Euro 2024 semi-finals update: is football coming home?

As the European Championship nears its conclusion, each remaining team has roughly a 50% probability of being in the final in Berlin on Sunday. France and Spain have around a 30% chance of lifting the trophy, while the English and Dutch have about a 20% chance.

The 2024 European Championship enters its final week with three games yet to play. Three crucial games towards determining who will win – and hence how the tournament will be remembered.

But what of it so far? Plenty can already be said: since 48 of 51 matches have been played, most of the statistics for the tournament won’t change dramatically now. There have been 108 goals to date, hence 2.25 goals per game on average. This is some way below the Euro 2020 average of 2.8 goals per game, though higher than the 2016 average of 2.1.

The tournaments in 2004, 2008 and 2012 all delivered 2.5 goals per game, and Euro 2000 provided an average of 2.7 goals. As Euro 1996 only delivered 2.1 goals per game, the overall average for the 16-team Euros is 2.45 goals per game, while for the expanded 24-team format since 2016, the average is 2.38. That suggests no clear impact of the expansion of the competition on numbers of goals.

Figure 1: Historical goals per game at European Championships

Source: UEFA, author's calculations

How tight a tournament has it been?

The final four countries competing in the last three matches are four of the six strongest at the tournament’s starting point. France have eliminated the third (Belgium) and fifth (Portugal) best sides en route to their semi-final against Spain, who are now the strongest team in the tournament thanks to their impressive progress to this point.

Furthermore, the final four countries couldn’t have been paired more closely in their semi-finals. Spain’s Elo rating is a mere nine points better than France’s; and the Dutch rating is less than two points better than England’s. A team with a two-point better Elo has a 50.3% chance of winning any such encounter; and a team with a nine-point better Elo has a 51.3% chance of winning. So roughly speaking, each nation remaining in the tournament has a 50% chance of being in the final in Berlin on Sunday.

But the superiority of France and Spain over England and the Netherlands is such that the combatants in the first semi-final have a roughly 30% chance of lifting the trophy compared with the English or Dutch, who have about a 20% chance.

In getting to this point, England and the Netherlands have beaten sides that they would be expected to beat: Romania, Slovakia, Switzerland and Turkey. France and Spain have had to defeat strong opposition: Belgium and Portugal, plus hosts Germany, who drew some benefit from home advantage.

But most of the matches on both sides of the draw, and indeed during the group stages, have been tight. Over the whole tournament, the goal difference between the two teams in a match has averaged 1.08, the lowest such difference since 1996. That suggests that matches have been tighter in this competition, and hence more exciting, than they have been for quite a long time.

The tightness has led to some quirks. In the quarter-finals, all four of the teams that progressed did so despite recording a lower expected goals than their opponents. Additionally, three of the four clashes went to extra time – and two had to be decided by a penalty shoot-out.

Figure 2: Team strengths

Source: Elo ratings
Note: Bars show the pre-tournament team Elo ratings; points show the latest team Elo ratings, following the quarter-final matches. Remaining teams are highlighted in blue.

What do we know about penalty shoot-outs?

Penalty shoot-outs have long been a huge focus of attention. Some of it is more speculative in nature, but plenty of research by economists has gone into shoot-outs over the years.

Ignacio Palacios-Huerta, an economist at the London School of Economics, is perhaps most readily associated with the study of penalty kicks, pointing out the extent to which they represent a classic two-player, instantaneous-move game – the type that game theorists commonly use when thinking about optimal strategies. It turns out that England have been consulting Palacios-Huerta since 2018, coinciding with a notable upturn in their success in this kind of tie-breaker.

The study of penalties continues, not least since they happen so often. In the Copa America, Latin America’s equivalent of the Euros and taking place concurrently, three of the four quarter-final ties went to penalties. And last month, the University of Zurich hosted a workshop featuring a number of talks on the topic, including a study by Leo Morabito (University of Milan) considering gender differences in shoot-outs.

Zeki Amdouni’s fifth penalty kick for Switzerland against England last Saturday was placed down the middle of the goal. Indeed, much of the development of the modelling of shoot-outs has broadened out the action set that each player has from two moves (shoot left or right) to three, adding shoot down the centre (sometimes called the panenka). Morabito finds that in women’s football, almost nobody shoots down the middle, and hence it is still a two-by-two game, statistically speaking.

Is a bigger or smaller tournament most appealing to fans?

Returning to the broader set of statistics for Euro 2024, does the relative lack of goals along with the tightness of results indicate that matches are becoming more unbalanced, and that the weaker teams are simply defending? Has the optimal response to opening up the tournament been that weaker teams become better at defending, rather than better at attacking?

Slovenia (the 19^th best nation in the tournament by Elo rating) were clear exemplars of this possibility, conceding just two goals in their four matches. But Georgia (23^rd best) and Turkey (17^th best) showed much more attacking flair and both progressed at least as far as the Slovenians. We can investigate this a little further by creating an uncertainty index over the tournaments using the Elo ratings and predictions we create to make our forecasts of match outcomes.

The Elo prediction for a match is a number between 0 and 1 representing how likely the team listed first is to win – 1 being that they are certain to win; 0 that they will lose. Hence 0.5 is the most balanced possible match – incidentally, that is essentially what both semi-finals are this week.

One balance measure is to multiply the Elo prediction for match i, elo_i , by 1-elo_i, hence elo_i (1-elo_i). This measure achieves its maximum, 0.25, when the Elo prediction is 0.5 – that perfectly balanced match. Movements away from there will yield lower values of this measure.

This Elo balance measure achieved its highest value, 0.24, in the early editions of the European Championship, and most recently in 1984 and 1992. This is consistent with matches having an Elo prediction of between 0.4 and 0.6, hence quite close to 0.5 and balanced matches. But this measure has dropped to 0.21 since 2016, indicative of matches having Elo predictions on average of 0.3 or 0.7, a less evenly balanced contest.

UEFA thus has the kind of quandary that faces the organiser of any tournament: how big should it be? A smaller and more elite competition may discourage other nations in their sporting development. A larger and less restrictive competition may improve the quality of less developed nations, but it may well affect the attractiveness of the competition if those lesser developed teams pursue defensive rather than attacking strategies.

Defensive strategies are often optimal in football due to the low-scoring nature of the game – and as we have seen, there have been fewer goals at Euro 2024 relative to Euro 2020. But is this just a sample of one? Waiting for a few more observations may not be ideal if the commercial impacts of more defensive football are that fewer fans tune in.