Football culminates in the FIFA World Cup, the premier showcase of global talent. We investigated factors influencing the number of international caps and goal efficiency of players selected for the 2014 World Cup, focussing on differences between continents. We found that goal efficiency is affected by whether players play in their home country, and age but these relationships differ markedly between continents. We also found that international caps are affected by continent, and the interaction between age and whether players play in their home country. We recommend further analysis be conducted on larger longitudinal data to determine if these relationships hold for the careers of individual players.
With the FIFA World Cup now in full swing fans all over the world are debating who should start for their country, and betting on who will be top goalscorer. Here, we have taken a more analytical approach, using data on all thirty-two national squads to determine what really impacts number of caps and goals of international players, and whether these relationships differ between players from different continents. In the U.S sport as a whole has embraced statistics as a way of enriching the discussions of fans and boosting the success of clubs. European clubs are beginning to catch up, but the statistical revolution here has yet to hit full stride. Delving into data such as this could provide insight into a number of areas. An example comes from suspected age fraud in African players, with nations such as Nigeria unable to match their numerous successes at under-17 level in the senior game. Recent improvements in ageing testing at the under-17 world cup led to a large withdrawal of Nigerian players in 2009, and retrospective testing suggests as many as 35% of players were overage in the three previous tournaments. Rumours have surfaced that professional African players may be as many as ten years older than their official age. Using a large database it may be possible to use the relationship between age and goals for players of a known age to compare to the same relationship for players suspected to be older than they claim to determine their actual age. In recent years some of the traditional dogma of football has begun to be challenged, with work such as the book Soccernomics using data to test numerous assertions, such as racial discrimination in the U.K in the 80’s, and the true age at which players peak. The author’s analysis does an excellent job of highlighting the immense value of information such as this to clubs. Regarding age of peak performance, if a club knows which factors influence this, they can make tailored predictions of when each players will reach his peak, and possibly sell players for large fees when they start their statistical decline, but before other teams notice this decline. The two variables we have chosen to focus on is number of caps and number of goals, with a separate measure created for goals per cap to account for some player scoring lots of goals purely because they have played lots of games. Age is of course a factor we would expect to have a considerable impact on both number of caps and number of goals, but we will also be investigating how this relationship changes depending on which continent a player is from, his height, and whether he plays in his home country. With the top European leagues heavily overrepresented in the World Cup it is inevitable that many countries will rely on players that don’t play their club football in their home country. Debate continues in many countries on the effect this may have, and whether in some countries these players are disadvantaged due to playing football in possibly a different style to other members of the team. This factor may also interact with factors such as age in determining the number of caps a player gets. National coaches may be under pressure to pick young players who have performed well in their domestic league, rather than ageing players playing abroad. Our chief aim here is to assess whether any of these effects exist, and whether the relationships can be quantified to allow predictions to be made about the future caps and goals of players based on factors such as age, height and where they play their club football.
Methods We have used the database of all 736 players originally selected for the 23 player world cup squad of one of the 32 competing nations.
Descriptive Statistics To begin we have graphed summary statistics on continental differences between key factors of the database, which has revealed some interesting differences.
What factors affect goal efficiency of players? To investigate the factors affecting goal efficiency we decide to look only at players listed as forwards, as a substantial number of players have scored very few international goals due to age or position, skewing the data. We created a general linear model (GLM) with goals per game (natural log transformed to normalise) as our response variable and explanatory variables age, continent, whether a player plays in his home country, and height.
What factors affect the number of caps a player gets? To investigate factors affecting the number of caps a player gets we constructed a quasipoisson GLM, using all players with at least 1 international cap. We fitted number of caps as the response variable alongside explanatory variables age, continent, height and whether a player plays in his home country.
Results Descriptive Statistics One interesting factor we investigated is height. At a national level there was considerable variation, with Ghana’s squad standing a diminutive 173.6cm tall on average, compared to the average German squad member at 185.2cm tall. Height is a trait with considerable global variation, impacted by numerous factors. We compared average height of a nation’s squad with average height of an adult male from that nation and found some striking differences.
Figure 1.0 – Height difference between average member of the 2014 World Cup squad of a nation and average adult male from that nation. Data shown for the five smallest and five largest differences.
African and Australian squads picked the youngest sides for the tournament (Figure 2.0) with Ghana, Nigeria and Australia all picking teams with an average age of under 26. Squads from the Americas were much older on average, with Argentina boasting the highest average age of 28.5 years. There was substantial goal efficiency variation within continents, translating to small differences in averages between the continents, with South America achieving the highest goal efficiency (Figure 3.0) alongside highest average age (Figure 2.0). Brazil, Mexico and Germany have the highest goal efficiency of all nations, with Greece, Costa Rica and Italy having the smallest.
There was much more variation in number of caps between continents, although on the whole continents with older players on average also unsurprisingly had players with more international experience on average. It is important to note at this stage that the continent of Australia is only represented by the nation of Australia, whereas there are 13 countries representing Europe so there are some issues in comparing these continents. The most experienced nations are Spain, Uruguay, Honduras and Germany all with an average of over 40 caps for each player. At the other end of the spectrum is Algeria, Australia and France with under 25 caps on average for each player.
Average number of goals was highly variable, and as a result it is difficult to suggest substantial differences between continents. At a national level, Germany and Spain we big leaders in average goals, with the average player from that nation having over 9 goals. Algeria, Colombia and Nigeria brought the squads with the fewest average goals.
What factors affect goal efficiency of players? Goal efficiency was significantly influenced by an interaction between home continent and player age (F5,140=2.4199, p=0.03871) and an interaction between continent and whether they played for their home country (F4,140=3.6123, p=0.007796)
What factors affect the number of caps a player gets? Number of caps was found to be significantly influenced by continent (Quasipoisson GLM, F1,627= 4.3008, p=0.0007445) and an interaction between player age and whether they played for their home country or not (F1,627 =10.579, p=0.001205). There was a trend for increasing number of caps with increasing height, but this relationship was not significant (F1,626 =3.3297,p=0.06852).
Does it matter if you play in your home country? In terms of number of international caps we found that for players playing their home country, number of caps increases slightly with age, but for players not playing in their home country number of caps massively increases with age. Again it is important to note that this data isn’t following the ageing of individual players, rather it is a snapshot in time of players of different ages. Nonetheless, it still shows that playing away from your home country can do wonders for your international career, and unsurprising trend that is probably driven by non-European nations due to the likes of Spain, England and Germany having players from their home country playing in the best leagues in the world. In this model number of caps was also significantly influenced by continent, unsurprising given that we know some continents brought more experienced teams than others (Figure 4). More interestingly there was a trend that was not quite significant of number of caps increasing with increasing height. More work on this is needed to confirm this relationship, particularly by controlling for position as goalkeepers and defenders may be taller and have longer international lifespan for reasons other than their height. Figure 1 shows the five lowest and five greatest differences in height between the nation and its football stars. Interestingly the Netherlands is the only country where footballers were smaller on average than males from that nation, although as arguably the tallest nation in the world that is somewhat unsurprising. At the other end of the spectrum both Ecuador and Nigeria named squads where players were on average over 15cm taller than the average adult male. With the heritability of human height at around 0.6-0.8 perhaps genetic factors are more important in becoming an international footballer in countries such as Nigerian, Ecuador and Iran compared to nations such as Spain, Ghana and Germany, where it takes much more that height to get spotted.
Do South America and Asian players get less efficient with age? Figure 6.0 shows that older players from South America and Asia show a decreased efficiency in terms of goals per game. It is important to first note that this is not data following the ageing of individual players, rather it is a cross section of efficiency of players in the squad as a whole. Furthermore this analysis was performed using only strikers, so cannot be explained by any effect of these continents tending to pick younger strikers and older defenders. Several possibilities remain. Firstly, young strikers may only be considered for selection if they have proved themselves efficient, whereas older players may still make the squad based on previous form, experience and leadership. Secondly players from South America and Asia may indeed peak at a young age. This could be easily tested using data from previous international or domestic competitions and investigating the change in performance of individual players with age. Perhaps the style of football utilized by South American teams favours younger players, or perhaps Africa has been unfairly scrutinized when investigating age fraud in young players. Further analysis is definitely warranted here.
Does it matter if you play in your home country? For goal efficiency it may depend on continent. We found a significant interaction between continent and home country affecting goal efficiency. In Europe and Asia playing in your home country led to decreased goals per game, whereas in the other continents it led to increased goals per game. This is in contrast to the general view that players in many continents benefit from playing abroad in the best European leagues, as they may actually have higher goal efficiency if they play in their home country. This could potentially be influenced by small sample sizes of some players, and a larger database is needed to see if this effect is real, and players perform better at international level if they play in their home country. In terms of number of international caps we found that for players playing their home country, number of caps increases slightly with age, but for players not playing in their home country number of caps massively increases with age. Again it is important to note that this data isn’t following the ageing of individual players, rather it is a snapshot in time of players of different ages. Nonetheless, it still shows that playing away from your home country can do wonders for your international career, and unsurprising trend that is probably driven by non-European nations due to the likes of Spain, England and Germany having players from their home country playing in the best leagues in the world. In this model number of caps was also significantly influenced by continent, unsurprising given that we know some continents brought more experienced teams than others (Figure 4). More interestingly there was a trend that was not quite significant of number of caps increasing with increasing height. More work on this is needed to confirm this relationship, particularly by controlling for position as goalkeepers and defenders may be taller and have longer international lifespan for reasons other than their height. In this report we have highlighted several intriguing statistical relationships, particularly related to how the factors affecting goal efficiency and number of international caps vary depending on which continent the players are from. These conclusions are based on a relatively small sample of the 736 players selected for the world cup, unevenly split up into continents. In the information age much larger datasets exist, on a much deeper array of football stats. This relatively untapped data source could enhance comparison of player ability and improve efficiency in the transfer market.