The first of the big games are behind us, with a few surprises to keep us on our toes. As you know, Blue Yonder’s founder and chief data scientist is watching Euro 2016 not only as a football fan cheering Germany to the final, but also with anticipation on the predictions created using data from all historical international football games. This data includes 36,000 international games dating back to 1873 and the data points include venue, date and score.
Blue Yonder plugged this data into an advanced machine learning algorithm, which then generated probabilities of each team’s chances of winning each game. As you know from our last update on this, there are 94 billion possible ways the tournament could play out. In light of this, we used the Monte Carlo method; a probability simulation often used in the Financial sector, to help figure out each team’s chances of lifting the trophy on 10th July.
Based on the historical data, our algorithm predicted:
- France are expected to win the tournament and have a 34.1 per cent probability of winning. This is double their closest rival, Spain, who only have a 13.4 percent chance of winning, in comparison
- England only have an 11% chance of winning the tournament, but have a probability of 56.7 percent that they will finish top in the group stages, and a 93.3 per cent chance of qualify for the last 16
- England (11%) have a better chance to win the tournament than Germany (9.8%)
- Unfortunately for the other home nations, Wales only have a 0.14% chance to win the championship, while Northern Ireland are least likely to win the tournament, with a 0.04% chance
In the opening stages there have been a few surprises, from Wales’ victory over Slovakia to the last minute goal secured by Russia against England. We’ve plugged the updated data into our algorithm to see how the probabilities have changed. There have been a few shifts, but this has not impacted the overriding predictions.
Here’s a summary:
Wales-Slovakia: 2: 1
Wales’ victory had a probability of 29% and a defeat had the probability of 47%. This was the first surprise.
England-Russia 1: 1
This draw was predicted with a probability of 29%. A victory of England had the higher probability of 47.7% before the game. Here Russia's late goal made a significant difference.
France (68.1%), Poland (63% against Northern Ireland), Spain (59%), Germany (58.9%) and Switzerland (52.1%) all won their games as the probability statements from the algorithm predicted. For the Croatia-Turkey game the probabilities were almost equal at 37.5%: 31.5%, but the favoured team – Croatia – did win. The similarly predicted game Ireland vs. Sweden (37% Ireland, 28% Draw, 35% Sweden) finished 1:1. Finally, Italy won despite as predicted (36%) against Belgium (32%).
Based on the new information, Welsh fans in particular will be delighted to hear that the probability of Wales getting through the knockout stages has increased from 29% to 76%, and the probability of reaching the quarter-finals, from 8% to 25%.
The chances of Germany have been marginally improved by their victory and the chance to become group winners has increased from 52.6% to 75.9%.
Nothing has changed in the overall winners’ probabilities. France still has the best chance with 35% chance of winning Euro 2016.
Forecasts for today's games:
Hungary - Austria: 26% Hungary, 37% Draw, 37% Austria
Iceland - Portugal: 14% Iceland, 18% Draw, 68% Portugal