Foreigners and goals in English football

More foreigners = greater success?

One of the striking features of the Premier League, and of league football in England as a whole, is the number of non-English players. Famously, on 26th December 1999, Chelsea started a game with no English players at all. Current estimates I've seen suggest that a little over 70% of Premier League players aren't English. This begs the question, what effect do foreign players have on the game? Do teams with more foreign players score more? In this blog post, I'm going to look at the data.

(Canva)

Some background

The data for this analysis comes from transfermarkt.com (https://www.transfermarkt.com/), who have player nationality data going back to the early 2000s for many leagues. The foreigner count data is only updated at the start of each season.

The obvious issue is definitional: who is a foreigner? TransferMarkt appears to be using the definition that any player whose allegiance is to a national team other than England. This of course means that Scottish, Welsh, and Northern Irish players count as foreigners. It's a complicated area and I'll leave the definition to others and just use the data for analysis.

The charts

For each team in each league for each season, I calculated the total number of goals scored ("For goals"), the number of goals they conceded ("Against goals"), and the goal difference ("Net goals"). From TransferMarkt, I got the number of foreign players for each team at the start of the season. I plotted "For goals" etc. against "foreigner count" in the charts below.

I've put a linear fit on the charts, with the r² value and p-value (for the predictor variable foreigner count) in the chart heading. The band around the fit is the 95% confidence interval.

These charts are interactive:

• From the menu on the left hand side you can zoom in and move around the plots and even download them.
• You can hover over the points to see the data.
• You can select the league tier to look at from the control below the chart.
• You can change the year you're looking at using the year slider.

What the charts show

Let's start with the obvious, there's no strong correlation between foreigners and goals in any league and for any year in the data set. Almost all r² values are low, and in the handful of cases where they're high (above 0.8), that's due to small data set. The p-values are telling us largely the same story.

Is it worth including foreigner count as feature in a machine learning model? My answer is no. If there is a correlation between foreigners and goals, it's a very small one.

Was there a correlation in the past? I don't have enough data to say. We'd really need to go back to the 1970s or earlier to get a bigger picture. The number of foreign players grew from small numbers in the 1970s, and the growth was presumably due to the belief that foreign talent could help a team to win. In principle then, we should see some form of correlation between foreign players and goals during the growth years for foreign players, if, of course, we could get the data.

One thing that surprised me was the absence of a trend even in the lower leagues. For example, there's no trend in tier 5 (the National League). It's noticeable that there are substantial numbers of foreign players even at the lower levels of English football.

Is this level of foreign players good for the English game? For fans, I would argue the answer is yes. The Premier League is in effect, a global league with many of the world's best players running on to English pitches every week. This holds for the lower tiers too. I've heard argument that too many foreign players are bad for the English national team, but I'm not convinced. I think the problems of the English national team are deep rooted in process and organization.