Questions and answers about
the economy.

Is football coming home?

England football fans are starting to sing ‘It’s coming home’; while followers of Scotland and Wales hope their teams can cause a few shocks at Euro 2020. But the chances of these outcomes are slim, despite what’s suggested by the bookmakers’ odds.

The delayed UEFA European Football Championship starts today, 11 June 2021. The media are ratcheting up the anticipation and pundits are filling the airwaves. Some football fans will fervently believe it is their nation’s time to taste success. Others will just be hoping that their team does well.

The definition of ‘well’, though, is a matter of expectations. For some, making it out of the group stage is a success. For others, anything less than the final is failure. The England manager, Gareth Southgate, has marked out the semi-finals as the minimum his team should achieve – and his job could be on the line if they don’t get that far.

Our analysis – which is based on measures of relative team strengths based on past performance plugged into a statistical model – suggests that the favourite is Belgium with a 24% probability of winning the title. But that still means that there is a 76% chance that Belgium will not lift the trophy.

Read on for the chances of success we calculate for every other team at Euro 2020 – and the opportunity to take part in our Euro 2020 predictions game.

Forecasting the results of Euro 2020

As the physics Nobel laureate Niels Bohr once said: ‘Prediction is very difficult, especially if it's about the future’. Even so, we decided to simulate the European Championships a few thousand times anyway. Why? Because it is fun, and as fans we can get some sense of how excited, or surprised, we might be about our teams’ chances of winning.

Our forecasting model is simple — it asks how many goals each team are expected to score in each match, given how strong they and their opponents are, and assuming that goals arrive at a particular rate (based on a probability distribution called the ‘Poisson’ – a very standard assumption in this context).

To measure relative team strengths, we use the 'Elo ratings' of each team, which are based on the entire history of international football – a recent win against a strong team is worth more than a win against a weak team. On the eve of the tournament, Belgium has the highest Elo rating in Europe, followed by France and Spain. North Macedonia is at the bottom of the list among the countries participating in Euro 2020.

Our model generates probability forecasts for the final score in a match (using the ‘Bivariate Poisson distribution’). For example, a 2-0 win for Italy in their opening game against Turkey is one quite likely outcome. After generating a match outcome from this distribution, we carry on to the next match, updating each team’s Elo ratings, and so on until we reach the final game. We do this whole process 10,000 times and count how often each team wins the tournament or gets to a particular stage.

The order of actions in tournaments can have significant effects on performance. This is as true for a major international tournament as it is for a kickabout at the local park. We have taken this into account in the model. The teams in Groups E and F have an advantage over others because they play on the last day of the group stage (23 June 2021). They will know exactly what results they need to be among the four best-ranked third-placed teams from the groups (which would allow them to qualify for the next stage).

For example, in Euro 2016, Portugal played on the last day of the group stage and knew that a draw would get them into the last 16 from third place. Coincidentally or otherwise, this is exactly what happened. Such a system also occurred in the World Cups of 1986, 1990 and 1994, as well as at Euro 2016. In all these tournaments, the third-ranked teams that played on the last day of the group stage always qualified for the knock-out stage.

A notable example concerns the Russian team in 1994, which finished third in Group B, one of the groups to finish earliest. Russia then needed only one of the following results to qualify: Bulgaria to lose to Argentina, Netherlands to lose to Morocco, Nigeria to lose to Greece; or Saudi Arabia to lose to Belgium. None of these happened and Russia headed home.

To illustrate the impact of this quirk on tournament outcomes, we assume that teams from Groups E and F, which play on the last day of the group stage, always qualify for the knock-out stage. The remaining two qualifying third-ranked teams are randomly picked from Groups A to D. In the forecast model results below, we compare the group stage qualifying probabilities to a hypothetical ‘fair qualifying’ schedule.

The countries that gain most from this are the teams unlikely to qualify in first or second in Groups E and F: Poland most markedly, but also Germany, Hungary and Slovakia. The countries that suffer most are those in Groups A to D that may have qualified in third: Ukraine, Denmark, Austria and Turkey.

Group A

 Fair qualifying probability (%)Qualifying probability adjusted to schedule (%)Difference in probabilities caused by schedule

Group B

 Fair qualifying probability (%)Qualifying probability adjusted to schedule (%)Difference in probabilities caused by schedule

Group C

 Fair qualifying probability (%)Qualifying probability adjusted to schedule (%)Difference in probabilities caused by schedule
North Macedonia30.629.1-1.5

Group D

 Fair qualifying probability (%)Qualifying probability adjusted to schedule (%)Difference in probabilities caused by schedule
Czech Republic49.546.6-2.9

Group E

 Fair qualifying probability (%)Qualifying probability adjusted to schedule (%)Difference in probabilities caused by schedule

Group F

 Fair qualifying probability (%)Qualifying probability adjusted to schedule (%)Difference in probabilities caused by schedule

Knock-out stage

If a knock-out match finishes level in the model, we decide the penalty shootout by assuming that the probability each team wins would be equal to the Elo prediction for that match (so the better team wins more than the worse team, but not always). This is known as a ‘binomial event’.

Below is a table showing the model’s forecast probabilities of winning Euro 2020 for each country, as well as the probabilities that each country makes the final, semi-finals and quarter-finals.

Czech Republic0.130.873.2015.38
North Macedonia0.

The model suggests that the tournament favourite is Belgium, with a 24% probability of winning the title. But it still means that there is a 76% chance that Belgium will not lift the trophy. The other teams with a greater than 10% chance of winning are France and Spain.

England fans can be happy that their 7% chance of winning the trophy is higher than Germany’s, but it does suggest that the bookmakers are hedging their bets against patriotic fervour, given some have England as equal tournament favourites with France.

Scotland have never reached the knock-out stages of a major tournament, and so will be heartened by an approximately 36% chance of reaching the last 16, and even a 9% chance of reaching the quarter-finals. Wales have a 9% chance of repeating their heroics of 2016 when they raced to the semi-finals.

How hard is it to predict football match outcomes?

The beauty of football, especially at the European Championships, is that seemingly anything can happen. In 1992, Denmark replaced Yugoslavia at the last minute and won the whole tournament. In 2004, Greece beat hosts Portugal in the final. Greece’s pre-tournament odds that year were 150-to-1 to win, which implies a probability of less than 1% – not dissimilar to our forecast for the chances of Wales winning this time. Scotland fans need to hope for a miracle.

Home advantage is not factored into our model, and is typically large in professional football. On a related note, the impact of travel fatigue could come into play. Due to the design of Euro 2020, which is taking place all over Europe, there could be significant scheduling effects. There is evidence from the post-Covid-19 period that the presence of fans does significantly affect football match outcomes, through social pressure and its effects on the referee.

Finally, the pre-tournament Elo ratings that drive the model are mostly based on recent non-tournament match outcomes. Some might argue that Belgium, for example, are overrated and may be underprepared for the grind of tournament football due to having less depth in their playing squad compared with the likes of France and Spain. These are all factors that, to some extent, make a mockery of our forecasts. But they are also why football is such an exciting and unpredictable game.

Where can I find out more?

Who are experts on this question?

Do you think you can beat the experts at predicting the Euros?

If so, click this link to submit scoreline predictions for every match during the tournament!

Authors: Alex Krumer, James Reade and Carl Singleton
Photo by Edoardo Busti on Unsplash
Recent Questions
View all articles
Do you have a question surrounding any of these topics? Or are you an economist and have an answer?
Ask a Question
Submit Evidence