Club and Country
Does club form or previous international performance better predict a footballer's performance at an international tournament?
Going into the 2022 World Cup, Harry Maguire posed a problem for Gareth Southgate, the England football manager. At last year’s European Championships he was one of England’s better players. He anchored the defence, made no obvious errors, and was good in the air. He won 21 aerial duels, the second most in the tournament.
However, since last summer Maguire’s club performances have been substandard. When Manchester United, his club, lost 4-0 to Brentford, fans ranked his performance 2 out of 10. He has since been out of favour and has only started 3 of 14 games this season.
Many international managers face similar problems to Southgate’s Maguire problem. Do you stick with someone who has performed for you in previous internationals? Or do you drop a player who is underperforming for their club?
I explored this question using data on 85 outfield players who played at Euro 2020 (which took place in 2021) or Copa America 2021. On a range of key metrics, I tested whether their recent club performance or recent international performance better predicted their performance at the tournaments.
The graph below summarises the results. It shows how much of the variation between different players on a given metric (for instance, expected goals) at the Euros and Copa America can be explained using different information. If the score is 100 this means that you can perfectly predict performance on that metric using the given information. If it is 0 it means your predictions are basically useless.
For each metric, I use three sets of information. Firstly, the performance of each player on a given metric in previous internationals between 2018 and 2020. Secondly, the performance of each player on the same metric in their domestic league (club football) for 2020. Finally, I allow the model to combine the two bits of information. (For the technically minded: I run a linear regression with each variable and then both variables. The x value in the chart below is the R squared of those regressions.)
For most, but not all, metrics domestic league performance can explain more of the variation in tournament performance than previous internationals. This suggests that managers should put more weight on domestic league performance than previous international performance. You can see this in the two plots below. For all metrics there is a strong positive relationship between domestic league performance and tournament performance, with the points clustered around the trendline. The relationship is statistically significant for every metric.
While there are positive correlations between previous international performance on a metric and tournament performance, there is more noise around them. The points are less tightly clustered around the trendline. This makes sense. Players typically play more minutes in a domestic season than in international competitions, so their domestic statistics are less likely to be affected by a few fluke performances. Domestic competition is also more recent, so it is likely to be a more up-to-date account of that player’s ability.
However, for most metrics the best prediction uses a combination of both pieces of information. While domestic league performance is on balance the more useful metric, previous international performance is also useful. It shows a player in a more similar context than domestic football (some players may suit the style of their national team much better than their domestic team). A sensible international manager will weigh up both pieces of information alongside other factors when picking their team.
Gareth Southgate leaned on Maguire’s previous international performances and picked him in England’s first World Cup game. Maguire rewarded Southgate with a solid performance and an assist. As an England fan, I hope Maguire steers clear of his domestic demons and proves Southgate right. I worry he won’t.
I share the concerns about the World Cup in Qatar others have raised. I do not condone the treatment of LGBTQ+ people and migrant workers in Qatar. I’ve decided to watch it, as I love the World Cup and would struggle to stop myself watching it. I respect those who’ve decided to boycott it.