Skip to main content

Turning the UEFA Euro into math

The Finnish company Weoptit, a company in Visma, has turned the UEFA Euro tournament into math and simulations. Based on a model originally built prior to the World Cup 2006, they have played out the tournament 1,000 000 times. Why? To find out what results each team can expect from this summer’s football festival.

""

This article has been written by Jukka in Weopit, and it was first published on Weoptit’s blog.

On the eve of the UEFA Euros, historical to Finnish football, we decided to create a simulation of the tournament to find out what kind of success each team could expect from this summer’s football festival.

The company is issuing no warranties (and takes no responsibility for any inflated expectation/hope levels), with these probability estimates. 

The model

We used an old and somewhat simple model that one of our analysts originally built for the World Cup 2006, to play out the tournament 1 000 000 times.

As the input evaluations of the strengths of the teams are subjective, we decided to use the FIFA rankings to classify the teams before the start of the tournament.

At first, we divided the teams into five main categories from A to E and then divided these into three subcategories (eg. A+, A, A-) by ranking points. A difference in the starting points of hundreds would imply a difference in the main category and each of these hundreds of points is divided into thirds to create the subcategories.

This will yield the following ranking (as per FIFA ranking at the start of the tournament):

Fifa rankingCountryFifa coefficientCategory
1BEL1783,38A+
2FRA1757,30A
4ENG1686,78B+
5POR1666,12B
6ESP1648,13B
7ITA1642,06B
10DEN1631,55B-
12GER1609,12B-
13SUI1606,21B-
14CRO1605,75B-
Fifa rankingCountryFifa coefficientCategory
16NED1598,04C+
17WAL1570,36C+
18SWE1569,81C
21POL1549,87C
23AUT1523,42C-
24UKR1514,64C-
29TUR1505,05C-
36SVK1475,24D+
37HUN1468,75D
38RUS1462,65D
Fifa rankingCountryFifa coefficientCategory
40CZE1458,81D
44SCO1441,43D
54FIN1410,82D-
62MKD1374,73E+

The matches between teams of equal strength are modelled to follow a probability distribution of; 37% home win, 26% draw and 37% away win and a difference of a single main category would tilt the probabilities roughly 15 points in favour (ie. 45-26-29) of the higher ranked team.

Since the tournament is played throughout Europe during Covid, we decided to not give any home team advantages for the hosting nations.

Additionally, in real life, the last group stage games might be affected by either previous matches or matches in other groups. These were not included in the model since these scenarios will become negligible as the number of simulations goes up.

The result

Here’s the breakdown of our simulation results (apologies to our English colleagues in advance):

CountryWinning probability%
BEL20,39%
FRA15,09%
ENG8,84%
SPA6,55%
ITA6,28%
POR6,09%
DEN4,82%
CRO4,71%
SUI4,45%
GER4,24%
CountryWinning probability%
NED3,35%
WAL3,02%
SWE2,14%
POL2,12%
UKR1,54%
AUT1,51%
TUR1,31%
SVK0,9%
CZE0,57%
RUS0,56%
CountryWinning probability%
SCO0,56%
HUN0,46%
FIN0,31%
MKD0,18%

Belgium and France seem to get an advantage from their tough groups since by surviving their group they are going to avoid their group opponents until the semifinal stage.

Furthermore, if they win their group, they will be guaranteed to face a team coming into the playoffs as a third-place finisher in their group. This will give them a higher chance to reach at least the quarter-finals than facing a runner-up of another group.

These outright favourites will then be followed by a range of teams that are quite close in strengths to each other implying final rounds for the tournament depending on the knockout bracket.

How will the Eagle-Owls do?

The Finnish team will play in group B with Belgium, Denmark and Russia. Our model predicted the following point distribution:

PointsProbability%
015,07%
120,42%
28,52%
322,86%
417,34%
53,18%
68,62%
73,06%
90,94%

The team will qualify from the group stage with a 44,6% probability, reach the quarter-finals with a 13,77% probability, the semifinal with a 4,5% probability and the final with 1,3% probability.

The probability for a big gathering at Senaatintori (where Finn’s go to celebrate big sporting wins) with the trophy would be 0,31%.

Want to know more about how we work with technology, AI and machine learning? Visit our technology category page.

Visit our technology category page