Home or Rome?
Who needs Ronaldo and Mbappé when you have the big lad Data Science in your team?
This is the closest of all the match probabilities that we have modelled throughout Euro 2020. Expect a tense and nervy affair.
The finalists' story
Looking back at the model we can see that both England and Italy were overwhelmingly favourites at all stages to qualify from their groups.
Both teams benefited from the draw but England more so. Italy were drawn on the harder side but had easier opponents in Austria in the round of 16. Italy had a 75% probability of winning that game although they needed extra time to do so. England were drawn on the easier side playing their potentially most difficult fixture, Germany, at home. England had a 53% probability of beating Germany which was their lowest probability of winning any match up until the final.
Given that the model predicts a more or less 50-50 game it's worth looking back in history to see who has the upper hand in meetings. England has never beaten Italy at a major tournament. In the 2014 World Cup, Italy won 2-1. At Euro 2012 the score was 0-0 with Italy winning on penalties. At Euro 1980, Italy won 1-0. History favours the Azzurri.
What about the dreaded penalty shootout? Italy has faced penalties 10 times in major competitions (World Cup and Euros) winning four times. All four of their wins have come in the last six competitions.
England has faced penalties eight times in major competitions, winning twice. England fans will be hoping it doesn't go to penalties.
How has data science done at the Euros?
As we reach the final match in Euro 2020, it's time to review the performance of our predictive modelling. Did it win the golden boot for accuracy or did it let the team down and let itself down?
Before a ball had been kicked we calculated that finishing the group stage 3rd with 3 points and a goal difference of no worse than -1 would be sufficient to qualify. We gave it a 67% probability and this is exactly the combination that Ukraine achieved to go through as the 4th ranked 3rd place team.
After the 1st round of group matches we calculated that there was a 71% probability of Italy, Belgium, Netherlands, England, Spain, France, Portugal and Germany all qualifying for the knockouts. All did qualify despite both Germany and Spain failing to win their opening matches.
Ahead of the final group games we revisited the probability of finishing in 3rd place and qualifying. The model calculated that the four 3rd place qualifiers would come from groups C, D, E and F. In fact they came from groups A, C, D and F. The model calculated that there was a 56% probability of the 3rd place qualifier coming from group A and 60% probability from group E. Although the model was wrong it was still very close.
There were some shock results that went against the probabilities in the round of 16. Czech Republic was given a 29% chance against the Netherlands and Switzerland a 34% chance against France.
In the quarter finals, the model calculated that Belgium would beat Italy but the probability was 52%-48% so Italy winning was no real surprise. All of the remaining quarter final and semi final matches went with the model's probability scores.
Overall, the model done great and more than justified its selection.
You can catch up with all our Euro 2020 probabilities from throughout the tournament here.
