive revenue disadvantage -- Midtjylland are seemingly removed from the financial pressures of English soccer, so they’re able to steer even further outside the box.

“If you can outspend everyone else, then there's no need to start thinking about how you can out-think them,” Midtjylland chairman Rasmus Ankersen told me. “An abundance of money, I think, kills creativity and innovation very often.” Midtjylland don’t have as much money as their competitors in Denmark, and so creativity has become their currency.

The Dutch journalist Michiel De Hoog wrote a great story about the club back in 20015. (His stories aren’t always in English, and I can’t read Dutch, but I’m a huge, huge fan of his work.)  At Benham’s clubs, they refer to what they call a “table of justice”, rather than the standings; it’s sorted by an xG model, and decisions are made against that. “There's so much randomness that the same performance can swing you 15 points one way or the other. So we don't work with the league table,” Ankersen told me. They also identified a midfielder named Tim Sparv for no other reason than “he played a lot of minutes for a second-division German team that Midtjylland’s model rated highly”. They figured he was a winning player, even though there was no real data to suggest it, and Sparv went on to be a key figure for FCM as they won a bunch of league titles and famously beat Manchester United in the Europa League. Perhaps most notably, they’ve made it a mission to dominate set pieces -- the one phase of the game you can script and practice over and over again.

When they won the Danish league title in 2014-15, they scored 25 set piece goals, and no one else in the league broke 11. As StatsBomb founder Ted Knutson, who used to work for Midtjylland and Brentford, told me for a piece I wrote for ESPN:

It was almost an unintentional economics experiment. Midtjylland won the title in 2014-15 and then discussed how set piece training and execution was a huge part of how they won it. Other teams in the league immediately started paying more attention to this phase of the game, and most of the teams in the league started scoring more goals- - in some cases A LOT more goals -- because of it. Critics initially suggested that FCM only scored as much as they did on set pieces because Danish teams were bad at defending them, but the Danes were fine at defending them BEFORE we found an edge, and it's not like the whole league could become bad at defending one phase of the game at the same time.

Midtjylland won the league again in 17-18 with 25 goals again, but two other teams broke 20 and eight others broke double-digits this time ‘round. Overall goal-scoring in the league increased, too, which essentially proved that every team is capable of scoring more set piece goals without sacrificing any other aspect of their play. It seems clear that anyone -- in any league -- who isn’t doing this is leaving points on the table.

What’s next for Midtjylland? I recently spoke to Liverpool throw-in coach Thomas Gronnemark, and he told me that Midtjylland were now dominating the Danish league with, yes, their throw-ins.

So, the future of a more aggressively analytic-ized game includes way more set piece design, even fewer long range shots, even less open-play crossing, a de-emphasis on what happens on the ball in the midfield, and a way more aggressive approach to substitutions. There’s one far-off future where all those factors combine to create a product that’s absolutely awful -- frantic up-and-down play that’s mainly dependent on getting the ball near the opponent’s goal as fast as possible and hopefully winning free kicks. But just as analytical thinking still hasn’t come close to breaking through in soccer; the analytics themselves are still only scratching the surface. Beyond shots, we still don’t have a good handle on how valuable all of the things that happen with the ball. Plus, on-ball actions are only a fraction of what happens in a game, and we’re even further away from really understanding just how meaningful all of the things that happen off the ball are. And so, that, to me, is what makes these ideas so compelling; it’s why I started this newsletter. What will the future of soccer look like? Just about anything is possible.