As many did, I participated in pre-tournament prediction competitions for the Euros. As I didn’t know much about most of the teams, I largely based my predictions on the Elo rankings. Elo ratings are essentially Chess ranking applied to international football teams. Each team has a numerical score which is increased or decreased based on the results of their matches, with the new ranking being based on what would have been expected from their current rankings. So for example if team A with a rating of 2000 plays team B with a rating of 1700, even a win for team A will not increase their ratings by a large amount, but even a draw would significantly increase the rating of team B.

So here’s what I made of each group’s eventual results with respect to their pre-tournament elo ratings.

Group A

  • Germany 1920
  • Hungary 1832
  • Switzerland 1805
  • Scotland 1770

From the Elo ratings here, Germany were approximately 100 points ahead of the others who were relatively closely matched, albeit with Scotland a little behind. Hence one would expect Germany to win the group and second place to be something of a scramble. I think this prediction holds up pretty well with reality. Germany did win the group, although required a late goal against Switzerland to do so. The other three teams were indeed relatively closely matched, although Switzerland did somewhat outplay Hungary in their first match.

Overall, I’d say the ratings held up pretty well here. You probably would have expected Germany to beat Switzerland, and Hungary to have drawn with Switzerland, but both well within the bounds of variance.

Group B

  • Spain 2020
  • Croatia 1969
  • Italy 1950
  • Albania 1624

Three relatively closely matched teams with a rank outsider. Spain were certainly the favourites for this group, but nonetheless could be said to have outperformed the ratings here, winning all three matches. Even the two 1-0 victories were in truth a bit more dominant than that suggests. Croatia were perhaps a little unfortunate to go out here, but not by much, and the draw with Italy, very much predicted from their slender lead in the ratings, it was the draw with Albania that saw Croatia exit the tournament. Albania, despite amassing a single point, have, if anything, outperformed their pre-tournament ranking here, in which one would have expected 3 losses, and even losing by only a single goal in both other matches was admirable.

Overall though, I’d say this group also conformed fairly well to the ratings, Albania’s snatched draw against Croatia the only result to really go against what would have been predicted.

Group C

  • England 1981
  • Denmark 1834
  • Serbia 1801
  • Slovenia 1733

The ratings do not predict anything about how exciting the football will be. England with a ratings lead of 150 over second rated Denmark should have been expected to dominate the group more than they did. One might expect a draw to Denmark or Slovenia, but not both. Serbia, if anything a touch unfortunate to be knocked out here, since they drew both their matches with Denmark and Slovenia, and were only consigned to last place because England failed to beat either of those teams as well. The draws between Denmark, Slovenia, and Serbia, mean that Slovenia somewhat outperformed their pre-tournament ranking, drawing with all three teams, despite large ratings gaps to England and Denmark.

Overall then, this group, largely by failing to produce anything resembling interesting/exciting football didn’t quite conform to the pre-tournament ratings, but there weren’t any out-and-out shocks. The main thing is England only getting 5 points from what should have been an easier group than that.

Group D

  • France 2077
  • Netherlands 1972
  • Austria 1863
  • Poland 1738

Austria winning this group represents something of a shock. Even losing only 1-0 to France, a team more than 200 ranking points ahead of them, was pretty decent. That Poland were eliminated after two games was no shock, with 125 points difference to their closest ranked competitors. The shock here was that Poland held France to a 1-1 draw, which meant France didn’t win the group. To draw with the Netherlands was a little disappointing given the 100 point ranking lead, but to draw with Poload despite a 350 point ranking lead is a huge disappointment, albeit one that came after France had already qualified, but not winning that game has put France in the tougher side of the draw.

Austria beating the Poland was fair enough with a 125 point gap, but beating the Netherlands, verges on a shock result. In truth the game was something of a mad-cap game, the Netherlands actually won the xG battle 1.73 to 0.95.

Overall though, Austria were better than might have been predicted, and France were worse. Netherlands a touch unfortunate to have done worse than would have been expected, and Poland, if anything, outplayed their pre-tournament ranking.

Group E

  • Belgium 1988
  • Ukraine 1850
  • Slovakia 1671
  • Romania 1647

This group started with a major shock, Slovakia beating Belgium, despite the latter having more than a 300 point ranking advantage at the start of play. Generally, looking at the rankings, one wouldn’t have expected Romania and Slovakia to both get out of this group, albeit the latter by virtue of 3rd place. Both Belgium and Ukraine significantly underperformed their pre-tournament rankings. One might suggest that Belgium just had a bad day at the office against Slovakia, they won the xG battle 1.7 vs 0.57, but that was done spending a lot of that match at 1-0 down, and the more than 300 points gap might have suggested more from Belgium. Ukraine, were outplayed by Romania, even if the scoreline somewhat flattered Romania as a couple of belting long-range efforts found the back of the net.

Overall, the pre-tournament ratings didn’t predict this one at all.

Group F

  • Portugal 2001
  • Czechia 1774
  • Turkey 1757
  • Georgia 1666

The biggest shock in this was clearly Georgia beating Portugal 2-0, though this came in the final game that Portugal played knowing not only that they were through, but definitely through as group winners. Czechia were a disappointment based on pre-tournament ratings, losing to Portugal with a 225 point advantage was absolutely fair enough, but losing to Turkey ranked slightly below them and drawing with Georgia ranked over 100 points below is poor form. To be fair, the game against Turkey was somewhat ill-tempered and had Czechia playing more than half of it a man-down after a first half red-card, and still only lost the game in the dying minutes as they became more and more adventurous in search of a goal that would have ensured their progression. They also significantly outplayed Georgia winning the xG battle 3.07 vs 1.11, enough to feel hard done by in only getting a draw out of the game.

Overall, then, the ratings here didn’t predict the results very well, but did predict how the matches would be balanced reasonably well.

Conclusion

The pre-tournament ratings clearly did better than random chance at predicting the outcomes of the groups. But random chance isn’t the appropriate benchmark. Most of the favourites in the groups did well enough, with perhaps only Spain really impressing. This is despite Spain having the lowest ratings gap to the second rated team, and the third rated team was more closely ranked than all the other second ranked teams. Four of the group favourites won their group, with France and Belgium failing. Both have generally felt as though they have started somewhat sluggishly.

Still, I feel the ratings did a pretty decent job of deciphering what might occur with many of the non-favoured teams, in particular that you might not have seen much of, or know much about before the tournament.

The group games have changed things a little, with Spain now better ranked than France. Austria have somewhat improved gaining 49 ranking points, but are still the 9th highest ranked team left in the tournament (and ranked lower than Croatia who are out).